id
stringlengths 9
16
| submitter
stringlengths 1
64
⌀ | authors
stringlengths 3
62.8k
| title
stringlengths 1
382
| comments
stringlengths 1
1.15k
⌀ | journal-ref
stringlengths 1
557
⌀ | doi
stringlengths 7
153
⌀ | report-no
stringlengths 1
509
⌀ | categories
stringlengths 5
125
| license
stringclasses 9
values | abstract
stringlengths 6
6.09k
| versions
listlengths 1
187
| update_date
timestamp[s] | authors_parsed
sequencelengths 1
2.83k
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2502.14836 | Hans-Thomas Janka | H.-Thomas Janka (MPI Astrophysics, Garching) | Long-Term Multidimensional Models of Core-Collapse Supernovae: Progress
and Challenges | 40 pages, 8 figures; minor revisions due to referee comments;
accepted by Annual Review of Nuclear and Particle Science; article DOI:
https://doi.org/10.1146/annurev-nucl-121423-100945 (the link will become
active upon publication) | null | null | null | astro-ph.HE hep-ph nucl-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Self-consistent, multidimensional core-collapse supernova (SN) simulations,
especially in 3D, have achieved tremendous progress over the past 10 years.
They are now able to follow the entire evolution from core collapse through
bounce, neutrino-triggered shock revival, shock breakout at the stellar surface
to the electromagnetic SN outburst and the subsequent SN remnant phase. Thus
they provide general support for the neutrino-driven explosion mechanism by
reproducing observed SN energies, neutron-star (NS) kicks, and diagnostically
relevant radioactive isotope yields; they allow to predict neutrino and
gravitational-wave signals for many seconds of proto-NS cooling; they confirm
correlations between explosion and progenitor or remnant properties already
expected from previous spherically symmetric (1D) and 2D models; and they carve
out various scenarios for stellar-mass black-hole (BH) formation. Despite these
successes it is currently unclear which stars explode or form BHs, because
different modeling approaches disagree and suggest the possible importance of
the 3D nature of the progenitors and of magnetic fields. The role of neutrino
flavor conversion in SN cores still needs to be better understood, the nuclear
equation of state including potential phase transitions implies major
uncertainties, the SN 1987A neutrino measurements raise new puzzles, and
tracing a possible correlation of NS spins and kicks requires still more
refined SN simulations.
| [
{
"version": "v1",
"created": "Thu, 20 Feb 2025 18:49:33 GMT"
},
{
"version": "v2",
"created": "Wed, 19 Mar 2025 20:12:33 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Janka",
"H. -Thomas",
"",
"MPI Astrophysics, Garching"
]
] |
2502.14957 | Gautham Narayana Sabhahit Mr. | Gautham N. Sabhahit, Jorick S. Vink, Andreas A. C. Sander, Matheus
Bernini-Peron, Paul A. Crowther, Roel R. Lefever, and Tomer Shenar | A new mass estimate method with hydrodynamical atmospheres for very
massive WNh stars | Accepted for publication in A&A, 20 pages, 18 figures, 4 tables | null | null | null | astro-ph.SR astro-ph.GA astro-ph.HE | http://creativecommons.org/licenses/by-sa/4.0/ | Very massive stars with masses over 100 Msun are key objects in the Universe
for our understanding of chemical and energetic feedback in the Universe, but
their evolution and fate are almost entirely determined by their wind mass
loss. We aim to determine the mass of the most massive star known in the Local
Group R136a1. For this we compute the first hydrodynamically consistent
non-local thermodynamical equilibrium atmosphere models for both R136a1 (WN5h)
as well as the binary system R144 (WN5/6h+WN6/7h) in the Tarantula nebula.
Using the Potsdam Wolf-Rayet code, we simultaneously empirically derive and
theoretically predict mass-loss rates and wind velocities. By fitting synthetic
spectra derived from these models to multi-wavelength observations, we
constrain the stellar and wind properties of R144 and R136a1. We first
determine the clumping stratification required by our hydro-models to fit the
spectra of R144 by using the available dynamical mass estimates for the two
components. We then utilise this clumping stratification in hydrodynamic models
of R136a1 and estimate a mass of $M_\mathrm{Hydro}$ of 233 Msun. Remarkably,
the estimated mass is close to and entirely consistent with chemical
homogeneous mass relations. This present-day mass of 233 Msun provides a lower
limit to the initial stellar mass, that could be far higher due to previous
wind mass loss.
| [
{
"version": "v1",
"created": "Thu, 20 Feb 2025 19:00:01 GMT"
},
{
"version": "v2",
"created": "Wed, 19 Mar 2025 18:15:56 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Sabhahit",
"Gautham N.",
""
],
[
"Vink",
"Jorick S.",
""
],
[
"Sander",
"Andreas A. C.",
""
],
[
"Bernini-Peron",
"Matheus",
""
],
[
"Crowther",
"Paul A.",
""
],
[
"Lefever",
"Roel R.",
""
],
[
"Shenar",
"Tomer",
""
]
] |
2502.15308 | Yichen Liu | Yichen Liu, San-Dong Guo, Yongpan Li and Cheng-Cheng Liu | Two-dimensional fully-compensated Ferrimagnetism | Phys. Rev. Lett. (2025); 7-page text + 7-page Supplemental Material | Phys. Rev. Lett. 134, 116703 (2025) | 10.1103/PhysRevLett.134.116703 | null | cond-mat.mtrl-sci cond-mat.mes-hall | http://creativecommons.org/licenses/by/4.0/ | Antiferromagnetic spintronics has long been a subject of intense research
interest, and the recent introduction of altermagnetism has further ignited
enthusiasm in the field. However, fully-compensated ferrimagnetism, which
exhibits band spin splitting but zero net magnetization, has yet to receive
enough attention. Since the experimental preparation of two-dimensional (2D)
magnetic van der Waals (vdW) materials in 2017, 2D magnetic materials, thanks
to their super tunability, have quickly become an important playground for
spintronics. Here, we extend the concept of fully-compensated ferrimagnetism
(fFIM) to two dimensions and propose 2D \textit{filling-enforced} fFIM,
demonstrate its stability and ease of manipulation, and present three feasible
realization schemes with respective exemplary candidate materials. A simple
model for 2D fully-compensated ferrimagnets (fFIMs) is developed. Further
investigation of 2D fFIMs' physical properties reveals that they not only
exhibit significant magneto-optical response but also show fully spin-polarized
currents and the anomalous Hall effect in the half-metallic states, displaying
characteristics previously almost exclusive to ferromagnetic materials, greatly
broadening the research and application prospects of spintronic materials.
| [
{
"version": "v1",
"created": "Fri, 21 Feb 2025 09:03:07 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Liu",
"Yichen",
""
],
[
"Guo",
"San-Dong",
""
],
[
"Li",
"Yongpan",
""
],
[
"Liu",
"Cheng-Cheng",
""
]
] |
2502.15451 | Yuan Sun | Yuan Sun | Binary-Integer-Programming Based Algorithm for Expert Load Balancing in
Mixture-of-Experts Models | null | null | null | null | cs.LG cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | For pre-training of MoE (Mixture-of-Experts) models, one of the main issues
is unbalanced expert loads, which may cause routing collapse or increased
computational overhead. Existing methods contain the Loss-Controlled method and
the Loss-Free method, where both the unbalanced degrees at first several
training steps are still high and decrease slowly. In this work, we propose
BIP-Based Balancing, an expert load balancing algorithm based on binary integer
programming (BIP). The algorithm maintains an additional vector q on each MoE
layer that can help change the top-K order of s by solving a binary integer
programming with very small time costs. We implement the algorithm on two MoE
language models: 16-expert (0.3B) and 64-expert (1.1B). The experimental
results show that on both models comparing with the Loss-Controlled method and
the Loss-Free method, our algorithm trains models with the lowest perplexities,
while saves at least 13% of pre-training time compared with the Loss-Controlled
method. Within our current knowledge, this is the first routing algorithm that
achieves maintaining load balance status on every expert in every MoE layer
from the first step to the last step during the whole pre-training process,
while the trained MoE models also perform well. The code material of this work
is available at https://github.com/sunyuanLLM/bip_routing_algorithm.
| [
{
"version": "v1",
"created": "Fri, 21 Feb 2025 13:25:00 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 14:10:27 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Sun",
"Yuan",
""
]
] |
2502.15540 | Milad Sefidgaran | Milad Sefidgaran and Abdellatif Zaidi and Piotr Krasnowski | Generalization Guarantees for Representation Learning via Data-Dependent
Gaussian Mixture Priors | Accepted as a Spotlight Paper at ICLR 2025 | null | null | null | stat.ML cs.IT cs.LG math.IT | http://creativecommons.org/licenses/by/4.0/ | We establish in-expectation and tail bounds on the generalization error of
representation learning type algorithms. The bounds are in terms of the
relative entropy between the distribution of the representations extracted from
the training and "test'' datasets and a data-dependent symmetric prior, i.e.,
the Minimum Description Length (MDL) of the latent variables for the training
and test datasets. Our bounds are shown to reflect the "structure" and
"simplicity'' of the encoder and significantly improve upon the few existing
ones for the studied model. We then use our in-expectation bound to devise a
suitable data-dependent regularizer; and we investigate thoroughly the
important question of the selection of the prior. We propose a systematic
approach to simultaneously learning a data-dependent Gaussian mixture prior and
using it as a regularizer. Interestingly, we show that a weighted attention
mechanism emerges naturally in this procedure. Our experiments show that our
approach outperforms the now popular Variational Information Bottleneck (VIB)
method as well as the recent Category-Dependent VIB (CDVIB).
| [
{
"version": "v1",
"created": "Fri, 21 Feb 2025 15:43:31 GMT"
},
{
"version": "v2",
"created": "Wed, 19 Mar 2025 22:37:44 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Sefidgaran",
"Milad",
""
],
[
"Zaidi",
"Abdellatif",
""
],
[
"Krasnowski",
"Piotr",
""
]
] |
2502.16182 | Shivank Garg | Shivank Garg, Ayush Singh, Shweta Singh, Paras Chopra | IPO: Your Language Model is Secretly a Preference Classifier | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Reinforcement learning from human feedback (RLHF) has emerged as the primary
method for aligning large language models (LLMs) with human preferences. While
it enables LLMs to achieve human-level alignment, it often incurs significant
computational and financial costs due to its reliance on training external
reward models or human-labeled preferences. In this work, we propose Implicit
Preference Optimization (IPO), an alternative approach that leverages
generative LLMs as preference classifiers, thereby reducing the dependence on
external human feedback or reward models to obtain preferences. We conduct a
comprehensive evaluation on the preference classification ability of LLMs using
RewardBench, assessing models across different sizes, architectures, and
training levels to validate our hypothesis. Furthermore, we investigate the
self-improvement capabilities of LLMs by generating multiple responses for a
given instruction and employing the model itself as a preference classifier for
Direct Preference Optimization (DPO)-based training. Our findings demonstrate
that models trained through IPO achieve performance comparable to those
utilizing state-of-the-art reward models for obtaining preferences.
| [
{
"version": "v1",
"created": "Sat, 22 Feb 2025 10:59:11 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 10:52:45 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Garg",
"Shivank",
""
],
[
"Singh",
"Ayush",
""
],
[
"Singh",
"Shweta",
""
],
[
"Chopra",
"Paras",
""
]
] |
2502.17031 | Piotr Pokora | Piotr Pokora | If Terao's conjecture fails in $\mathbb{P}^{2}$, then it fails in
$\mathbb{P}^{n > 2}$ | 6 pages, Version 1.1, a new title and minor editorial changes have
been made | null | null | null | math.AG math.AC math.CO | http://creativecommons.org/licenses/by/4.0/ | Using an elementary technique, we construct a first example of Ziegler pairs
of hyperplane arrangements in $\mathbb{P}^{3}$. Then, using this construction,
we show how to obtain possible counterexamples to Terao's freeness conjecture
in $\mathbb{P}^{n}$ with $n>2$ using arrangements of lines in $\mathbb{P}^{2}$.
| [
{
"version": "v1",
"created": "Mon, 24 Feb 2025 10:31:16 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 07:50:30 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Pokora",
"Piotr",
""
]
] |
2502.17179 | Vyshnav Mohan | Fri{\dh}rik Freyr Gautason, Vyshnav Mohan and L\'arus Thorlacius | Late-Time Saturation of Black Hole Complexity | 35 pages and 9 figures. v2: typos fixed and references added | null | null | null | hep-th gr-qc | http://creativecommons.org/licenses/by/4.0/ | The holographic complexity of a static spherically symmetric black hole,
defined as the volume of an extremal surface, grows linearly with time at late
times in general relativity. The growth comes from a region at a constant
transverse area inside the black hole and continues forever in the classical
theory. In this region the volume complexity of any spherically symmetric black
hole in $d+1$ spacetime dimensions reduces to a geodesic length in an effective
two-dimensional JT-gravity theory. The length in JT-gravity has been argued to
saturate at very late times via non-perturbative corrections obtained from a
random matrix description of the gravity theory. The same argument, applied to
our effective JT-gravity description of the volume complexity, leads to
complexity saturation at times of exponential order in the Bekenstein-Hawking
entropy of a $d+1$-dimensional black hole. Along the way, we explore a simple
toy model for complexity growth, based on a discretisation of Nielsen
complexity geometry, that can be analytically shown to exhibit the expected
late-time complexity saturation.
| [
{
"version": "v1",
"created": "Mon, 24 Feb 2025 14:14:08 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 16:31:29 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Gautason",
"Friðrik Freyr",
""
],
[
"Mohan",
"Vyshnav",
""
],
[
"Thorlacius",
"Lárus",
""
]
] |
2502.17984 | Keivan Shariatmadar PhD | Keivan Shariatmadar, Neil Yorke-Smith, Ahmad Osman, Fabio Cuzzolin,
Hans Hallez, David Moens | Generalized Decision Focused Learning under Imprecise
Uncertainty--Theoretical Study | 13 pages | null | null | null | cs.LG math.OC math.PR | http://creativecommons.org/licenses/by-sa/4.0/ | Decision Focused Learning has emerged as a critical paradigm for integrating
machine learning with downstream optimisation. Despite its promise, existing
methodologies predominantly rely on probabilistic models and focus narrowly on
task objectives, overlooking the nuanced challenges posed by epistemic
uncertainty, non-probabilistic modelling approaches, and the integration of
uncertainty into optimisation constraints. This paper bridges these gaps by
introducing innovative frameworks: (i) a non-probabilistic lens for epistemic
uncertainty representation, leveraging intervals (the least informative
uncertainty model), Contamination (hybrid model), and probability boxes (the
most informative uncertainty model); (ii) methodologies to incorporate
uncertainty into constraints, expanding Decision-Focused Learning's utility in
constrained environments; (iii) the adoption of Imprecise Decision Theory for
ambiguity-rich decision-making contexts; and (iv) strategies for addressing
sparse data challenges. Empirical evaluations on benchmark optimisation
problems demonstrate the efficacy of these approaches in improving decision
quality and robustness and dealing with said gaps.
| [
{
"version": "v1",
"created": "Tue, 25 Feb 2025 08:53:02 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 12:02:58 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Shariatmadar",
"Keivan",
""
],
[
"Yorke-Smith",
"Neil",
""
],
[
"Osman",
"Ahmad",
""
],
[
"Cuzzolin",
"Fabio",
""
],
[
"Hallez",
"Hans",
""
],
[
"Moens",
"David",
""
]
] |
2502.18372 | Darvin Wanisch | Darvin Wanisch, Nora Reini\'c, Daniel Jaschke, Simone Montangero and
Pietro Silvi | Entanglement transitions in a boundary-driven open quantum many-body
system | 7 pages, 4 figures | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | We present a numerical framework based on tree tensor operators that enables
large-scale simulation of out-of-equilibrium open quantum many-body systems. By
design, it protects density operator positivity and provides direct access to
entanglement monotones, such as entanglement of formation and logarithmic
negativity. To demonstrate the framework's ability to probe entanglement in
open quantum many-body systems and distinguish it from other correlations, we
study a paradigmatic open system problem: the boundary-driven XXZ spin-chain.
We uncover entanglement transitions driven by both the coupling to the
environment and the anisotropy parameter. These transitions reveal an immediate
connection between entanglement and spin-current, and link the known transport
regimes of the model to distinct entanglement regimes, i.e., separable,
area-law, and volume-law. Our work enables the analysis of entanglement in open
quantum many-body systems out of equilibrium, a necessary step for developing
scalable quantum technologies.
| [
{
"version": "v1",
"created": "Tue, 25 Feb 2025 17:09:13 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 11:35:36 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Wanisch",
"Darvin",
""
],
[
"Reinić",
"Nora",
""
],
[
"Jaschke",
"Daniel",
""
],
[
"Montangero",
"Simone",
""
],
[
"Silvi",
"Pietro",
""
]
] |
2502.18435 | Yizhe Zhang | Yizhe Zhang, Richard Bai, Zijin Gu, Ruixiang Zhang, Jiatao Gu,
Emmanuel Abbe, Samy Bengio, Navdeep Jaitly | Reversal Blessing: Thinking Backward May Outpace Thinking Forward in
Multi-choice Questions | null | null | null | null | cs.CL cs.IT cs.LG math.IT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Language models usually use left-to-right (L2R) autoregressive factorization.
However, L2R factorization may not always be the best inductive bias.
Therefore, we investigate whether alternative factorizations of the text
distribution could be beneficial in some tasks. We investigate right-to-left
(R2L) training as a compelling alternative, focusing on multiple-choice
questions (MCQs) as a test bed for knowledge extraction and reasoning. Through
extensive experiments across various model sizes (2B-8B parameters) and
training datasets, we find that R2L models can significantly outperform L2R
models on several MCQ benchmarks, including logical reasoning, commonsense
understanding, and truthfulness assessment tasks. Our analysis reveals that
this performance difference may be fundamentally linked to multiple factors
including calibration, computability and directional conditional entropy. We
ablate the impact of these factors through controlled simulation studies using
arithmetic tasks, where the impacting factors can be better disentangled. Our
work demonstrates that exploring alternative factorizations of the text
distribution can lead to improvements in LLM capabilities and provides
theoretical insights into optimal factorization towards approximating human
language distribution, and when each reasoning order might be more
advantageous.
| [
{
"version": "v1",
"created": "Tue, 25 Feb 2025 18:30:25 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 03:25:21 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Zhang",
"Yizhe",
""
],
[
"Bai",
"Richard",
""
],
[
"Gu",
"Zijin",
""
],
[
"Zhang",
"Ruixiang",
""
],
[
"Gu",
"Jiatao",
""
],
[
"Abbe",
"Emmanuel",
""
],
[
"Bengio",
"Samy",
""
],
[
"Jaitly",
"Navdeep",
""
]
] |
2502.18637 | Dominik Va\v{s}inka | Dominik Va\v{s}inka, Filip Jur\'a\v{n}, Jarom\'ir B\v{e}hal, and
Miroslav Je\v{z}ek | From Stars to Molecules: AI Guided Device-Agnostic Super-Resolution
Imaging | 10 pages, 7 figures | null | null | null | physics.optics astro-ph.IM quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Super-resolution imaging has revolutionized the study of systems ranging from
molecular structures to distant galaxies. However, existing super-resolution
methods require extensive calibration and retraining for each imaging setup,
limiting their practical deployment. We introduce a device-agnostic
deep-learning framework for super-resolution imaging of point-like emitters
that eliminates the need for calibration data or explicit knowledge of optical
system parameters. Our model is trained on a diverse, numerically simulated
dataset encompassing a broad range of imaging conditions, enabling
generalization across different optical setups. Once trained, it reconstructs
super-resolved images directly from a single resolution-limited camera frame
with superior accuracy and computational efficiency compared to
state-of-the-art methods. We experimentally validate our approach using a
custom microscopy setup with ground-truth emitter positions. We also
demonstrate its versatility on astronomical and single-molecule localization
microscopy datasets, achieving unprecedented resolution without prior
information. Our findings establish a pathway toward universal,
calibration-free super-resolution imaging, expanding its applicability across
scientific disciplines.
| [
{
"version": "v1",
"created": "Tue, 25 Feb 2025 20:54:27 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 17:15:36 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Vašinka",
"Dominik",
""
],
[
"Juráň",
"Filip",
""
],
[
"Běhal",
"Jaromír",
""
],
[
"Ježek",
"Miroslav",
""
]
] |
2502.19227 | Yunyang Li | Yunyang Li, Zaishuo Xia, Lin Huang, Xinran Wei, Han Yang, Sam Harshe,
Zun Wang, Chang Liu, Jia Zhang, Bin Shao, Mark B. Gerstein | Enhancing the Scalability and Applicability of Kohn-Sham Hamiltonians
for Molecular Systems | null | null | null | null | physics.chem-ph cond-mat.mtrl-sci cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Density Functional Theory (DFT) is a pivotal method within quantum chemistry
and materials science, with its core involving the construction and solution of
the Kohn-Sham Hamiltonian. Despite its importance, the application of DFT is
frequently limited by the substantial computational resources required to
construct the Kohn-Sham Hamiltonian. In response to these limitations, current
research has employed deep-learning models to efficiently predict molecular and
solid Hamiltonians, with roto-translational symmetries encoded in their neural
networks. However, the scalability of prior models may be problematic when
applied to large molecules, resulting in non-physical predictions of
ground-state properties. In this study, we generate a substantially larger
training set (PubChemQH) than used previously and use it to create a scalable
model for DFT calculations with physical accuracy. For our model, we introduce
a loss function derived from physical principles, which we call Wavefunction
Alignment Loss (WALoss). WALoss involves performing a basis change on the
predicted Hamiltonian to align it with the observed one; thus, the resulting
differences can serve as a surrogate for orbital energy differences, allowing
models to make better predictions for molecular orbitals and total energies
than previously possible. WALoss also substantially accelerates
self-consistent-field (SCF) DFT calculations. Here, we show it achieves a
reduction in total energy prediction error by a factor of 1347 and an SCF
calculation speed-up by a factor of 18%. These substantial improvements set new
benchmarks for achieving accurate and applicable predictions in larger
molecular systems.
| [
{
"version": "v1",
"created": "Wed, 26 Feb 2025 15:36:25 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 17:54:16 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Li",
"Yunyang",
""
],
[
"Xia",
"Zaishuo",
""
],
[
"Huang",
"Lin",
""
],
[
"Wei",
"Xinran",
""
],
[
"Yang",
"Han",
""
],
[
"Harshe",
"Sam",
""
],
[
"Wang",
"Zun",
""
],
[
"Liu",
"Chang",
""
],
[
"Zhang",
"Jia",
""
],
[
"Shao",
"Bin",
""
],
[
"Gerstein",
"Mark B.",
""
]
] |
2502.19322 | Shruti Menon | Suvankar Dutta, Shruti Menon and Aayush Srivastav | Bosonisation and BTZ Black Hole Microstates | 22 pages, 1 figure | null | null | null | hep-th | http://creativecommons.org/publicdomain/zero/1.0/ | When the boundary dynamics of \(AdS_3\) gravity is governed by the collective
field theory Hamiltonian proposed by Jevicki and Sakita, its asymptotic
symmetry algebra becomes the centerless \(U(1)\) Kac-Moody algebra. We quantize
this system using the quantum bosonization of relativistic free fermions and
relate these to the dynamical fields of \(AdS_3\) gravity. This leads to a
correspondence where different bulk configurations correspond to distinct
states (particle-hole pair excitations) in the fermionic Hilbert space. This
mapping allows us to construct BTZ black hole microstates, represented by Young
diagrams of irreducible \(U(\infty)\) representations. Notably, the logarithm
of the microstate degeneracy exactly reproduces the classical entropy of the
BTZ black hole.
| [
{
"version": "v1",
"created": "Wed, 26 Feb 2025 17:15:37 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 04:12:04 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Dutta",
"Suvankar",
""
],
[
"Menon",
"Shruti",
""
],
[
"Srivastav",
"Aayush",
""
]
] |
2502.19458 | Yue-Liang Wu | Yue-Liang Wu | A general theory of the standard model and the revelation of the dark
side of the universe | 8 pages, typos corrected | null | 10.1016/j.scib.2025.03.026 | null | physics.gen-ph | http://creativecommons.org/licenses/by/4.0/ | A general theory of the Standard Model (GSM) is presented in a spin-related
gravigauge spacetime, grounded on conformal inhomogeneous spin gauge symmetry
WS$_c$(1,3)=SP(1,3)$\rtimes$W$^{1,3}$$\rtimes$SP$_c$(1,1) and scaling gauge
symmetry SG(1), along with symmetry U$_Y$(1)$\times$SU$_L$(2)$\times$SU(3)$_c$
in the standard model. It introduces new interactions: spin, chirality
boost-spin, chiral conformal-spin, and scaling gauge forces. A gravitization
equation is obtained to describe the gravitational effects emerging from the
non-commutative derivative operator in gravigauge spacetime. In the framework
of gravitational quantum field theory(GQFT), the gravitational force is
mediated by the gravigauge field, identified as the massless graviton. Both
gauge-type and geometric-type gravitational equations are derived to describe
gravidynamics beyond general relativity. A massive chirality boost-spin gauge
boson, termed the dark graviton, acts as a dark matter candidate, interacting
with leptons and quarks via a spin gauge boson. A fundamental scalar field
serves as a source of primordial energy and dark energy, driving the early
inflation and current accelerated expansion of the universe.
| [
{
"version": "v1",
"created": "Wed, 26 Feb 2025 10:01:16 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 03:03:03 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Wu",
"Yue-Liang",
""
]
] |
2502.19523 | Xavier Roulleau | David Broadhurst, Xavier Roulleau | Number of partitions of modular integers (with an Appendix by P.
Deligne) | 23 pages, Conjecture 1 of the first version is a theorem; a proof is
given in an Appendix by Pierre Deligne | null | null | null | math.NT hep-th math.CO | http://creativecommons.org/licenses/by-nc-sa/4.0/ | For integers $n,k,s$, we give a formula for the number $T(n,k,s)$ of order
$k$ subsets of the ring $\mathbb{Z}/n\mathbb{Z}$ whose sum of elements is $s$
modulo $n$. To do so, we describe explicitly a sequence of matrices $M(k)$, for
positive integers $k$, such that the size of $M(k)$ is the number of divisors
of $k$, and for two coprime integers $k_{1},k_{2}$, the matrix $M(k_{1}k_{2})$
is the Kronecker product of $M(k_{1})$ and $M(k_{2})$. For $s=0, 1, 2$, and for
$s=k/2$ when $k$ is even, the sequences $T(n,k,s)$ are related to the number of
necklaces with $k$ black beads and $n-k$ white beads, and to Lyndon words. This
work begins with empirical determinations of $M(k)$ up to $k=10000$, from which
we infer a closed formula that encompasses many entries in the Encyclopedia of
Integer Sequences. Its proof comes from work on Ramanujan sums, by Ramanathan,
with a generalization to wider problems linked to representation theory and
recently described by Deligne.
| [
{
"version": "v1",
"created": "Wed, 26 Feb 2025 19:47:05 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 02:10:16 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Broadhurst",
"David",
""
],
[
"Roulleau",
"Xavier",
""
]
] |
2502.19537 | Joshua Kazdan | Joshua Kazdan, Lisa Yu, Rylan Schaeffer, Chris Cundy, Sanmi Koyejo,
Krishnamurthy Dvijotham | No, of course I can! Refusal Mechanisms Can Be Exploited Using Harmless
Fine-Tuning Data | null | null | null | null | cs.CR cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Leading language model (LM) providers like OpenAI and Google offer
fine-tuning APIs that allow customers to adapt LMs for specific use cases. To
prevent misuse, these LM providers implement filtering mechanisms to block
harmful fine-tuning data. Consequently, adversaries seeking to produce unsafe
LMs via these APIs must craft adversarial training data that are not
identifiably harmful. We make three contributions in this context: 1. We show
that many existing attacks that use harmless data to create unsafe LMs rely on
eliminating model refusals in the first few tokens of their responses. 2. We
show that such prior attacks can be blocked by a simple defense that pre-fills
the first few tokens from an aligned model before letting the fine-tuned model
fill in the rest. 3. We describe a new data-poisoning attack, ``No, Of course I
Can Execute'' (NOICE), which exploits an LM's formulaic refusal mechanism to
elicit harmful responses. By training an LM to refuse benign requests on the
basis of safety before fulfilling those requests regardless, we are able to
jailbreak several open-source models and a closed-source model (GPT-4o). We
show an attack success rate (ASR) of 57% against GPT-4o; our attack earned a
Bug Bounty from OpenAI. Against open-source models protected by simple
defenses, we improve ASRs by an average of 3.25 times compared to the best
performing previous attacks that use only harmless data. NOICE demonstrates the
exploitability of repetitive refusal mechanisms and broadens understanding of
the threats closed-source models face from harmless data.
| [
{
"version": "v1",
"created": "Wed, 26 Feb 2025 20:20:01 GMT"
},
{
"version": "v2",
"created": "Wed, 19 Mar 2025 17:50:21 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Kazdan",
"Joshua",
""
],
[
"Yu",
"Lisa",
""
],
[
"Schaeffer",
"Rylan",
""
],
[
"Cundy",
"Chris",
""
],
[
"Koyejo",
"Sanmi",
""
],
[
"Dvijotham",
"Krishnamurthy",
""
]
] |
2502.19763 | Yang Sun | Chen Gao, Kai-Ming Ho, Renata M. Wentzcovitch, Yang Sun | Understanding the two-step nucleation of iron at Earth's inner core
conditions: a comparative molecular dynamics study | PRB in press | null | null | null | physics.geo-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Metastable phases can lead to multistep nucleation processes, influencing the
liquid-to-solid transition in various systems. In this study, we investigate
the homogeneous nucleation of iron's crystalline phases under Earth's inner
core conditions, employing two previously developed interatomic potentials. We
compare the thermodynamic and kinetic properties of iron relevant to the
nucleation as predicted by these potentials. While the potentials differ in
their predictions of melting temperature by a few hundred Kelvin, they show a
consistent description of the relative Gibbs free energy between solid and
liquid phases with respect to the undercooling. Both potentials also predict
that the metastable bcc phase exhibits a significantly higher nucleation rate
than the hcp phase over a wide range of undercooling temperatures below the
melting point. This substantially lowers the undercooling thresholds required
for the initial nucleation of Earth's inner core. The results validate the
commonality of the two-step nucleation mechanism of iron under Earth's inner
core conditions for two different potentials, providing a foundation for future
studies about the influence of other elements on the nucleation of Earth's
core.
| [
{
"version": "v1",
"created": "Thu, 27 Feb 2025 05:03:00 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 01:00:04 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Gao",
"Chen",
""
],
[
"Ho",
"Kai-Ming",
""
],
[
"Wentzcovitch",
"Renata M.",
""
],
[
"Sun",
"Yang",
""
]
] |
2502.20580 | Maher Hanut | Maher Hanut, Jonathan Kadmon | Training Large Neural Networks With Low-Dimensional Error Feedback | null | null | null | null | cs.LG q-bio.NC | http://creativecommons.org/licenses/by/4.0/ | Training deep neural networks typically relies on backpropagating high
dimensional error signals a computationally intensive process with little
evidence supporting its implementation in the brain. However, since most tasks
involve low-dimensional outputs, we propose that low-dimensional error signals
may suffice for effective learning. To test this hypothesis, we introduce a
novel local learning rule based on Feedback Alignment that leverages indirect,
low-dimensional error feedback to train large networks. Our method decouples
the backward pass from the forward pass, enabling precise control over error
signal dimensionality while maintaining high-dimensional representations. We
begin with a detailed theoretical derivation for linear networks, which forms
the foundation of our learning framework, and extend our approach to nonlinear,
convolutional, and transformer architectures. Remarkably, we demonstrate that
even minimal error dimensionality on the order of the task dimensionality can
achieve performance matching that of traditional backpropagation. Furthermore,
our rule enables efficient training of convolutional networks, which have
previously been resistant to Feedback Alignment methods, with minimal error.
This breakthrough not only paves the way toward more biologically accurate
models of learning but also challenges the conventional reliance on
high-dimensional gradient signals in neural network training. Our findings
suggest that low-dimensional error signals can be as effective as
high-dimensional ones, prompting a reevaluation of gradient-based learning in
high-dimensional systems. Ultimately, our work offers a fresh perspective on
neural network optimization and contributes to understanding learning
mechanisms in both artificial and biological systems.
| [
{
"version": "v1",
"created": "Thu, 27 Feb 2025 22:45:41 GMT"
},
{
"version": "v2",
"created": "Mon, 17 Mar 2025 12:41:58 GMT"
},
{
"version": "v3",
"created": "Thu, 20 Mar 2025 11:00:14 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Hanut",
"Maher",
""
],
[
"Kadmon",
"Jonathan",
""
]
] |
2502.20673 | Xiantao Li | Pegah Mohammadipour and Xiantao Li | Direct Analysis of Zero-Noise Extrapolation: Polynomial Methods, Error
Bounds, and Simultaneous Physical-Algorithmic Error Mitigation | null | null | null | null | quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Zero-noise extrapolation (ZNE) is a widely used quantum error mitigation
technique that artificially amplifies circuit noise and then extrapolates the
results to the noise-free circuit. A common ZNE approach is Richardson
extrapolation, which relies on polynomial interpolation. Despite its
simplicity, efficient implementations of Richardson extrapolation face several
challenges, including approximation errors from the non-polynomial behavior of
noise channels, overfitting due to polynomial interpolation, and exponentially
amplified measurement noise. This paper provides a comprehensive analysis of
these challenges, presenting bias and variance bounds that quantify
approximation errors. Additionally, for any precision $\varepsilon$, our
results offer an estimate of the necessary sample complexity. We further extend
the analysis to polynomial least squares-based extrapolation, which mitigates
measurement noise and avoids overfitting. Finally, we propose a strategy for
simultaneously mitigating circuit and algorithmic errors in the Trotter-Suzuki
algorithm by jointly scaling the time step size and the noise level. This
strategy provides a practical tool to enhance the reliability of near-term
quantum computations. We support our theoretical findings with numerical
experiments.
| [
{
"version": "v1",
"created": "Fri, 28 Feb 2025 02:59:44 GMT"
},
{
"version": "v2",
"created": "Wed, 19 Mar 2025 23:54:28 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Mohammadipour",
"Pegah",
""
],
[
"Li",
"Xiantao",
""
]
] |
2502.20881 | Andrea Cipriani | Andrea Cipriani, Alessandro De Santis, Giorgio Di Russo, Alfredo
Grillo, Luca Tabarroni | Hamiltonian Neural Networks approach to fuzzball geodesics | 25 pages + Appendices, 39 figures | null | null | null | hep-th cs.LG gr-qc | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The recent increase in computational resources and data availability has led
to a significant rise in the use of Machine Learning (ML) techniques for data
analysis in physics. However, the application of ML methods to solve
differential equations capable of describing even complex physical systems is
not yet fully widespread in theoretical high-energy physics. Hamiltonian Neural
Networks (HNNs) are tools that minimize a loss function defined to solve
Hamilton equations of motion. In this work, we implement several HNNs trained
to solve, with high accuracy, the Hamilton equations for a massless probe
moving inside a smooth and horizonless geometry known as D1-D5 circular
fuzzball. We study both planar (equatorial) and non-planar geodesics in
different regimes according to the impact parameter, some of which are
unstable. Our findings suggest that HNNs could eventually replace standard
numerical integrators, as they are equally accurate but more reliable in
critical situations.
| [
{
"version": "v1",
"created": "Fri, 28 Feb 2025 09:25:49 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 17:54:41 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Cipriani",
"Andrea",
""
],
[
"De Santis",
"Alessandro",
""
],
[
"Di Russo",
"Giorgio",
""
],
[
"Grillo",
"Alfredo",
""
],
[
"Tabarroni",
"Luca",
""
]
] |
2502.21001 | Woo Kyoung Han | Woo Kyoung Han, Byeonghun Lee, Hyunmin Cho, Sunghoon Im, Kyong Hwan
Jin | Towards Lossless Implicit Neural Representation via Bit Plane
Decomposition | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | We quantify the upper bound on the size of the implicit neural representation
(INR) model from a digital perspective. The upper bound of the model size
increases exponentially as the required bit-precision increases. To this end,
we present a bit-plane decomposition method that makes INR predict bit-planes,
producing the same effect as reducing the upper bound of the model size. We
validate our hypothesis that reducing the upper bound leads to faster
convergence with constant model size. Our method achieves lossless
representation in 2D image and audio fitting, even for high bit-depth signals,
such as 16-bit, which was previously unachievable. We pioneered the presence of
bit bias, which INR prioritizes as the most significant bit (MSB). We expand
the application of the INR task to bit depth expansion, lossless image
compression, and extreme network quantization. Our source code is available at
https://github.com/WooKyoungHan/LosslessINR
| [
{
"version": "v1",
"created": "Fri, 28 Feb 2025 12:43:46 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 10:50:43 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Han",
"Woo Kyoung",
""
],
[
"Lee",
"Byeonghun",
""
],
[
"Cho",
"Hyunmin",
""
],
[
"Im",
"Sunghoon",
""
],
[
"Jin",
"Kyong Hwan",
""
]
] |
2502.21010 | Xiaoli Hu | Jianming Zhou, Xiaoli Hu, Honglian Zhang and Naihuan Jing | Analytic Formulas for Quantum Discord of Special Families of N-Qubit
States | 14 pages, 3 figures | Quant. Inf. Process. (2025) 24:15 | 10.1007/s11128-024-04625-1 | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | Quantum discord, a key indicator of non-classical correlations in bipartite
systems, has been recently extended to multipartite scenarios [Phys. Rev. Lett.
2020, 124:110401]. We present exact analytic formulas for the quantum discord
of special families of N-qubit states, including generalized class of GHZ
states. Our formulations span $2$, $3$, $4n$, $4n+1$, $4n+2$, and $4n+3$-qubit
configurations where $n\in 1, 2, \ldots$, which refine the assessment of
quantum correlations and provide an analytical tool in quantum computation.
Moreover, we uncover a ``discord freezing'' in even-qubit systems under phase
flip decoherence which provides a means for preserving quantum coherence in
environmental perturbations.
| [
{
"version": "v1",
"created": "Fri, 28 Feb 2025 12:55:16 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Zhou",
"Jianming",
""
],
[
"Hu",
"Xiaoli",
""
],
[
"Zhang",
"Honglian",
""
],
[
"Jing",
"Naihuan",
""
]
] |
2502.21114 | Krzysztof Podlaski | Krzysztof Podlaski, Michal Beczkowski, Katharina Simbeck, Katrin
Dziergwa, Derek O'Reilly, Shane Dowdall, Joao Monteiro, Catarina Oliveira
Lucas, Johanna Hautamaki, Heikki Ahonen, Hiram Bollaert, Philippe Possemiers,
and Zofia Stawska | Software development projects as a way for multidisciplinary soft and
future skills education | null | null | null | null | cs.CY | http://creativecommons.org/licenses/by/4.0/ | Soft and future skills are in high demand in the modern job market. These
skills are required for both technical and non-technical people. It is
difficult to teach these competencies in a classical academic environment.
The paper presents a possible approach to teaching in soft and future skills
in a short, intensive joint project. In our case, it is a project within the
Erasmus+ framework, but it can be organized in many different frameworks.
In the project we use problem based learning, active learning and group-work
teaching methodologies. Moreover, the approach put high emphasizes diversity.
We arrange a set of multidisciplinary students in groups. Each group is working
on software development tasks. This type of projects demand diversity, and only
a part of the team needs technical skills. In our case less than half of
participants had computer science background. Additionally, software
development projects are usually interesting for non-technical students.
The multicultural, multidisciplinary and international aspects are very
important in a modern global working environment. On the other hand, short time
of the project and its intensity allow to simulate stressful situations in a
real word tasks. The effects of the project on the required competencies are
measured using the KYSS method.
The results prove that the presented method increased participants soft
skills in communication, cooperation, digital skills and self reflection.
| [
{
"version": "v1",
"created": "Fri, 28 Feb 2025 14:52:40 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 15:42:53 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Podlaski",
"Krzysztof",
""
],
[
"Beczkowski",
"Michal",
""
],
[
"Simbeck",
"Katharina",
""
],
[
"Dziergwa",
"Katrin",
""
],
[
"O'Reilly",
"Derek",
""
],
[
"Dowdall",
"Shane",
""
],
[
"Monteiro",
"Joao",
""
],
[
"Lucas",
"Catarina Oliveira",
""
],
[
"Hautamaki",
"Johanna",
""
],
[
"Ahonen",
"Heikki",
""
],
[
"Bollaert",
"Hiram",
""
],
[
"Possemiers",
"Philippe",
""
],
[
"Stawska",
"Zofia",
""
]
] |
2503.00057 | Ranjan Sapkota | Ranjan Sapkota, Manoj Karkee | Improved YOLOv12 with LLM-Generated Synthetic Data for Enhanced Apple
Detection and Benchmarking Against YOLOv11 and YOLOv10 | 8 pages, 5 Figures, 2 Tables | null | null | null | cs.CV cs.CL | http://creativecommons.org/licenses/by/4.0/ | This study evaluated the performance of the YOLOv12 object detection model,
and compared against the performances YOLOv11 and YOLOv10 for apple detection
in commercial orchards based on the model training completed entirely on
synthetic images generated by Large Language Models (LLMs). The YOLOv12n
configuration achieved the highest precision at 0.916, the highest recall at
0.969, and the highest mean Average Precision (mAP@50) at 0.978. In comparison,
the YOLOv11 series was led by YOLO11x, which achieved the highest precision at
0.857, recall at 0.85, and mAP@50 at 0.91. For the YOLOv10 series, YOLOv10b and
YOLOv10l both achieved the highest precision at 0.85, with YOLOv10n achieving
the highest recall at 0.8 and mAP@50 at 0.89. These findings demonstrated that
YOLOv12, when trained on realistic LLM-generated datasets surpassed its
predecessors in key performance metrics. The technique also offered a
cost-effective solution by reducing the need for extensive manual data
collection in the agricultural field. In addition, this study compared the
computational efficiency of all versions of YOLOv12, v11 and v10, where
YOLOv11n reported the lowest inference time at 4.7 ms, compared to YOLOv12n's
5.6 ms and YOLOv10n's 5.9 ms. Although YOLOv12 is new and more accurate than
YOLOv11, and YOLOv10, YOLO11n still stays the fastest YOLO model among YOLOv10,
YOLOv11 and YOLOv12 series of models. (Index: YOLOv12, YOLOv11, YOLOv10,
YOLOv13, YOLOv14, YOLOv15, YOLOE, YOLO Object detection)
| [
{
"version": "v1",
"created": "Wed, 26 Feb 2025 20:24:01 GMT"
},
{
"version": "v2",
"created": "Wed, 19 Mar 2025 18:04:39 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Sapkota",
"Ranjan",
""
],
[
"Karkee",
"Manoj",
""
]
] |
2503.00371 | Xuehao Gao | Xuehao Gao, Yang Yang, Shaoyi Du, Guo-Jun Qi, Junwei Han | Jointly Understand Your Command and Intention:Reciprocal Co-Evolution
between Scene-Aware 3D Human Motion Synthesis and Analysis | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | As two intimate reciprocal tasks, scene-aware human motion synthesis and
analysis require a joint understanding between multiple modalities, including
3D body motions, 3D scenes, and textual descriptions. In this paper, we
integrate these two paired processes into a Co-Evolving Synthesis-Analysis
(CESA) pipeline and mutually benefit their learning. Specifically, scene-aware
text-to-human synthesis generates diverse indoor motion samples from the same
textual description to enrich human-scene interaction intra-class diversity,
thus significantly benefiting training a robust human motion analysis system.
Reciprocally, human motion analysis would enforce semantic scrutiny on each
synthesized motion sample to ensure its semantic consistency with the given
textual description, thus improving realistic motion synthesis. Considering
that real-world indoor human motions are goal-oriented and path-guided, we
propose a cascaded generation strategy that factorizes text-driven
scene-specific human motion generation into three stages: goal inferring, path
planning, and pose synthesizing. Coupling CESA with this powerful cascaded
motion synthesis model, we jointly improve realistic human motion synthesis and
robust human motion analysis in 3D scenes.
| [
{
"version": "v1",
"created": "Sat, 1 Mar 2025 06:56:58 GMT"
},
{
"version": "v2",
"created": "Fri, 7 Mar 2025 01:20:01 GMT"
},
{
"version": "v3",
"created": "Thu, 20 Mar 2025 13:20:09 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Gao",
"Xuehao",
""
],
[
"Yang",
"Yang",
""
],
[
"Du",
"Shaoyi",
""
],
[
"Qi",
"Guo-Jun",
""
],
[
"Han",
"Junwei",
""
]
] |
2503.00705 | Evangelos Paouris | Evangelos Paouris, Angelos Vourlidas, Manolis K. Georgoulis, Phillip
Hess, Guillermo Stenborg | How the CME on 21 April 2023 Triggered the First Severe Geomagnetic
Storm of Solar Cycle 25 | 20 pages, 9 Figures | null | null | null | astro-ph.SR astro-ph.EP physics.space-ph | http://creativecommons.org/licenses/by/4.0/ | The first severe (G4) geomagnetic storm of Solar Cycle 25 occurred on 23-24
April 2023, following the arrival of a Coronal Mass Ejection (CME) on 23 April.
The characteristics of this CME, measured from coronagraphs (speed and mass),
did not indicate that it would trigger such an intense geomagnetic storm. In
this work, our aim is to understand why this CME led to such a geoeffective
outcome. Our analysis spans from the source active region to the corona and
inner heliosphere through 1 au using multiwavelength, multi-viewpoint remote
sensing observations and in situ data. We find that rotation and possibly
deflection of the CME resulted in an axial magnetic field nearly parallel to
the ecliptic plane during the Earth encounter, which might explain the storm's
severity. Additionally, we find that imaging away from the Sun-Earth line is
crucial in hindcasting the CME Time-of-Arrival at Earth. The position (0.39 au)
and detailed images from the SoloHI telescope onboard the Solar Orbiter
mission, in combination with SOHO and STEREO images, helped decisively with the
three-dimensional (3D) reconstruction of the CME.
| [
{
"version": "v1",
"created": "Sun, 2 Mar 2025 03:04:06 GMT"
},
{
"version": "v2",
"created": "Wed, 19 Mar 2025 18:56:35 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Paouris",
"Evangelos",
""
],
[
"Vourlidas",
"Angelos",
""
],
[
"Georgoulis",
"Manolis K.",
""
],
[
"Hess",
"Phillip",
""
],
[
"Stenborg",
"Guillermo",
""
]
] |
2503.00861 | Taewoong Kang | Sohyun Jeong, Taewoong Kang, Hyojin Jang and Jaegul Choo | Zero-Shot Head Swapping in Real-World Scenarios | CVPR'25 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | With growing demand in media and social networks for personalized images, the
need for advanced head-swapping techniques, integrating an entire head from the
head image with the body from the body image, has increased. However,
traditional head swapping methods heavily rely on face-centered cropped data
with primarily frontal facing views, which limits their effectiveness in real
world applications. Additionally, their masking methods, designed to indicate
regions requiring editing, are optimized for these types of dataset but
struggle to achieve seamless blending in complex situations, such as when the
original data includes features like long hair extending beyond the masked
area. To overcome these limitations and enhance adaptability in diverse and
complex scenarios, we propose a novel head swapping method, HID, that is robust
to images including the full head and the upper body, and handles from frontal
to side views, while automatically generating context aware masks. For
automatic mask generation, we introduce the IOMask, which enables seamless
blending of the head and body, effectively addressing integration challenges.
We further introduce the hair injection module to capture hair details with
greater precision. Our experiments demonstrate that the proposed approach
achieves state-of-the-art performance in head swapping, providing visually
consistent and realistic results across a wide range of challenging conditions.
| [
{
"version": "v1",
"created": "Sun, 2 Mar 2025 11:44:23 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 04:38:17 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Jeong",
"Sohyun",
""
],
[
"Kang",
"Taewoong",
""
],
[
"Jang",
"Hyojin",
""
],
[
"Choo",
"Jaegul",
""
]
] |
2503.01130 | Runmao Yao | Runmao Yao, Yi Du, Zhuoqun Chen, Haoze Zheng, Chen Wang | AirRoom: Objects Matter in Room Reidentification | Paper accepted at CVPR 2025 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Room reidentification (ReID) is a challenging yet essential task with
numerous applications in fields such as augmented reality (AR) and homecare
robotics. Existing visual place recognition (VPR) methods, which typically rely
on global descriptors or aggregate local features, often struggle in cluttered
indoor environments densely populated with man-made objects. These methods tend
to overlook the crucial role of object-oriented information. To address this,
we propose AirRoom, an object-aware pipeline that integrates multi-level
object-oriented information-from global context to object patches, object
segmentation, and keypoints-utilizing a coarse-to-fine retrieval approach.
Extensive experiments on four newly constructed datasets-MPReID, HMReID,
GibsonReID, and ReplicaReID-demonstrate that AirRoom outperforms
state-of-the-art (SOTA) models across nearly all evaluation metrics, with
improvements ranging from 6% to 80%. Moreover, AirRoom exhibits significant
flexibility, allowing various modules within the pipeline to be substituted
with different alternatives without compromising overall performance. It also
shows robust and consistent performance under diverse viewpoint variations.
| [
{
"version": "v1",
"created": "Mon, 3 Mar 2025 03:20:08 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 01:13:23 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Yao",
"Runmao",
""
],
[
"Du",
"Yi",
""
],
[
"Chen",
"Zhuoqun",
""
],
[
"Zheng",
"Haoze",
""
],
[
"Wang",
"Chen",
""
]
] |
2503.01196 | Alain Moise Dikande Pr. | Alain M. Dikand\'e | On a hyperbolic Duffing oscillator with linear damping and periodic
forcing | 18 pages, 34 figures | null | null | null | nlin.CD math-ph math.MP physics.comp-ph | http://creativecommons.org/licenses/by/4.0/ | The Duffing oscillator describes the dynamics of a mass suspended on a spring
with position-dependent stiffness. The mass is assumed to experience a linear
damping and a time-dependent external forcing. The model has been instrumental
in theoretical investigations of dynamical properties of systems with
parity-conserving symmetry, where a double-well substrate connects two
metastable states separated by a barrier. Physical systems of interest include
nonlinear feedback-controlled mass-spring-damper oscillators, active hysteresis
circuits (e.g. memristors), protein chains prone to hydrogen bond-mediated
conformational transitions, centro-symmetric crystals and so on. In this work
we consider a Duffing-type oscillator with a double-well potential represented
by a hyperbolic function of mass position. The hyperbolic double-well potential
has two degenerate minima that can be smoothly tuned by varying a deformability
parameter, leaving unchanged the barrier height. We investigate solutions of
the equation of motion in the absence and presence of damping and forcing. In
the absence of perturbations numerical solutions lead to a periodic train of
anharmonic oscillations featuring a crystal of pulse solitons of sech types.
However, when the hyperbolic double-well potential is inverted, analytical
solutions can be obtained which turn out to be kink-soliton crystals described
by Jacobi elliptic functions. When damping and forcing are taken into
consideration, the system dynamics can transit from periodic to chaotic phases
or vice-versa via period-doubling or period-halving bifurcations, by simply
varying the deformability parameter. The Poincar\'e map of the proposed model
carries the well-known characteristic signatures of chaos presursors of the
standard Duffing model, which happens to be just a particular case of the
bistable oscillator model with the hyperbolic double-well potential.
| [
{
"version": "v1",
"created": "Mon, 3 Mar 2025 05:47:46 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 16:59:24 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Dikandé",
"Alain M.",
""
]
] |
2503.01448 | Xiangjun Tang | Xiangjun Tang, Biao Zhang and Peter Wonka | Generative Human Geometry Distribution | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Realistic human geometry generation is an important yet challenging task,
requiring both the preservation of fine clothing details and the accurate
modeling of clothing-pose interactions. Geometry distributions, which can model
the geometry of a single human as a distribution, provide a promising
representation for high-fidelity synthesis. However, applying geometry
distributions for human generation requires learning a dataset-level
distribution over numerous individual geometry distributions. To address the
resulting challenges, we propose a novel 3D human generative framework that,
for the first time, models the distribution of human geometry distributions.
Our framework operates in two stages: first, generating the human geometry
distribution, and second, synthesizing high-fidelity humans by sampling from
this distribution. We validate our method on two tasks: pose-conditioned 3D
human generation and single-view-based novel pose generation. Experimental
results demonstrate that our approach achieves the best quantitative results in
terms of realism and geometric fidelity, outperforming state-of-the-art
generative methods.
| [
{
"version": "v1",
"created": "Mon, 3 Mar 2025 11:55:19 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 08:48:44 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Tang",
"Xiangjun",
""
],
[
"Zhang",
"Biao",
""
],
[
"Wonka",
"Peter",
""
]
] |
2503.01478 | Yijie Xu | Lu Dai, Yijie Xu, Jinhui Ye, Hao Liu, Hui Xiong | SePer: Measure Retrieval Utility Through The Lens Of Semantic Perplexity
Reduction | ICLR 2025 Spotlight | null | null | null | cs.CL cs.AI cs.LG | http://creativecommons.org/licenses/by-sa/4.0/ | Large Language Models (LLMs) have demonstrated improved generation
performance by incorporating externally retrieved knowledge, a process known as
retrieval-augmented generation (RAG). Despite the potential of this approach,
existing studies evaluate RAG effectiveness by 1) assessing retrieval and
generation components jointly, which obscures retrieval's distinct
contribution, or 2) examining retrievers using traditional metrics such as
NDCG, which creates a gap in understanding retrieval's true utility in the
overall generation process. To address the above limitations, in this work, we
introduce an automatic evaluation method that measures retrieval quality
through the lens of information gain within the RAG framework. Specifically, we
propose Semantic Perplexity (SePer), a metric that captures the LLM's internal
belief about the correctness of the retrieved information. We quantify the
utility of retrieval by the extent to which it reduces semantic perplexity
post-retrieval. Extensive experiments demonstrate that SePer not only aligns
closely with human preferences but also offers a more precise and efficient
evaluation of retrieval utility across diverse RAG scenarios.
| [
{
"version": "v1",
"created": "Mon, 3 Mar 2025 12:37:34 GMT"
},
{
"version": "v2",
"created": "Tue, 4 Mar 2025 07:51:56 GMT"
},
{
"version": "v3",
"created": "Wed, 5 Mar 2025 05:24:54 GMT"
},
{
"version": "v4",
"created": "Wed, 12 Mar 2025 08:49:58 GMT"
},
{
"version": "v5",
"created": "Thu, 20 Mar 2025 11:28:41 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Dai",
"Lu",
""
],
[
"Xu",
"Yijie",
""
],
[
"Ye",
"Jinhui",
""
],
[
"Liu",
"Hao",
""
],
[
"Xiong",
"Hui",
""
]
] |
2503.01697 | Da-Jian Zhang | Da-Jian Zhang and D. M. Tong | Krylov shadow tomography: Efficient estimation of quantum Fisher
information | To be published in Phys. Rev. Lett | Phys. Rev. Lett. 134, 110802 (2025) | 10.1103/PhysRevLett.134.110802 | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | Efficiently estimating the quantum Fisher information (QFI) is pivotal in
quantum information science but remains an outstanding challenge for large
systems due to its high nonlinearity. In this Letter, we tackle this
long-standing challenge by integrating the Krylov subspace method--a celebrated
tool from applied mathematics--into the framework of shadow tomography. The
integrated technique, dubbed Krylov shadow tomography (KST), enables us to
formulate a strict hierarchy of non-polynomial lower bounds on the QFI, among
which the highest one matches the QFI exactly. We show that all the bounds can
be expressed as expected values of the inverses of Hankel matrices, which are
accessible via shadow tomography. Our KST therefore opens up a
resource-efficient and experimentally feasible avenue to estimate not only
non-polynomial lower bounds but also the QFI itself.
| [
{
"version": "v1",
"created": "Mon, 3 Mar 2025 16:10:30 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Zhang",
"Da-Jian",
""
],
[
"Tong",
"D. M.",
""
]
] |
2503.01754 | Guande Wu | Guande Wu, Huan Song, Yawei Wang, Qiaojing Yan, Yijun Tian, Lin Lee
Cheong, Panpan Xu | SDRT: Enhance Vision-Language Models by Self-Distillation with Diverse
Reasoning Traces | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Reasoning is increasingly crucial for various tasks. While chain-of-thought
prompting enables large language models to leverage reasoning effectively,
harnessing the reasoning capabilities of Vision-Language Models (VLMs) remains
challenging. To solve this problem, we propose a novel self-distillation
framework that enhances the reasoning capabilities of the model. The proposed
framework introduces several key innovations. We start by employing a prompt
library tailored to visual reasoning tasks to generate diverse in-context
questions and utilize a two-step reasoning procedure to derive reasoning-guided
responses. These responses are then used for self-distillation, enabling the
model to internalize the reasoning process. Additionally, we improve the model
architecture with several innovative components, including an intervention
adapter for efficient parameter updates, a cross-modal skip connection to
facilitate information exchange between modalities, and an ensemble learning
algorithm to integrate diverse reasoning from multiple in-context questions.
Extensive experiments show that our method significantly improves the baseline
performance across five VQA datasets.
| [
{
"version": "v1",
"created": "Mon, 3 Mar 2025 17:24:42 GMT"
},
{
"version": "v2",
"created": "Tue, 18 Mar 2025 08:05:25 GMT"
},
{
"version": "v3",
"created": "Wed, 19 Mar 2025 18:35:44 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Wu",
"Guande",
""
],
[
"Song",
"Huan",
""
],
[
"Wang",
"Yawei",
""
],
[
"Yan",
"Qiaojing",
""
],
[
"Tian",
"Yijun",
""
],
[
"Cheong",
"Lin Lee",
""
],
[
"Xu",
"Panpan",
""
]
] |
2503.01761 | Seung Suk Kang | Steven Seung-Suk Kang, Joseph Bodenheimer, Kayley Morris, Tracey
Butler | A comprehensive and reliable protocol for manual segmentation of the
human claustrum using high-resolution MRI | null | null | null | null | q-bio.NC | http://creativecommons.org/licenses/by/4.0/ | The claustrum is a thin gray matter structure in each brain hemisphere,
characterized by exceptionally high connectivity with nearly all brain regions.
Despite extensive animal studies on its anatomy and function and growing
evidence of claustral deficits in neuropsychiatric disorders, its specific
roles in normal and abnormal human brain function remain largely unknown. This
is primarily due to its thin and complex morphology, which limits accurate
anatomical delineation and neural activity isolation in conventional in vivo
neuroimaging. To facilitate future neuroimaging studies, we developed a
comprehensive and reliable manual segmentation protocol based on a
cellular-resolution brain atlas and high-resolution (0.7^3 mm) MRI data. The
protocols involve detailed guidelines to delineate the entire claustrum,
including the inferior parts that have not been clearly described in earlier
MRI studies. Additionally, we propose a geometric method to parcellate the
claustrum into three subregions (the dorsal, ventral, and temporal claustrum)
along the superior-to-inferior axis. The mean bilateral claustrum volume in 10
young adults was 3307.5 mm^3, approximately 0.21% of total intracranial volume.
Our segmentation protocol demonstrated high inter- and intra-rater reliability
(ICC > 0.89, DSC > 0.85), confirming its replicability. This comprehensive and
reliable claustrum segmentation protocols will provide a cornerstone for future
neuroimaging studies of systematic, large-scale investigations of the anatomy
and the functions of the human claustrum in normal and pathological
populations.
| [
{
"version": "v1",
"created": "Mon, 3 Mar 2025 17:34:47 GMT"
},
{
"version": "v2",
"created": "Wed, 19 Mar 2025 23:30:33 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Kang",
"Steven Seung-Suk",
""
],
[
"Bodenheimer",
"Joseph",
""
],
[
"Morris",
"Kayley",
""
],
[
"Butler",
"Tracey",
""
]
] |
2503.01924 | Yuhang Wang | Wang YuHang, Junkang Guo, Aolei Liu, Kaihao Wang, Zaitong Wu, Zhenyu
Liu, Wenfei Yin, Jian Liu | TAET: Two-Stage Adversarial Equalization Training on Long-Tailed
Distributions | Text: 8 pages of main content, 5 pages of appendices have been
accepted by CVPR2025 | null | null | null | cs.LG cs.AI stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Adversarial robustness is a critical challenge in deploying deep neural
networks for real-world applications. While adversarial training is a widely
recognized defense strategy, most existing studies focus on balanced datasets,
overlooking the prevalence of long-tailed distributions in real-world data,
which significantly complicates robustness. This paper provides a comprehensive
analysis of adversarial training under long-tailed distributions and identifies
limitations in the current state-of-the-art method, AT-BSL, in achieving robust
performance under such conditions. To address these challenges, we propose a
novel training framework, TAET, which integrates an initial stabilization phase
followed by a stratified equalization adversarial training phase. Additionally,
prior work on long-tailed robustness has largely ignored the crucial evaluation
metric of balanced accuracy. To bridge this gap, we introduce the concept of
balanced robustness, a comprehensive metric tailored for assessing robustness
under long-tailed distributions. Extensive experiments demonstrate that our
method surpasses existing advanced defenses, achieving significant improvements
in both memory and computational efficiency. This work represents a substantial
advancement in addressing robustness challenges in real-world applications. Our
code is available at:
https://github.com/BuhuiOK/TAET-Two-Stage-Adversarial-Equalization-Training-on-Long-Tailed-Distributions.
| [
{
"version": "v1",
"created": "Sun, 2 Mar 2025 12:07:00 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 08:49:42 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"YuHang",
"Wang",
""
],
[
"Guo",
"Junkang",
""
],
[
"Liu",
"Aolei",
""
],
[
"Wang",
"Kaihao",
""
],
[
"Wu",
"Zaitong",
""
],
[
"Liu",
"Zhenyu",
""
],
[
"Yin",
"Wenfei",
""
],
[
"Liu",
"Jian",
""
]
] |
2503.02291 | Songting Li | Songting Li, Wenting Wang, Sergey E. Koposov, Ting S. Li, Youjia Wu,
Monica Valluri, Joan Najita, Carlos Allende Prieto, Amanda Bystr\"om,
Christopher J. Manser, Jiaxin Han, Carles G. Palau, Hao Yang, Andrew P.
Cooper, Namitha Kizhuprakkat, Alexander H.Riley, Jessica Nicole Aguilar,
Steven Ahlen, David Bianchi, David Brooks, Todd Claybaugh, Axel de la
Macorra, John Della Costa, Arjun Dey, Peter Doel, Jaime E. Forero-Romero,
Enrique Gazta\~naga, Satya Gontcho A Gontcho, Gaston Gutierrez, Klaus
Honscheid, Mustapha Ishak, Stephanie Juneau, Robert Kehoe, Theodore Kisner,
Martin Landriau, Laurent Le Guillou, Michael Levi, Marc Manera, Aaron
Meisner, Ramon Miquel, John Moustakas, Nathalie Palanque-Delabrouille, Will
Percival, Claire Poppett, Francisco Prada, Ignasi P\'erez-R\`afols, Graziano
Rossi, Eusebio Sanchez, David Schlegel, Michael Schubnell, Hee-Jong Seo,
Joseph Harry Silber, David Sprayberry, Gregory Tarl\'e, Benjamin Alan Weaver,
Rongpu Zhou, Hu Zou | SpecDis: Value added distance catalogue for 4 million stars from DESI
Year-1 data | 24 pages,20 figures,2 tables This version updates the correct file
link of VAC | null | null | null | astro-ph.GA astro-ph.SR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present the SpecDis value added stellar distance catalogue accompanying
DESI DR1. SpecDis trains a feed-forward Neural Network (NN) on a large sample
of stars with Gaia parallaxes, but without applying selections on parallax
error or signal-to-noise (S/N) of the stellar spectra. We incorporate parallax
error into the loss function for training. This approach ensures the training
sample not suffering from biases. Moreover, SpecDis predicts the reciprocal of
the square root of luminosity, which is linearly proportional to parallax and
helps to avoid excluding negative parallaxes. To enhance the precision of
distance predictions, we employ Principal Component Analysis (PCA) to reduce
the noise and dimensionality of stellar spectra. Validated by independent
external samples of member stars with precise distances from globular clusters,
dwarf galaxies, and stellar streams, combined with BHB stars, we demonstrate
that our distance measurements show no significant bias up to 100 kpc, and are
much more precise than Gaia parallax beyond 7 kpc. The median distance
uncertainties are 23 %, 19 %, 11 % and 7 % for S/N$<$20, 20$\leq$S/N$<$ 60,
60$\leq$ S/N $<$ 100 and S/N$\geq$100. Selecting stars with $\log g<3.8$ and
distance uncertainties smaller than 25 %, we have more than 74,000 giant
candidates within 50 kpc to the Galactic center and 1,500 candidates beyond
this distance. Additionally, we develop a Gaussian mixture model to identify
binaries and identify 120,000 possible binaries, and discover that the binary
fraction increases with [Fe/H] and $\log g$ and declines with [$\alpha$/Fe] and
$T_\mathrm{eff}$, indicating stars with low Fe and high $\alpha$, which form
early, may have experienced more encounters and tidal effects to disrupt
binaries. Our final catalogue provides distances and distance uncertainties for
$>$4 million stars, offering a valuable resource for Galactic astronomy.
| [
{
"version": "v1",
"created": "Tue, 4 Mar 2025 05:30:22 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 04:55:53 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Li",
"Songting",
""
],
[
"Wang",
"Wenting",
""
],
[
"Koposov",
"Sergey E.",
""
],
[
"Li",
"Ting S.",
""
],
[
"Wu",
"Youjia",
""
],
[
"Valluri",
"Monica",
""
],
[
"Najita",
"Joan",
""
],
[
"Prieto",
"Carlos Allende",
""
],
[
"Byström",
"Amanda",
""
],
[
"Manser",
"Christopher J.",
""
],
[
"Han",
"Jiaxin",
""
],
[
"Palau",
"Carles G.",
""
],
[
"Yang",
"Hao",
""
],
[
"Cooper",
"Andrew P.",
""
],
[
"Kizhuprakkat",
"Namitha",
""
],
[
"Riley",
"Alexander H.",
""
],
[
"Aguilar",
"Jessica Nicole",
""
],
[
"Ahlen",
"Steven",
""
],
[
"Bianchi",
"David",
""
],
[
"Brooks",
"David",
""
],
[
"Claybaugh",
"Todd",
""
],
[
"de la Macorra",
"Axel",
""
],
[
"Della Costa",
"John",
""
],
[
"Dey",
"Arjun",
""
],
[
"Doel",
"Peter",
""
],
[
"Forero-Romero",
"Jaime E.",
""
],
[
"Gaztañaga",
"Enrique",
""
],
[
"Gontcho",
"Satya Gontcho A",
""
],
[
"Gutierrez",
"Gaston",
""
],
[
"Honscheid",
"Klaus",
""
],
[
"Ishak",
"Mustapha",
""
],
[
"Juneau",
"Stephanie",
""
],
[
"Kehoe",
"Robert",
""
],
[
"Kisner",
"Theodore",
""
],
[
"Landriau",
"Martin",
""
],
[
"Guillou",
"Laurent Le",
""
],
[
"Levi",
"Michael",
""
],
[
"Manera",
"Marc",
""
],
[
"Meisner",
"Aaron",
""
],
[
"Miquel",
"Ramon",
""
],
[
"Moustakas",
"John",
""
],
[
"Palanque-Delabrouille",
"Nathalie",
""
],
[
"Percival",
"Will",
""
],
[
"Poppett",
"Claire",
""
],
[
"Prada",
"Francisco",
""
],
[
"Pérez-Ràfols",
"Ignasi",
""
],
[
"Rossi",
"Graziano",
""
],
[
"Sanchez",
"Eusebio",
""
],
[
"Schlegel",
"David",
""
],
[
"Schubnell",
"Michael",
""
],
[
"Seo",
"Hee-Jong",
""
],
[
"Silber",
"Joseph Harry",
""
],
[
"Sprayberry",
"David",
""
],
[
"Tarlé",
"Gregory",
""
],
[
"Weaver",
"Benjamin Alan",
""
],
[
"Zhou",
"Rongpu",
""
],
[
"Zou",
"Hu",
""
]
] |
2503.02295 | Ai-Chao Wang | Ai-Chao Wang, Neng-Chang Wei, Fei Huang | Comprehensive analysis of the $\gamma p \to K^+ \Sigma^0(1385)$, $\gamma
n \to K^+ \Sigma^-(1385)$, and $\pi^+ p \to K^+ \Sigma^+(1385)$ reactions | null | null | null | null | hep-ph nucl-th | http://creativecommons.org/licenses/by/4.0/ | A comprehensive analysis of all available data on cross sections and
spin-dependent observables for the $\gamma p \to K^+ \Sigma^0(1385)$, $\gamma n
\to K^+ \Sigma^-(1385)$, and $\pi^+ p \to K^+ \Sigma^+(1385)$ reactions is
performed within an effective Lagrangian framework. In addition to the
$s$-channel $N$ exchange, $t$-channel $K$ and $K^\ast$ exchanges, $u$-channel
$\Lambda$, $\Sigma$, and $\Sigma^\ast$ exchanges, and the interaction current,
the $s$-channel contributions from a minimal set of $N$ and $\Delta$ resonances
are included in constructing the reaction amplitudes to describe the data. The
analysis reveals that all available data for both photon- and hadron-induced
$K\Sigma(1385)$ production reactions can be simultaneously and satisfactorily
reproduced when the contribution of the $s$-channel $\Delta(1930){5/2}^-$
resonance is taken into account. The reaction mechanisms, particularly the
resonance contributions in each considered reaction, are thoroughly analyzed
and discussed in detail.
| [
{
"version": "v1",
"created": "Tue, 4 Mar 2025 05:36:14 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Wang",
"Ai-Chao",
""
],
[
"Wei",
"Neng-Chang",
""
],
[
"Huang",
"Fei",
""
]
] |
2503.02593 | Yanlong Xu | Yanlong Xu, Haoxuan Qu, Jun Liu, Wenxiao Zhang, Xun Yang | CMMLoc: Advancing Text-to-PointCloud Localization with
Cauchy-Mixture-Model Based Framework | Accepted by CVPR 2025 | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | The goal of point cloud localization based on linguistic description is to
identify a 3D position using textual description in large urban environments,
which has potential applications in various fields, such as determining the
location for vehicle pickup or goods delivery. Ideally, for a textual
description and its corresponding 3D location, the objects around the 3D
location should be fully described in the text description. However, in
practical scenarios, e.g., vehicle pickup, passengers usually describe only the
part of the most significant and nearby surroundings instead of the entire
environment. In response to this $\textbf{partially relevant}$ challenge, we
propose $\textbf{CMMLoc}$, an uncertainty-aware
$\textbf{C}$auchy-$\textbf{M}$ixture-$\textbf{M}$odel ($\textbf{CMM}$) based
framework for text-to-point-cloud $\textbf{Loc}$alization. To model the
uncertain semantic relations between text and point cloud, we integrate CMM
constraints as a prior during the interaction between the two modalities. We
further design a spatial consolidation scheme to enable adaptive aggregation of
different 3D objects with varying receptive fields. To achieve precise
localization, we propose a cardinal direction integration module alongside a
modality pre-alignment strategy, helping capture the spatial relationships
among objects and bringing the 3D objects closer to the text modality.
Comprehensive experiments validate that CMMLoc outperforms existing methods,
achieving state-of-the-art results on the KITTI360Pose dataset. Codes are
available in this GitHub repository https://github.com/kevin301342/CMMLoc.
| [
{
"version": "v1",
"created": "Tue, 4 Mar 2025 13:17:17 GMT"
},
{
"version": "v2",
"created": "Wed, 5 Mar 2025 02:11:25 GMT"
},
{
"version": "v3",
"created": "Thu, 20 Mar 2025 00:06:14 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Xu",
"Yanlong",
""
],
[
"Qu",
"Haoxuan",
""
],
[
"Liu",
"Jun",
""
],
[
"Zhang",
"Wenxiao",
""
],
[
"Yang",
"Xun",
""
]
] |
2503.03051 | Shyam Mohan Subbiah Pillai | Nadhir Ben Rached, Shyam Mohan Subbiah Pillai, Ra\'ul Tempone | Optimal power procurement for green cellular wireless networks under
uncertainty and chance constraints | null | null | 10.3390/e27030308 | null | math.OC stat.AP | http://creativecommons.org/licenses/by/4.0/ | Given the increasing global emphasis on sustainable energy usage and the
rising energy demands of cellular wireless networks, this work seeks an optimal
short-term, continuous-time power procurement schedule to minimize operating
expenditure and the carbon footprint of cellular wireless networks equipped
with energy storage capacity, and hybrid energy systems comprising uncertain
renewable energy sources. Despite the stochastic nature of wireless fading
channels, the network operator must ensure a certain quality-of-service (QoS)
constraint with high probability. This probabilistic constraint prevents using
the dynamic programming principle to solve the stochastic optimal control
problem. This work introduces a novel time-continuous Lagrangian relaxation
approach tailored for real-time, near-optimal energy procurement in cellular
networks, overcoming tractability problems associated with the probabilistic
QoS constraint. The numerical solution procedure includes an efficient upwind
finite-difference solver for the Hamilton--Jacobi--Bellman equation
corresponding to the relaxed problem, and an effective combination of the
limited memory bundle method (LMBM) for handling nonsmooth optimization and the
stochastic subgradient method (SSM) to navigate the stochasticity of the dual
problem. Numerical results, based on the German power system and daily cellular
traffic data, demonstrate the computational efficiency of the proposed
numerical approach, providing a near-optimal policy in a practical timeframe.
| [
{
"version": "v1",
"created": "Tue, 4 Mar 2025 23:30:23 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Rached",
"Nadhir Ben",
""
],
[
"Pillai",
"Shyam Mohan Subbiah",
""
],
[
"Tempone",
"Raúl",
""
]
] |
2503.03376 | Grzegorz Centa{\l}a | Grzegorz Centa{\l}a and Jaros{\l}aw W. K{\l}os | Magneto-rotation coupling for ferromagnetic nanoelement embedded in
elastic substrate | null | null | null | null | cond-mat.mes-hall | http://creativecommons.org/licenses/by/4.0/ | This study investigates magneto-rotational coupling as a distinct
contribution to magnetoelastic interactions, which can be influenced by
magnetic anisotropy. We determine magneto-rotational coupling coefficients that
incorporate the shape anisotropy of a magnetic nanoelement (strip) and
demonstrate that this type of coupling can be modified through geometric
adjustments. Furthermore, we analyze the magneto-rotational contribution to the
magnetoelastic field in a ferromagnetic strip embedded in a nonmagnetic
substrate. Both Rayleigh and Love waves are considered sources of the
magnetoelastic field, and we examine how the strength of the magneto-rotational
coupling varies with the direction of the in-plane applied magnetic field. We
found that in the absence of magnetocrystalline anisotropy the
magneto-rotational contribution to the magnetoelastic field decreases with a
reduction in the thickness-to-width ratio of the strip for a Rayleigh wave,
whereas for a Love wave, it changes non-monotonically. These findings enhance
the understanding of magneto-rotational coupling in magnonic nanostructures.
| [
{
"version": "v1",
"created": "Wed, 5 Mar 2025 10:53:14 GMT"
},
{
"version": "v2",
"created": "Wed, 19 Mar 2025 20:37:01 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Centała",
"Grzegorz",
""
],
[
"Kłos",
"Jarosław W.",
""
]
] |
2503.03404 | Katja Fahrion | K. Fahrion, M. A. Beasley, A. Gvozdenko, S. Guerra Arencibia, T.
Jerabkova, J. Fensch, and E. Emsellem | Revisiting the globular clusters of NGC1052-DF2 | 13 pages, 10 figures, updated to version accepted for publication in
A&A | null | null | null | astro-ph.GA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The ultra-diffuse galaxy (UDG) NGC1052-DF2 has captured the interest of
astronomers ever since the low velocity dispersion measured from ten globular
clusters (GCs) suggested a low dark matter fraction. Also, its GC system was
found to be unusually bright, with a GC luminosity function peak at least one
magnitude brighter than expected for a galaxy at a distance of 20 Mpc. In this
work we present an updated view of the GC system of NGC1052-DF2. We analysed
archival MUSE data of NGC1052-DF2 to confirm the membership of four additional
GCs based on their radial velocities, thereby raising the number of
spectroscopically confirmed GCs to 16. We measured the ages and metallicities
of 11 individual GCs, finding them to be old ($> 9$ Gyr) and with a range of
metallicities from [M/H] = $-0.7$ to $-1.8$ dex. The majority of GCs are found
to be more metal-poor than the host galaxy, with some metal-rich GCs sharing
the metallicity of the host ([M/H] = $-$1.09$^{+0.09}_{-0.07}$ dex). The host
galaxy shows a flat age and metallicity gradient out to 1 $R_\text{e}$. Using a
distance measurement based on the internal GC velocity dispersions ($D = 16.2$
Mpc), we derived photometric GC masses and find that the peak of the GC mass
function compares well with that of the Milky Way. From updated GC velocities,
we estimated the GC system velocity dispersion of NGC1052-DF2 with a simple
kinematic model and find $\sigma_\text{GCS} = 14.86^{+3.89}_{-2.83}$ km
s$^{-1}$. However, this value is reduced to $\sigma_\text{GCS} =
8.63^{+2.88}_{-2.14}$ km s$^{-1}$ when the GC that has the highest relative
velocity based on a low S/N spectrum is considered an interloper. We discuss
the possible origin of NGC1502-DF2, taking the lower distance, spread in GC
metallicities, flat stellar population profiles, and dynamical mass estimate
into consideration.
| [
{
"version": "v1",
"created": "Wed, 5 Mar 2025 11:30:49 GMT"
},
{
"version": "v2",
"created": "Fri, 7 Mar 2025 07:55:20 GMT"
},
{
"version": "v3",
"created": "Thu, 20 Mar 2025 08:38:21 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Fahrion",
"K.",
""
],
[
"Beasley",
"M. A.",
""
],
[
"Gvozdenko",
"A.",
""
],
[
"Arencibia",
"S. Guerra",
""
],
[
"Jerabkova",
"T.",
""
],
[
"Fensch",
"J.",
""
],
[
"Emsellem",
"E.",
""
]
] |
2503.03409 | Abdollah Langari | S. Sadeghizade, R. Jafari and A. Langari | Anti Kibble-Zurek behavior in the quantum XY spin-1/2 chain driven by
correlated noisy magnetic field and anisotropy | 13 pages, 12 figures, References updated | Phys. Rev. B 111, 104310 (2025) | 10.1103/PhysRevB.111.104310 | null | cond-mat.str-el cond-mat.stat-mech quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In the non-adiabatic dynamics across a quantum phase transition, the
Kibble-Zurek paradigm describes that the average number of topological defects
is suppressed as a universal power law with the quench time scale. A
conflicting observation, which termed anti-Kibble-Zurek dynamics has been
reported in several studies, specifically in the driven systems with an
uncorrelated stochastic (white) noise. Here, we study the defect generation in
the driven transverse field/anisotropy quantum $XY$ model in the presence of a
correlated (colored) Gaussian noise. We propose a generic conjecture that
properly capture the noise-induced excitation features, which shows good
agreement with the numerical simulations. We show that, the dynamical features
of defect density are modified by varying the noise correlation time. Our
numerical simulations confirm that, for fast noises, the dynamics of the defect
density is the same as that of the uncorrelated (white) noise, as is expected.
However, the larger ratio of noise correlation time to the annealing time
results in larger defects density formation and reforms the universal dynamical
features. Our finding reveals that, the noise-induced defects scale linearly
with the annealing time for fast noises, while in the presence of the slow
noises, the noise-induced defects scale linearly with the square of the
annealing time. The numerical simulations confirm that, the optimal annealing
time, at which the defects density is minimum, scales linearly in logarithmic
scale with the total noise power having different exponents for the fast and
slow noises.
| [
{
"version": "v1",
"created": "Wed, 5 Mar 2025 11:35:49 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 17:56:41 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Sadeghizade",
"S.",
""
],
[
"Jafari",
"R.",
""
],
[
"Langari",
"A.",
""
]
] |
2503.03644 | Shuo Li | Xiaojun Bi, Shuo Li, Ziyue Wang, Fuwen Luo, Weizheng Qiao, Lu Han,
Ziwei Sun, Peng Li, Yang Liu | DongbaMIE: A Multimodal Information Extraction Dataset for Evaluating
Semantic Understanding of Dongba Pictograms | Our dataset can be obtained from:
https://github.com/thinklis/DongbaMIE | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Dongba pictographs are the only pictographs still in use in the world. They
have pictorial ideographic features, and their symbols carry rich cultural and
contextual information. Due to the lack of relevant datasets, existing research
has difficulty in advancing the study of semantic understanding of Dongba
pictographs. To this end, we propose \textbf{DongbaMIE}, the first multimodal
dataset for semantic understanding and extraction of Dongba pictographs,
consisting of Dongba pictograph images and corresponding Chinese semantic
annotations. DongbaMIE contains 23,530 sentence-level and 2,539 paragraph-level
images, covering four semantic dimensions: objects, actions, relations, and
attributes. We systematically evaluate multimodal large language models
(MLLMs), such as GPT-4o, Gemini-2.0, and Qwen2-VL. Experimental results show
that best F1 scores of proprietary models, GPT-4o and Gemini, for object
extraction task are only 3.16 and 3.11 respectively. For the open-source model
Qwen2-VL, it achieves only 11.49 after supervised fine-tuning. These suggest
that current MLLMs still face significant challenges in accurately recognizing
diverse semantic information in Dongba pictographs.
| [
{
"version": "v1",
"created": "Wed, 5 Mar 2025 16:20:53 GMT"
},
{
"version": "v2",
"created": "Thu, 6 Mar 2025 11:36:33 GMT"
},
{
"version": "v3",
"created": "Thu, 20 Mar 2025 12:16:23 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Bi",
"Xiaojun",
""
],
[
"Li",
"Shuo",
""
],
[
"Wang",
"Ziyue",
""
],
[
"Luo",
"Fuwen",
""
],
[
"Qiao",
"Weizheng",
""
],
[
"Han",
"Lu",
""
],
[
"Sun",
"Ziwei",
""
],
[
"Li",
"Peng",
""
],
[
"Liu",
"Yang",
""
]
] |
2503.04784 | Jiexiong Liu | Cheng Li, Jiexiong Liu, Yixuan Chen, Yanqin Jia, Zhepeng Li | KunlunBaize: LLM with Multi-Scale Convolution and Multi-Token Prediction
Under TransformerX Framework | 21 pages | null | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Large language models have demonstrated remarkable performance across various
tasks, yet they face challenges such as low computational efficiency, gradient
vanishing, and difficulties in capturing complex feature interactions. To
address these limitations, a novel framework has been proposed. This framework
incorporates a learnable dense residual skip connection mechanism, a
TransformerX module a transformer based component integrating multiscale
convolution and adaptive activation functions and a multitoken prediction
interaction module. The learnable dense residual connections enhance
information flow and feature capture across layers. Within the TransformerX
module, large convolutional kernels aggregate semantic information from
extensive text segments, while smaller convolutions focus on local word order
and syntactic structures. The adaptive activation function dynamically adjusts
its parameters based on the semantic features of the input text, improving the
model's ability to handle diverse semantic expressions and complex
relationships. The multitoken prediction module boosts data utilization and
accelerates inference by predicting multiple future tokens. These components
significantly enhance the performance and efficiency of large language models.
| [
{
"version": "v1",
"created": "Thu, 27 Feb 2025 01:56:09 GMT"
},
{
"version": "v2",
"created": "Tue, 11 Mar 2025 01:59:26 GMT"
},
{
"version": "v3",
"created": "Thu, 20 Mar 2025 03:04:01 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Li",
"Cheng",
""
],
[
"Liu",
"Jiexiong",
""
],
[
"Chen",
"Yixuan",
""
],
[
"Jia",
"Yanqin",
""
],
[
"Li",
"Zhepeng",
""
]
] |
2503.04997 | Paul Krassnig | Paul J. Krassnig and Dieter P. Gruber | ISP-AD: A Large-Scale Real-World Dataset for Advancing Industrial
Anomaly Detection with Synthetic and Real Defects | 26 pages, 6 figures, this preprint has been submitted to the Journal
of Intelligent Manufacturing, the dataset is available at
https://doi.org/10.5281/zenodo.14911043 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Automatic visual inspection using machine learning-based methods plays a key
role in achieving zero-defect policies in industry. Research on anomaly
detection approaches is constrained by the availability of datasets that
represent complex defect appearances and imperfect imaging conditions, which
are typical to industrial processes. Recent benchmarks indicate that most
publicly available datasets are biased towards optimal imaging conditions,
leading to an overestimation of the methods' applicability to real-world
industrial scenarios. To address this gap, we introduce the Industrial Screen
Printing Anomaly Detection dataset (ISP-AD). It presents challenging small and
weakly contrasted surface defects embedded within structured patterns
exhibiting high permitted design variability. To the best of our knowledge, it
is the largest publicly available industrial dataset to date, including both
synthetic and real defects collected directly from the factory floor. In
addition to the evaluation of defect detection performance of recent
unsupervised anomaly detection methods, experiments on a mixed supervised
training approach, incorporating both synthesized and real defects, were
conducted. Even small amounts of injected real defects prove beneficial for
model generalization. Furthermore, starting from training on purely synthetic
defects, emerging real defective samples can be efficiently integrated into
subsequent scalable training. Research findings indicate that supervision by
means of both synthetic and accumulated real defects can complement each other,
meeting demanded industrial inspection requirements such as low false positive
rates and high recall. The presented unsupervised and supervised dataset splits
are designed to emphasize research on unsupervised, self-supervised, and
supervised approaches, enhancing their applicability to industrial settings.
| [
{
"version": "v1",
"created": "Thu, 6 Mar 2025 21:56:31 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 08:40:35 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Krassnig",
"Paul J.",
""
],
[
"Gruber",
"Dieter P.",
""
]
] |
2503.05244 | Yuning Wu | Yuning Wu and Jiahao Mei and Ming Yan and Chenliang Li and Shaopeng
Lai and Yuran Ren and Zijia Wang and Ji Zhang and Mengyue Wu and Qin Jin and
Fei Huang | WritingBench: A Comprehensive Benchmark for Generative Writing | null | null | null | null | cs.AI cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent advancements in large language models (LLMs) have significantly
enhanced text generation capabilities, yet evaluating their performance in
generative writing remains a challenge. Existing benchmarks primarily focus on
generic text generation or limited in writing tasks, failing to capture the
diverse requirements of high-quality written contents across various domains.
To bridge this gap, we present WritingBench, a comprehensive benchmark designed
to evaluate LLMs across 6 core writing domains and 100 subdomains, encompassing
creative, persuasive, informative, and technical writing. We further propose a
query-dependent evaluation framework that empowers LLMs to dynamically generate
instance-specific assessment criteria. This framework is complemented by a
fine-tuned critic model for criteria-aware scoring, enabling evaluations in
style, format and length. The framework's validity is further demonstrated by
its data curation capability, which enables 7B-parameter models to approach
state-of-the-art (SOTA) performance. We open-source the benchmark, along with
evaluation tools and modular framework components, to advance the development
of LLMs in writing.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 08:56:20 GMT"
},
{
"version": "v2",
"created": "Tue, 11 Mar 2025 12:11:00 GMT"
},
{
"version": "v3",
"created": "Thu, 20 Mar 2025 05:13:53 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Wu",
"Yuning",
""
],
[
"Mei",
"Jiahao",
""
],
[
"Yan",
"Ming",
""
],
[
"Li",
"Chenliang",
""
],
[
"Lai",
"Shaopeng",
""
],
[
"Ren",
"Yuran",
""
],
[
"Wang",
"Zijia",
""
],
[
"Zhang",
"Ji",
""
],
[
"Wu",
"Mengyue",
""
],
[
"Jin",
"Qin",
""
],
[
"Huang",
"Fei",
""
]
] |
2503.05406 | Kaede Shintani | Kaede Shintani, Hamada Rizk and Hirozumi Yamaguchi | Energy-Free Sensing and Context Recognition Using Photovoltaic Cells | null | null | 10.1109/JSEN.2025.3549473 | null | eess.SP | http://creativecommons.org/licenses/by/4.0/ | The field of energy-free sensing and context recognition has recently gained
significant attention as it allows operating systems without external power
sources. Photovoltaic cells can convert light energy into electrical energy to
power sensing devices, but their power may not be sufficient to ensure
energy-free sensing due to the varying power needs of sensors and high
computational demands. In this paper, we propose the use of photovoltaic cells
as a standalone sensor for the recognition of different contexts, including
user identification, step counting, and location tracking. The system utilizes
the photocurrent readings generated by the photovoltaic cells to capture the
unique mobility patterns of different users. By analyzing these patterns, the
system can accurately identify the user, count the number of steps taken, and
track the user's location. We propose a computationally efficient DTW to match
the variable length sequences of photocurrent readings to a database of known
patterns and identify the closest subject and location matches. The system was
rigorously evaluated in a realistic environment, and the results indicate that
it can accurately estimate step count, identify subjects, and localize them
with an accuracy of 88%, 90%, and 43cm, respectively. This is achieved while
the proposed system is non-intrusive and can operate without external power
sources, making it a promising technology for energy-free sensing and context
recognition.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 13:28:54 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Shintani",
"Kaede",
""
],
[
"Rizk",
"Hamada",
""
],
[
"Yamaguchi",
"Hirozumi",
""
]
] |
2503.05931 | Piotr \.Zelasko | Piotr \.Zelasko, Kunal Dhawan, Daniel Galvez, Krishna C. Puvvada,
Ankita Pasad, Nithin Rao Koluguri, Ke Hu, Vitaly Lavrukhin, Jagadeesh Balam,
Boris Ginsburg | Training and Inference Efficiency of Encoder-Decoder Speech Models | null | null | null | null | cs.CL eess.AS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Attention encoder-decoder model architecture is the backbone of several
recent top performing foundation speech models: Whisper, Seamless, OWSM, and
Canary-1B. However, the reported data and compute requirements for their
training are prohibitive for many in the research community. In this work, we
focus on the efficiency angle and ask the questions of whether we are training
these speech models efficiently, and what can we do to improve? We argue that a
major, if not the most severe, detrimental factor for training efficiency is
related to the sampling strategy of sequential data. We show that negligence in
mini-batch sampling leads to more than 50% computation being spent on padding.
To that end, we study, profile, and optimize Canary-1B training to show gradual
improvement in GPU utilization leading up to 5x increase in average batch sizes
versus its original training settings. This in turn allows us to train an
equivalent model using 4x less GPUs in the same wall time, or leverage the
original resources and train it in 2x shorter wall time. Finally, we observe
that the major inference bottleneck lies in the autoregressive decoder steps.
We find that adjusting the model architecture to transfer model parameters from
the decoder to the encoder results in a 3x inference speedup as measured by
inverse real-time factor (RTFx) while preserving the accuracy and compute
requirements for convergence. The training code and models will be available as
open-source.
| [
{
"version": "v1",
"created": "Fri, 7 Mar 2025 20:57:43 GMT"
},
{
"version": "v2",
"created": "Wed, 19 Mar 2025 18:08:48 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Żelasko",
"Piotr",
""
],
[
"Dhawan",
"Kunal",
""
],
[
"Galvez",
"Daniel",
""
],
[
"Puvvada",
"Krishna C.",
""
],
[
"Pasad",
"Ankita",
""
],
[
"Koluguri",
"Nithin Rao",
""
],
[
"Hu",
"Ke",
""
],
[
"Lavrukhin",
"Vitaly",
""
],
[
"Balam",
"Jagadeesh",
""
],
[
"Ginsburg",
"Boris",
""
]
] |
2503.06764 | Zisheng Chen | Zisheng Chen, Chunwei Wang, Xiuwei Chen, Hang Xu, Jianhua Han, Xiaodan
Liang | SemHiTok: A Unified Image Tokenizer via Semantic-Guided Hierarchical
Codebook for Multimodal Understanding and Generation | Under Review, Refer to the latest version | null | null | null | cs.CV cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present SemHiTok, a unified image Tokenizer via Semantic-Guided
Hierarchical codebook that provides consistent discrete feature representations
for multimodal understanding and generation tasks. Recently, unified multimodal
large models (MLLMs) for understanding and generation have sparked exploration
within research community. Previous works attempt to train a unified image
tokenizer by combining loss functions for semantic feature reconstruction and
pixel reconstruction. However, due to the differing levels of features
prioritized by multimodal understanding and generation tasks, joint training
methods face significant challenges in achieving a good trade-off. SemHiTok
addresses this challenge through Semantic-Guided Hierarchical codebook which
builds texture sub-codebooks on pre-trained semantic codebook. This design
decouples the training of semantic reconstruction and pixel reconstruction and
equips the tokenizer with low-level texture feature extraction capability
without degradation of high-level semantic feature extraction ability. Our
experiments demonstrate that SemHiTok achieves excellent rFID score at
256X256resolution compared to other unified tokenizers, and exhibits
competitive performance on multimodal understanding and generation tasks.
| [
{
"version": "v1",
"created": "Sun, 9 Mar 2025 20:42:34 GMT"
},
{
"version": "v2",
"created": "Wed, 19 Mar 2025 05:24:47 GMT"
},
{
"version": "v3",
"created": "Thu, 20 Mar 2025 06:13:32 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Chen",
"Zisheng",
""
],
[
"Wang",
"Chunwei",
""
],
[
"Chen",
"Xiuwei",
""
],
[
"Xu",
"Hang",
""
],
[
"Han",
"Jianhua",
""
],
[
"Liang",
"Xiaodan",
""
]
] |
2503.07013 | Lei Zhang | Lei Zhang, Siddharth Das, Tanner Merry, Wenlong Zhang, Yi Ren | Learning Nash Equilibrial Hamiltonian for Two-Player Collision-Avoiding
Interactions | Accepted by 2025 ACC | null | null | null | cs.RO cs.GT cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We consider the problem of learning Nash equilibrial policies for two-player
risk-sensitive collision-avoiding interactions. Solving the
Hamilton-Jacobi-Isaacs equations of such general-sum differential games in real
time is an open challenge due to the discontinuity of equilibrium values on the
state space. A common solution is to learn a neural network that approximates
the equilibrium Hamiltonian for given system states and actions. The learning,
however, is usually supervised and requires a large amount of sample
equilibrium policies from different initial states in order to mitigate the
risks of collisions. This paper claims two contributions towards more
data-efficient learning of equilibrium policies: First, instead of computing
Hamiltonian through a value network, we show that the equilibrium co-states
have simple structures when collision avoidance dominates the agents' loss
functions and system dynamics is linear, and therefore are more data-efficient
to learn. Second, we introduce theory-driven active learning to guide data
sampling, where the acquisition function measures the compliance of the
predicted co-states to Pontryagin's Maximum Principle. On an uncontrolled
intersection case, the proposed method leads to more generalizable
approximation of the equilibrium policies, and in turn, lower collision
probabilities, than the state-of-the-art under the same data acquisition
budget.
| [
{
"version": "v1",
"created": "Mon, 10 Mar 2025 07:52:30 GMT"
},
{
"version": "v2",
"created": "Wed, 19 Mar 2025 21:47:07 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Zhang",
"Lei",
""
],
[
"Das",
"Siddharth",
""
],
[
"Merry",
"Tanner",
""
],
[
"Zhang",
"Wenlong",
""
],
[
"Ren",
"Yi",
""
]
] |
2503.07297 | Runxi Wang | Runxi Wang, Ziheng Wang, Ting Lin, Jacob M. Raby, Mircea R. Stan,
Xinfei Guo | Cool-3D: An End-to-End Thermal-Aware Framework for Early-Phase Design
Space Exploration of Microfluidic-Cooled 3DICs | null | null | null | null | cs.AR | http://creativecommons.org/licenses/by-nc-sa/4.0/ | The rapid advancement of three-dimensional integrated circuits (3DICs) has
heightened the need for early-phase design space exploration (DSE) to minimize
design iterations and unexpected challenges. Emphasizing the
pre-register-transfer level (Pre-RTL) design phase is crucial for reducing
trial-and-error costs. However, 3DIC design introduces additional complexities
due to thermal constraints and an expanded design space resulting from vertical
stacking and various cooling strategies. Despite this need, existing Pre-RTL
DSE tools for 3DICs remain scarce, with available solutions often lacking
comprehensive design options and full customization support. To bridge this
gap, we present Cool-3D, an end-to-end, thermal-aware framework for 3DIC design
that integrates mainstream architectural-level simulators, including gem5,
McPAT, and HotSpot 7.0, with advanced cooling models. Cool-3D enables broad and
fine-grained design space exploration, built-in microfluidic cooling support
for thermal analysis, and an extension interface for non-parameterizable
customization, allowing designers to model and optimize 3DIC architectures with
greater flexibility and accuracy. To validate the Cool-3D framework, we conduct
three case studies demonstrating its ability to model various hardware design
options and accurately capture thermal behaviors. Cool-3D serves as a
foundational framework that not only facilitates comprehensive 3DIC design
space exploration but also enables future innovations in 3DIC architecture,
cooling strategies, and optimization techniques. The entire framework, along
with the experimental data, is in the process of being released on GitHub.
| [
{
"version": "v1",
"created": "Mon, 10 Mar 2025 13:18:03 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 07:49:25 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Wang",
"Runxi",
""
],
[
"Wang",
"Ziheng",
""
],
[
"Lin",
"Ting",
""
],
[
"Raby",
"Jacob M.",
""
],
[
"Stan",
"Mircea R.",
""
],
[
"Guo",
"Xinfei",
""
]
] |
2503.07459 | Xiangru Tang | Xiangru Tang, Daniel Shao, Jiwoong Sohn, Jiapeng Chen, Jiayi Zhang,
Jinyu Xiang, Fang Wu, Yilun Zhao, Chenglin Wu, Wenqi Shi, Arman Cohan, Mark
Gerstein | MedAgentsBench: Benchmarking Thinking Models and Agent Frameworks for
Complex Medical Reasoning | null | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | Large Language Models (LLMs) have shown impressive performance on existing
medical question-answering benchmarks. This high performance makes it
increasingly difficult to meaningfully evaluate and differentiate advanced
methods. We present MedAgentsBench, a benchmark that focuses on challenging
medical questions requiring multi-step clinical reasoning, diagnosis
formulation, and treatment planning-scenarios where current models still
struggle despite their strong performance on standard tests. Drawing from seven
established medical datasets, our benchmark addresses three key limitations in
existing evaluations: (1) the prevalence of straightforward questions where
even base models achieve high performance, (2) inconsistent sampling and
evaluation protocols across studies, and (3) lack of systematic analysis of the
interplay between performance, cost, and inference time. Through experiments
with various base models and reasoning methods, we demonstrate that the latest
thinking models, DeepSeek R1 and OpenAI o3, exhibit exceptional performance in
complex medical reasoning tasks. Additionally, advanced search-based agent
methods offer promising performance-to-cost ratios compared to traditional
approaches. Our analysis reveals substantial performance gaps between model
families on complex questions and identifies optimal model selections for
different computational constraints. Our benchmark and evaluation framework are
publicly available at https://github.com/gersteinlab/medagents-benchmark.
| [
{
"version": "v1",
"created": "Mon, 10 Mar 2025 15:38:44 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 01:30:56 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Tang",
"Xiangru",
""
],
[
"Shao",
"Daniel",
""
],
[
"Sohn",
"Jiwoong",
""
],
[
"Chen",
"Jiapeng",
""
],
[
"Zhang",
"Jiayi",
""
],
[
"Xiang",
"Jinyu",
""
],
[
"Wu",
"Fang",
""
],
[
"Zhao",
"Yilun",
""
],
[
"Wu",
"Chenglin",
""
],
[
"Shi",
"Wenqi",
""
],
[
"Cohan",
"Arman",
""
],
[
"Gerstein",
"Mark",
""
]
] |
2503.07645 | Hongyuan Yang | Hongyuan Yang, Siqi Peng, Akihiro Yamamoto | BicliqueEncoder: An Efficient Method for Link Prediction in Bipartite
Networks using Formal Concept Analysis and Transformer Encoder | 33 pages, 8 figures | null | null | null | cs.LG cs.SI | http://creativecommons.org/licenses/by-nc-sa/4.0/ | We propose a novel and efficient method for link prediction in bipartite
networks, using \textit{formal concept analysis} (FCA) and the Transformer
encoder. Link prediction in bipartite networks finds practical applications in
various domains such as product recommendation in online sales, and prediction
of chemical-disease interaction in medical science. Since for link prediction,
the topological structure of a network contains valuable information, many
approaches focus on extracting structural features and then utilizing them for
link prediction. Bi-cliques, as a type of structural feature of bipartite
graphs, can be utilized for link prediction. Although several link prediction
methods utilizing bi-cliques have been proposed and perform well in rather
small datasets, all of them face challenges with scalability when dealing with
large datasets since they demand substantial computational resources. This
limits the practical utility of these approaches in real-world applications. To
overcome the limitation, we introduce a novel approach employing iceberg
concept lattices and the Transformer encoder. Our method requires fewer
computational resources, making it suitable for large-scale datasets while
maintaining high prediction performance. We conduct experiments on five large
real-world datasets that exceed the capacity of previous bi-clique-based
approaches to demonstrate the efficacy of our method. Additionally, we perform
supplementary experiments on five small datasets to compare with the previous
bi-clique-based methods for bipartite link prediction and demonstrate that our
method is more efficient than the previous ones.
| [
{
"version": "v1",
"created": "Thu, 6 Mar 2025 04:47:37 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 15:31:27 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Yang",
"Hongyuan",
""
],
[
"Peng",
"Siqi",
""
],
[
"Yamamoto",
"Akihiro",
""
]
] |
2503.07667 | Wei Dai | Wei Dai, Peilin Chen, Malinda Lu, Daniel Li, Haowen Wei, Hejie Cui,
Paul Pu Liang | CLIMB: Data Foundations for Large Scale Multimodal Clinical Foundation
Models | null | null | null | null | cs.LG cs.AI cs.CV eess.SP | http://creativecommons.org/licenses/by-sa/4.0/ | Recent advances in clinical AI have enabled remarkable progress across many
clinical domains. However, existing benchmarks and models are primarily limited
to a small set of modalities and tasks, which hinders the development of
large-scale multimodal methods that can make holistic assessments of patient
health and well-being. To bridge this gap, we introduce Clinical Large-Scale
Integrative Multimodal Benchmark (CLIMB), a comprehensive clinical benchmark
unifying diverse clinical data across imaging, language, temporal, and graph
modalities. CLIMB comprises 4.51 million patient samples totaling 19.01
terabytes distributed across 2D imaging, 3D video, time series, graphs, and
multimodal data. Through extensive empirical evaluation, we demonstrate that
multitask pretraining significantly improves performance on understudied
domains, achieving up to 29% improvement in ultrasound and 23% in ECG analysis
over single-task learning. Pretraining on CLIMB also effectively improves
models' generalization capability to new tasks, and strong unimodal encoder
performance translates well to multimodal performance when paired with
task-appropriate fusion strategies. Our findings provide a foundation for new
architecture designs and pretraining strategies to advance clinical AI
research. Code is released at https://github.com/DDVD233/climb.
| [
{
"version": "v1",
"created": "Sun, 9 Mar 2025 01:45:05 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 05:05:56 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Dai",
"Wei",
""
],
[
"Chen",
"Peilin",
""
],
[
"Lu",
"Malinda",
""
],
[
"Li",
"Daniel",
""
],
[
"Wei",
"Haowen",
""
],
[
"Cui",
"Hejie",
""
],
[
"Liang",
"Paul Pu",
""
]
] |
2503.08049 | Nadarasar Bahavan | Nadarasar Bahavan, Sachith Seneviratne, Saman Halgamuge | SphOR: A Representation Learning Perspective on Open-set Recognition for
Identifying Unknown Classes in Deep Learning Models | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | The widespread use of deep learning classifiers necessitates Open-set
recognition (OSR), which enables the identification of input data not only from
classes known during training but also from unknown classes that might be
present in test data. Many existing OSR methods are computationally expensive
due to the reliance on complex generative models or suffer from high training
costs. We investigate OSR from a representation-learning perspective,
specifically through spherical embeddings. We introduce SphOR, a
computationally efficient representation learning method that models the
feature space as a mixture of von Mises-Fisher distributions. This approach
enables the use of semantically ambiguous samples during training, to improve
the detection of samples from unknown classes. We further explore the
relationship between OSR performance and key representation learning properties
which influence how well features are structured in high-dimensional space.
Extensive experiments on multiple OSR benchmarks demonstrate the effectiveness
of our method, producing state-of-the-art results, with improvements up-to 6%
that validate its performance. Code at
https://github.com/nadarasarbahavan/SpHOR
| [
{
"version": "v1",
"created": "Tue, 11 Mar 2025 05:06:11 GMT"
},
{
"version": "v2",
"created": "Wed, 19 Mar 2025 22:03:31 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Bahavan",
"Nadarasar",
""
],
[
"Seneviratne",
"Sachith",
""
],
[
"Halgamuge",
"Saman",
""
]
] |
2503.08144 | Fei Wang | Fei Wang, Chengcheng Chen, Hongyu Chen, Yugang Chang, Weiming Zeng | Bring Remote Sensing Object Detect Into Nature Language Model: Using SFT
Method | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recently, large language models (LLMs) and vision-language models (VLMs) have
achieved significant success, demonstrating remarkable capabilities in
understanding various images and videos, particularly in classification and
detection tasks. However, due to the substantial differences between remote
sensing images and conventional optical images, these models face considerable
challenges in comprehension, especially in detection tasks. Directly prompting
VLMs with detection instructions often leads to unsatisfactory results. To
address this issue, this letter explores the application of VLMs for object
detection in remote sensing images. Specifically, we constructed supervised
fine-tuning (SFT) datasets using publicly available remote sensing object
detection datasets, including SSDD, HRSID, and NWPU-VHR-10. In these new
datasets, we converted annotation information into JSON-compliant natural
language descriptions, facilitating more effective understanding and training
for the VLM. We then evaluate the detection performance of various fine-tuning
strategies for VLMs and derive optimized model weights for object detection in
remote sensing images. Finally, we evaluate the model's prior knowledge
capabilities using natural language queries. Experimental results demonstrate
that, without modifying the model architecture, remote sensing object detection
can be effectively achieved using natural language alone. Additionally, the
model exhibits the ability to perform certain vision question answering (VQA)
tasks. Our datasets and related code will be released soon.
| [
{
"version": "v1",
"created": "Tue, 11 Mar 2025 08:02:54 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 13:21:00 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Wang",
"Fei",
""
],
[
"Chen",
"Chengcheng",
""
],
[
"Chen",
"Hongyu",
""
],
[
"Chang",
"Yugang",
""
],
[
"Zeng",
"Weiming",
""
]
] |
2503.08249 | Bo Zhao | Bo Zhao and Hongbin Zhang | High-throughput computation and machine learning modeling of magnetic
moments and M\"ossbauer parameters for Fe-based intermetallics | null | null | null | null | cond-mat.mtrl-sci | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Based on high-throughput density functional theory calculations, we evaluate
the local magnetic moments and M\"ossbauer properties for Fe-based
intermetallic compounds and employ machine learning to map the local
crystalline environments to such properties. It is observed that magnetic
moments and M\"ossbauer parameters provide complementary insights into the
local crystalline environment, where the statistical features cannot be
captured using phenomenological models. Furthermore, we find that the scarcity
of existing data in Materials Project (MP) poses a significant challenge in
developing predictive machine learning models, whereas SOAP-based descriptors
can be applied for reliable modeling of the enriched datasets with extra
structural prototypes. This work advances the mapping of local crystalline
structures to magnetic and spectroscopic properties, bridging the gap between
empirical observations and theoretical models.
| [
{
"version": "v1",
"created": "Tue, 11 Mar 2025 10:12:09 GMT"
},
{
"version": "v2",
"created": "Wed, 12 Mar 2025 08:01:32 GMT"
},
{
"version": "v3",
"created": "Thu, 20 Mar 2025 07:10:08 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Zhao",
"Bo",
""
],
[
"Zhang",
"Hongbin",
""
]
] |
2503.08250 | Jong Chul Ye | Jaa-Yeon Lee, Byunghee Cha, Jeongsol Kim, Jong Chul Ye | Aligning Text to Image in Diffusion Models is Easier Than You Think | null | null | null | null | cs.CV cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | While recent advancements in generative modeling have significantly improved
text-image alignment, some residual misalignment between text and image
representations still remains. Although many approaches have attempted to
address this issue by fine-tuning models using various reward models, etc., we
revisit the challenge from the perspective of representation alignment-an
approach that has gained popularity with the success of REPresentation
Alignment (REPA). We first argue that conventional text-to-image (T2I)
diffusion models, typically trained on paired image and text data (i.e.,
positive pairs) by minimizing score matching or flow matching losses, is
suboptimal from the standpoint of representation alignment. Instead, a better
alignment can be achieved through contrastive learning that leverages both
positive and negative pairs. To achieve this efficiently even with pretrained
models, we introduce a lightweight contrastive fine tuning strategy called
SoftREPA that uses soft text tokens. This approach improves alignment with
minimal computational overhead by adding fewer than 1M trainable parameters to
the pretrained model. Our theoretical analysis demonstrates that our method
explicitly increases the mutual information between text and image
representations, leading to enhanced semantic consistency. Experimental results
across text-to-image generation and text-guided image editing tasks validate
the effectiveness of our approach in improving the semantic consistency of T2I
generative models.
| [
{
"version": "v1",
"created": "Tue, 11 Mar 2025 10:14:22 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 04:35:44 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Lee",
"Jaa-Yeon",
""
],
[
"Cha",
"Byunghee",
""
],
[
"Kim",
"Jeongsol",
""
],
[
"Ye",
"Jong Chul",
""
]
] |
2503.08355 | Hugo Henneuse | Hugo Henneuse | Pointwise Minimax Vector Field Reconstruction from Noisy ODE | 17 pages | null | null | null | math.ST stat.TH | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This work addresses the problem of estimating a vector field from a noisy
Ordinary Differential Equation (ODE) in a non-parametric regression setting
with a random design for initial values. More specifically, given a vector
field $ f:\mathbb{R}^{D}\rightarrow \mathbb{R}^{D}$ governing a dynamical
system defined by the autonomous ODE: $y' = f(y)$, we assume that the
observations are $\tilde{y}_{X_{i}}(t_{j}) = y_{X_{i}}(t_{j}) +
\varepsilon_{i,j}$ where $y_{X_{i}}(t_{j})$ is the solution of the ODE at time
$t_{j}$ with initial condition $y(0) = X_{i}$, $X_{i}$ is sampled from a
probability distribution $\mu$, and $\varepsilon_{i,j}$ some noise. In this
context, we investigate, from a minimax perspective, the pointwise
reconstruction of $f$ within the envelope of trajectories originating from the
support of $\mu$. We propose an estimation strategy based on preliminary flow
reconstruction and techniques from derivative estimation in non-parametric
regression. Under mild assumptions on $f$, we establish convergence rates that
depend on the temporal resolution, the number of sampled initial values and the
mass concentration of $\mu$. Importantly, we show that these rates are minimax
optimal. Furthermore, we discuss the implications of our results in a manifold
learning setting, providing insights into how our approach can mitigate the
curse of dimensionality.
| [
{
"version": "v1",
"created": "Tue, 11 Mar 2025 12:13:02 GMT"
},
{
"version": "v2",
"created": "Mon, 17 Mar 2025 12:52:18 GMT"
},
{
"version": "v3",
"created": "Thu, 20 Mar 2025 10:25:59 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Henneuse",
"Hugo",
""
]
] |
2503.08380 | Kento Fujita | Kento Fujita | $t$-adic symmetric multiple zeta values for indices with alternating $1$
and $3$, starting with $1$ and ending with $3$ | 13 pages | null | null | null | math.NT | http://creativecommons.org/licenses/by/4.0/ | Hirose, Murahara, and Saito proved that some $t$-adic symmetric multiple zeta
values, for indices in which $1$ and $3$ appear alternately in succession, can
be expressed as polynomials in Riemann zeta values, and conjectured similar
formulas. In this paper, we prove the conjectured formula for indices that
start with $1$ and end with $3$, showing that they also can be expressed as
polynomials in Riemann zeta values.
| [
{
"version": "v1",
"created": "Tue, 11 Mar 2025 13:29:07 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 13:24:31 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Fujita",
"Kento",
""
]
] |
2503.08424 | Ziqiang Cai | Ziqiang Cai, Zhenglong Ban, Haiyuan Feng, Jinsong Yang | Thin accretion disk around Schwarzschild-like black hole in bumblebee
gravity | 15 pages, 13 figures | null | null | null | gr-qc | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate the physical properties and the optical appearance of a thin
accretion disk surrounding a Schwarzschild-like black hole (BH) within the
framework of bumblebee gravity. We study the impact of the Lorentz symmetry
breaking (LSB) parameter on various aspects of the disk, including its energy
flux, temperature distribution, and emission spectrum. By deriving and
analyzing both direct and secondary images of the accretion disk, we examine
how the observational inclination angle and the LSB parameter affect image
profile. We calculate and plot the distributions of both redshift and observed
flux for the disk as seen by distant observers at different inclination angles.
Our findings indicate that those distributions are influenced not only by the
LSB parameter but also the observer's inclination angle.
| [
{
"version": "v1",
"created": "Tue, 11 Mar 2025 13:38:23 GMT"
},
{
"version": "v2",
"created": "Wed, 19 Mar 2025 12:57:37 GMT"
},
{
"version": "v3",
"created": "Thu, 20 Mar 2025 14:25:19 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Cai",
"Ziqiang",
""
],
[
"Ban",
"Zhenglong",
""
],
[
"Feng",
"Haiyuan",
""
],
[
"Yang",
"Jinsong",
""
]
] |
2503.08522 | Ion Necoara | Lahcen El Bourkhissi and Ion Necoara | Convergence analysis of linearized $\ell_q$ penalty methods for
nonconvex optimization with nonlinear equality constraints | 28 pages. arXiv admin note: text overlap with arXiv:2402.15639 | null | null | null | math.OC | http://creativecommons.org/licenses/by/4.0/ | In this paper, we consider nonconvex optimization problems with nonlinear
equality constraints. We assume that the objective function and the functional
constraints are locally smooth. To solve this problem, we introduce a
linearized $\ell_q$ penalty based method, where $q \in (1,2]$ is the parameter
defining the norm used in the construction of the penalty function. Our method
involves linearizing the objective function and functional constraints in a
Gauss-Newton fashion at the current iteration in the penalty formulation and
introduces a quadratic regularization. This approach yields an easily solvable
subproblem, whose solution becomes the next iterate. By using a novel dynamic
rule for the choice of the regularization parameter, we establish that the
iterates of our method converge to an $\epsilon$-first-order solution in
$\mathcal{O}(1/{\epsilon^{2+ (q-1)/q}})$ outer iterations. Finally, we put
theory into practice and evaluate the performance of the proposed algorithm by
making numerical comparisons with existing methods from literature.
| [
{
"version": "v1",
"created": "Tue, 11 Mar 2025 15:15:28 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 17:40:50 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Bourkhissi",
"Lahcen El",
""
],
[
"Necoara",
"Ion",
""
]
] |
2503.08547 | Alexander Zubarev | Alexander P. Zubarev | On the analog of the Kolmogorov-Arnold superposition representation for
continuous functions of several $p$-adic variables | 8 pages | null | null | null | math-ph math.MP math.NT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | It is shown that any continuous function depending on several $p$-adic
variables, each of which is defined on $\mathbb{Z}_{p}$, can be represented as
a superposition of continuous functions of one $p$-adic variable. This
statement is true for both functions with values in $\mathbb{R}$ and functions
with values in $\mathbb{Q}_{p}$.
| [
{
"version": "v1",
"created": "Tue, 11 Mar 2025 15:35:23 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 06:16:09 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Zubarev",
"Alexander P.",
""
]
] |
2503.08679 | Iv\'an Arcuschin | Iv\'an Arcuschin, Jett Janiak, Robert Krzyzanowski, Senthooran
Rajamanoharan, Neel Nanda, Arthur Conmy | Chain-of-Thought Reasoning In The Wild Is Not Always Faithful | Accepted to the Reasoning and Planning for LLMs Workshop (ICLR 25),
10 main paper pages, 39 appendix pages | null | null | null | cs.AI cs.CL cs.LG | http://creativecommons.org/licenses/by/4.0/ | Chain-of-Thought (CoT) reasoning has significantly advanced state-of-the-art
AI capabilities. However, recent studies have shown that CoT reasoning is not
always faithful, i.e. CoT reasoning does not always reflect how models arrive
at conclusions. So far, most of these studies have focused on unfaithfulness in
unnatural contexts where an explicit bias has been introduced. In contrast, we
show that unfaithful CoT can occur on realistic prompts with no artificial
bias. Our results reveal non-negligible rates of several forms of unfaithful
reasoning in frontier models: Sonnet 3.7 (16.3%), DeepSeek R1 (5.3%) and
ChatGPT-4o (7.0%) all answer a notable proportion of question pairs
unfaithfully. Specifically, we find that models rationalize their implicit
biases in answers to binary questions ("implicit post-hoc rationalization").
For example, when separately presented with the questions "Is X bigger than Y?"
and "Is Y bigger than X?", models sometimes produce superficially coherent
arguments to justify answering Yes to both questions or No to both questions,
despite such responses being logically contradictory. We also investigate
restoration errors (Dziri et al., 2023), where models make and then silently
correct errors in their reasoning, and unfaithful shortcuts, where models use
clearly illogical reasoning to simplify solving problems in Putnam questions (a
hard benchmark). Our findings raise challenges for AI safety work that relies
on monitoring CoT to detect undesired behavior.
| [
{
"version": "v1",
"created": "Tue, 11 Mar 2025 17:56:30 GMT"
},
{
"version": "v2",
"created": "Thu, 13 Mar 2025 17:49:58 GMT"
},
{
"version": "v3",
"created": "Wed, 19 Mar 2025 19:20:42 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Arcuschin",
"Iván",
""
],
[
"Janiak",
"Jett",
""
],
[
"Krzyzanowski",
"Robert",
""
],
[
"Rajamanoharan",
"Senthooran",
""
],
[
"Nanda",
"Neel",
""
],
[
"Conmy",
"Arthur",
""
]
] |
2503.08738 | Janis Zenkner | Janis Zenkner, Tobias Sesterhenn, Christian Bartelt | Shedding Light in Task Decomposition in Program Synthesis: The Driving
Force of the Synthesizer Model | Accepted at ICLR 2025 Workshop Deep Learning for Code | null | null | null | cs.SE cs.LG cs.PL | http://creativecommons.org/licenses/by/4.0/ | Task decomposition is a fundamental mechanism in program synthesis, enabling
complex problems to be broken down into manageable subtasks. ExeDec, a
state-of-the-art program synthesis framework, employs this approach by
combining a Subgoal Model for decomposition and a Synthesizer Model for program
generation to facilitate compositional generalization. In this work, we develop
REGISM, an adaptation of ExeDec that removes decomposition guidance and relies
solely on iterative execution-driven synthesis. By comparing these two
exemplary approaches-ExeDec, which leverages task decomposition, and REGISM,
which does not-we investigate the interplay between task decomposition and
program generation. Our findings indicate that ExeDec exhibits significant
advantages in length generalization and concept composition tasks, likely due
to its explicit decomposition strategies. At the same time, REGISM frequently
matches or surpasses ExeDec's performance across various scenarios, with its
solutions often aligning more closely with ground truth decompositions. These
observations highlight the importance of repeated execution-guided synthesis in
driving task-solving performance, even within frameworks that incorporate
explicit decomposition strategies. Our analysis suggests that task
decomposition approaches like ExeDec hold significant potential for advancing
program synthesis, though further work is needed to clarify when and why these
strategies are most effective.
| [
{
"version": "v1",
"created": "Tue, 11 Mar 2025 06:30:49 GMT"
},
{
"version": "v2",
"created": "Fri, 14 Mar 2025 19:16:28 GMT"
},
{
"version": "v3",
"created": "Thu, 20 Mar 2025 08:23:40 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Zenkner",
"Janis",
""
],
[
"Sesterhenn",
"Tobias",
""
],
[
"Bartelt",
"Christian",
""
]
] |
2503.08748 | Andrzej Cichocki | Andrzej Cichocki, Toshihisa Tanaka and Sergio Cruces | Mirror Descent and Novel Exponentiated Gradient Algorithms Using
Trace-Form Entropies and Deformed Logarithms | 22 pages, 9 figures | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by-nc-nd/4.0/ | In this paper we propose and investigate a wide class of Mirror Descent
updates (MD) and associated novel Generalized Exponentiated Gradient (GEG)
algorithms by exploiting various trace-form entropies and associated deformed
logarithms and their inverses - deformed (generalized) exponential functions.
The proposed algorithms can be considered as extension of entropic MD and
generalization of multiplicative updates. In the literature, there exist
nowadays over fifty mathematically well defined generalized entropies, so
impossible to exploit all of them in one research paper. So we focus on a few
selected most popular entropies and associated logarithms like the Tsallis,
Kaniadakis and Sharma-Taneja-Mittal and some of their extension like Tempesta
or Kaniadakis-Scarfone entropies. The shape and properties of the deformed
logarithms and their inverses are tuned by one or more hyperparameters. By
learning these hyperparameters, we can adapt to distribution of training data,
which can be designed to the specific geometry of the optimization problem,
leading to potentially faster convergence and better performance. The using
generalized entropies and associated deformed logarithms in the Bregman
divergence, used as a regularization term, provides some new insight into
exponentiated gradient descent updates.
| [
{
"version": "v1",
"created": "Tue, 11 Mar 2025 10:50:07 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 06:40:17 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Cichocki",
"Andrzej",
""
],
[
"Tanaka",
"Toshihisa",
""
],
[
"Cruces",
"Sergio",
""
]
] |
2503.08827 | Donghee Lee | Donghee Lee, Hye-Sung Lee, Jaeok Yi | Synaptic Field Theory for Neural Networks | 6 pages, 2 figures. Major revisions including references | null | null | null | hep-th cond-mat.dis-nn hep-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Theoretical understanding of deep learning remains elusive despite its
empirical success. In this study, we propose a novel "synaptic field theory"
that describes the training dynamics of synaptic weights and biases in the
continuum limit. Unlike previous approaches, our framework treats synaptic
weights and biases as fields and interprets their indices as spatial
coordinates, with the training data acting as external sources. This
perspective offers new insights into the fundamental mechanisms of deep
learning and suggests a pathway for leveraging well-established field-theoretic
techniques to study neural network training.
| [
{
"version": "v1",
"created": "Tue, 11 Mar 2025 19:00:09 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 14:39:41 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Lee",
"Donghee",
""
],
[
"Lee",
"Hye-Sung",
""
],
[
"Yi",
"Jaeok",
""
]
] |
2503.08923 | Anand Menon | Anand Menon, Samit S Miftah, Shamik Kundu, Souvik Kundu, Amisha
Srivastava, Arnab Raha, Gabriel Theodor Sonnenschein, Suvadeep Banerjee,
Deepak Mathaikutty, Kanad Basu | Enhancing Large Language Models for Hardware Verification: A Novel
SystemVerilog Assertion Dataset | 29 Pages | null | null | null | cs.LG cs.CR cs.PL | http://creativecommons.org/licenses/by/4.0/ | Hardware verification is crucial in modern SoC design, consuming around 70%
of development time. SystemVerilog assertions ensure correct functionality.
However, existing industrial practices rely on manual efforts for assertion
generation, which becomes increasingly untenable as hardware systems become
complex. Recent research shows that Large Language Models (LLMs) can automate
this process. However, proprietary SOTA models like GPT-4o often generate
inaccurate assertions and require expensive licenses, while smaller open-source
LLMs need fine-tuning to manage HDL code complexities. To address these issues,
we introduce **VERT**, an open-source dataset designed to enhance SystemVerilog
assertion generation using LLMs. VERT enables researchers in academia and
industry to fine-tune open-source models, outperforming larger proprietary ones
in both accuracy and efficiency while ensuring data privacy through local
fine-tuning and eliminating costly licenses. The dataset is curated by
systematically augmenting variables from open-source HDL repositories to
generate synthetic code snippets paired with corresponding assertions.
Experimental results demonstrate that fine-tuned models like Deepseek Coder
6.7B and Llama 3.1 8B outperform GPT-4o, achieving up to 96.88% improvement
over base models and 24.14% over GPT-4o on platforms including OpenTitan, CVA6,
OpenPiton and Pulpissimo. VERT is available at
https://github.com/AnandMenon12/VERT.
| [
{
"version": "v1",
"created": "Tue, 11 Mar 2025 22:13:26 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Menon",
"Anand",
""
],
[
"Miftah",
"Samit S",
""
],
[
"Kundu",
"Shamik",
""
],
[
"Kundu",
"Souvik",
""
],
[
"Srivastava",
"Amisha",
""
],
[
"Raha",
"Arnab",
""
],
[
"Sonnenschein",
"Gabriel Theodor",
""
],
[
"Banerjee",
"Suvadeep",
""
],
[
"Mathaikutty",
"Deepak",
""
],
[
"Basu",
"Kanad",
""
]
] |
2503.09091 | Chen Zhao | Dong Li, Guihong Wan, Xintao Wu, Xinyu Wu, Xiaohui Chen, Yi He,
Christine G. Lian, Peter K. Sorger, Yevgeniy R. Semenov, Chen Zhao | Multi-Modal Foundation Models for Computational Pathology: A Survey | null | null | null | null | cs.CV cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Foundation models have emerged as a powerful paradigm in computational
pathology (CPath), enabling scalable and generalizable analysis of
histopathological images. While early developments centered on uni-modal models
trained solely on visual data, recent advances have highlighted the promise of
multi-modal foundation models that integrate heterogeneous data sources such as
textual reports, structured domain knowledge, and molecular profiles. In this
survey, we provide a comprehensive and up-to-date review of multi-modal
foundation models in CPath, with a particular focus on models built upon
hematoxylin and eosin (H&E) stained whole slide images (WSIs) and tile-level
representations. We categorize 32 state-of-the-art multi-modal foundation
models into three major paradigms: vision-language, vision-knowledge graph, and
vision-gene expression. We further divide vision-language models into
non-LLM-based and LLM-based approaches. Additionally, we analyze 28 available
multi-modal datasets tailored for pathology, grouped into image-text pairs,
instruction datasets, and image-other modality pairs. Our survey also presents
a taxonomy of downstream tasks, highlights training and evaluation strategies,
and identifies key challenges and future directions. We aim for this survey to
serve as a valuable resource for researchers and practitioners working at the
intersection of pathology and AI.
| [
{
"version": "v1",
"created": "Wed, 12 Mar 2025 06:03:33 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 16:43:54 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Li",
"Dong",
""
],
[
"Wan",
"Guihong",
""
],
[
"Wu",
"Xintao",
""
],
[
"Wu",
"Xinyu",
""
],
[
"Chen",
"Xiaohui",
""
],
[
"He",
"Yi",
""
],
[
"Lian",
"Christine G.",
""
],
[
"Sorger",
"Peter K.",
""
],
[
"Semenov",
"Yevgeniy R.",
""
],
[
"Zhao",
"Chen",
""
]
] |
2503.09267 | Enrique P\'erez-Montero Dr | E. P\'erez-Montero, J.A. Fern\'andez-Ontiveros, B. P\'erez-D\'iaz,
J.M. V\'ilchez and R. Amor\'in | Exploring the hardness of the ionizing radiation with the infrared
softness diagram. II. Bimodal distributions in both the ionizing continuum
slope and the excitation in active galactic nuclei | 12 pages, 6 figures. Accepted for publication by Astronomy &
Astrophysics. Version edited by A/A | null | null | null | astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | After exploring the infrared softness diagram to characterize the hardness of
the incident ionizing radiation in star-forming regions, we exploit the
availability of high-excitation lines in the same spectral regime to explore
its use for studying the narrow-line regions in AGN. We adapted the IR softness
diagram to consider very high-excitation lines, such as [Ne V] 14.3,24.3 $\mu$m
or [O IV] 25.9$\mu$m. The measured emission-line ratios were included as inputs
for the code HCm-Teff-IR, in order to provide a quantification for both the
ionization and the $\alpha_{\rm OX}$ parameters. The latter is sensitive to the
spectral shape of the incident continuum in AGN. We applied this to a large AGN
sample including different spectral types with available Spitzer/IRS,
Herschel/PACS and/or SOFIA/FIFI-LS mid-IR spectroscopic observations. The
combination of the ([Ne II] 12.8 $\mu$m+[Ne III] 15.6$\mu$m)/[Ne V] 14.3$\mu$m
and [O III] 52,88 $\mu$m/[O IV] 25.9$\mu$m emission line ratios is a robust
proxy for the shape of the ionizing continuum in AGN. An alternative based on
the [S III] 18.7 $\mu$m+[S IV] 10.5$\mu$m lines can be used instead. The
inclusion of very high-excitation lines in HCm-Teff-IR, to derive both
$\alpha_{\rm OX}$ and $U$ for the studied sample, points to a bimodal
distribution of galaxies. One of the peaks is characterized by relatively
harder values of $\alpha_{\rm OX}$ around -1.4 in combination with low values
for log $U$ around -2.4, while the other peak shows a softer $\alpha_{\rm OX}$
around -1.7 and high values of log $U$ around-1.5. This result is consistent
with the existence of two very marked AGN populations., one with a softer
ionizing continuum, possibly dominated by a radiatively efficient accretion
disk in bright Seyfert nuclei. In contrast, we observe a harder radiation field
in low-luminosity AGN, where the accretion disk is expected to recede.
| [
{
"version": "v1",
"created": "Wed, 12 Mar 2025 11:09:11 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 09:54:55 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Pérez-Montero",
"E.",
""
],
[
"Fernández-Ontiveros",
"J. A.",
""
],
[
"Pérez-Díaz",
"B.",
""
],
[
"Vílchez",
"J. M.",
""
],
[
"Amorín",
"R.",
""
]
] |
2503.09275 | Sergei Koniakhin | R. Ferrini and S.V. Koniakhin | Driven-dissipative turbulence in exciton-polariton quantum fluids | 12 pages, 10 figures, 1 table | null | null | null | cond-mat.mes-hall cond-mat.quant-gas | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The present paper is devoted to comprehensive theoretical studies of
exction-polariton quantum fluids specificities in the optics of their
utilization for quantum turbulence research. We show that a non-trivial
implementation of time-varying potential for excitation of quantum fluid
(injection of quantized vortices) via the stirring procedure can be efficiently
substituted with resonant excitation-based phase-imprinting techniques. The
most efficient phase pattern corresponds to imprinting of tiles with randomly
oriented plane waves in each. The resulting turbulent flows, spatial vortex
distributions, and clustering statistics resemble those for the case of a
conventional spoon-stirring scheme. We quantify the limitations on the lifetime
and density depletion for the development and sustainability of quantum
turbulence. The yield is the necessity to prevent the density depletion for
more than one order of magnitude. Finally, we demonstrate that turbulence is
robust with respect to alternating gain and loss at a certain range of
modulation parameters, which corresponds to laser operating above and below
condensation threshold.
| [
{
"version": "v1",
"created": "Wed, 12 Mar 2025 11:19:24 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 12:35:48 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Ferrini",
"R.",
""
],
[
"Koniakhin",
"S. V.",
""
]
] |
2503.09487 | Beier Zhu | Beier Zhu, Jiequan Cui, Hanwang Zhang, Chi Zhang | Project-Probe-Aggregate: Efficient Fine-Tuning for Group Robustness | Accepted by CVPR 2025 | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | While image-text foundation models have succeeded across diverse downstream
tasks, they still face challenges in the presence of spurious correlations
between the input and label. To address this issue, we propose a simple
three-step approach,Project-Probe-Aggregate (PPA), that enables
parameter-efficient fine-tuning for foundation models without relying on group
annotations. Building upon the failure-based debiasing scheme, our method, PPA,
improves its two key components: minority samples identification and the robust
training algorithm. Specifically, we first train biased classifiers by
projecting image features onto the nullspace of class proxies from text
encoders. Next, we infer group labels using the biased classifier and probe
group targets with prior correction. Finally, we aggregate group weights of
each class to produce the debiased classifier. Our theoretical analysis shows
that our PPA enhances minority group identification and is Bayes optimal for
minimizing the balanced group error, mitigating spurious correlations.
Extensive experimental results confirm the effectiveness of our PPA: it
outperforms the state-of-the-art by an average worst-group accuracy while
requiring less than 0.01% tunable parameters without training group labels.
| [
{
"version": "v1",
"created": "Wed, 12 Mar 2025 15:46:12 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 14:58:40 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Zhu",
"Beier",
""
],
[
"Cui",
"Jiequan",
""
],
[
"Zhang",
"Hanwang",
""
],
[
"Zhang",
"Chi",
""
]
] |
2503.09516 | Bowen Jin | Bowen Jin, Hansi Zeng, Zhenrui Yue, Dong Wang, Hamed Zamani, Jiawei
Han | Search-R1: Training LLMs to Reason and Leverage Search Engines with
Reinforcement Learning | 16 pages | null | null | null | cs.CL cs.AI cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Efficiently acquiring external knowledge and up-to-date information is
essential for effective reasoning and text generation in large language models
(LLMs). Prompting advanced LLMs with reasoning capabilities during inference to
use search engines is not optimal, since the LLM does not learn how to
optimally interact with the search engine. This paper introduces Search-R1, an
extension of the DeepSeek-R1 model where the LLM learns -- solely through
reinforcement learning (RL) -- to autonomously generate (multiple) search
queries during step-by-step reasoning with real-time retrieval. Search-R1
optimizes LLM rollouts with multi-turn search interactions, leveraging
retrieved token masking for stable RL training and a simple outcome-based
reward function. Experiments on seven question-answering datasets show that
Search-R1 improves performance by 26% (Qwen2.5-7B), 21% (Qwen2.5-3B), and 10%
(LLaMA3.2-3B) over strong baselines. This paper further provides empirical
insights into RL optimization methods, LLM choices, and response length
dynamics in retrieval-augmented reasoning. The code and model checkpoints are
available at https://github.com/PeterGriffinJin/Search-R1.
| [
{
"version": "v1",
"created": "Wed, 12 Mar 2025 16:26:39 GMT"
},
{
"version": "v2",
"created": "Wed, 19 Mar 2025 21:40:12 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Jin",
"Bowen",
""
],
[
"Zeng",
"Hansi",
""
],
[
"Yue",
"Zhenrui",
""
],
[
"Wang",
"Dong",
""
],
[
"Zamani",
"Hamed",
""
],
[
"Han",
"Jiawei",
""
]
] |
2503.09711 | Rebekah Rogers | Rebekah L. Rogers, John P. Wares, Jeffrey T. Garner | Genome evolution in an endangered freshwater mussel | 28 pages main text, 9 pages supplement, 6 main figures | null | null | null | q-bio.PE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Nearly neutral theory predicts that evolutionary processes will differ in
small populations compared to large populations, a key point of concern for
endangered species. The nearly-neutral threshold, the span of neutral
variation, and the adaptive potential from new mutations all differ depending
on N_e. To determine how genomes respond in small populations, we have created
a reference genome for a US federally endangered IUCN Red List freshwater
mussel, Elliptio spinosa, and compare it to genetic variation for a common and
successful relative, Elliptio crassidens. We find higher rates of background
duplication rates in E. spinosa consistent with proposed theories of duplicate
gene accumulation according to nearly-neutral processes. Along with these
changes we observe fewer cases of adaptive gene family amplification in this
endangered species. However, TE content is not consistent with nearly-neutral
theory. We observe substantially less recent TE proliferation in the endangered
species with over 500 Mb of newly copied TEs in Elliptio crassidens. These
results suggest a more complex interplay between TEs and duplicate genes than
previously proposed for small populations. They further suggest that TEs and
duplications require greater attention in surveys of genomic health for
endangered species.
| [
{
"version": "v1",
"created": "Wed, 12 Mar 2025 18:04:34 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 00:19:05 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Rogers",
"Rebekah L.",
""
],
[
"Wares",
"John P.",
""
],
[
"Garner",
"Jeffrey T.",
""
]
] |
2503.09965 | Min Ju | Min Ju, Xuhao Wu, Hong Shen | Effects of symmetry energy on the properties of hadron-quark mixed phase
in hybrid stars | null | null | null | null | nucl-th | http://creativecommons.org/licenses/by/4.0/ | We study how the nuclear symmetry energy slope affects the properties of
hadron-quark pasta phases and their sizes in massive hybrid stars ($> 2
M_{\odot}$). We utilize the relativistic mean-field model with a
density-dependent isovector coupling constant fitted by varying values of the
symmetry energy slope $L$ to describe the hadronic matter, while the quark
matter is described by a modified MIT bag model with vector interactions. We
construct the hadron-quark phase transition using the energy minimization (EM)
method and compare with the Gibbs construction (GC) in the hybrid star. We find
that the massive hybrid stars generally tend to have a hadron-quark pasta phase
core rather than a pure quark core for most values of $L$, except in some
special cases where they have a small pure quark core. Furthermore, we show
that the mass and size of the pasta phase core are strongly influenced by the
symmetry energy. The radius associated with the onset of pasta phases is also
influenced by the symmetry energy. Additionally, we also evaluate the effects
of the surface tension $\sigma$, the bag constant $B$, and the vector
interaction $G_{V}$.
| [
{
"version": "v1",
"created": "Thu, 13 Mar 2025 02:12:20 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 01:22:31 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Ju",
"Min",
""
],
[
"Wu",
"Xuhao",
""
],
[
"Shen",
"Hong",
""
]
] |
2503.09988 | Zijian Zhao | Zijian Zhao, Xuming Zhang, Jiayu Wen, Mingwen Liu, Xiaoteng Ma | Label Unbalance in High-frequency Trading | Technical Report | null | null | null | cs.LG cs.AI q-fin.CP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In financial trading, return prediction is one of the foundation for a
successful trading system. By the fast development of the deep learning in
various areas such as graphical processing, natural language, it has also
demonstrate significant edge in handling with financial data. While the success
of the deep learning relies on huge amount of labeled sample, labeling each
time/event as profitable or unprofitable, under the transaction cost,
especially in the high-frequency trading world, suffers from serious label
imbalance issue.In this paper, we adopts rigurious end-to-end deep learning
framework with comprehensive label imbalance adjustment methods and succeed in
predicting in high-frequency return in the Chinese future market. The code for
our method is publicly available at
https://github.com/RS2002/Label-Unbalance-in-High-Frequency-Trading .
| [
{
"version": "v1",
"created": "Thu, 13 Mar 2025 02:55:06 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 08:40:48 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Zhao",
"Zijian",
""
],
[
"Zhang",
"Xuming",
""
],
[
"Wen",
"Jiayu",
""
],
[
"Liu",
"Mingwen",
""
],
[
"Ma",
"Xiaoteng",
""
]
] |
2503.10008 | Abu Saleh Musa Patoary | Abu Musa Patoary, Amit Vikram and Victor Galitski | A discrete Fourier transform based quantum circuit for modular
multiplication in Shor's algorithm | 11 pages, 5 figures | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | Shor's algorithm for the prime factorization of numbers provides an
exponential speedup over the best known classical algorithms. However,
nontrivial practical applications have remained out of reach due to
experimental limitations. The bottleneck of the experimental realization of the
algorithm is the modular exponentiation operation. In this paper, based on a
relation between the modular multiplication operator and generalizations of
discrete Fourier transforms, we propose a quantum circuit for modular
exponentiation. A distinctive feature of our proposal is that our circuit can
be entirely implemented in terms of the standard quantum circuit for the
discrete Fourier transform and its variants. The gate-complexity of our
proposal is $O(L^3)$ where L is the number of bits required to store the number
being factorized. It is possible that such a proposal may provide easier
avenues for near-term generic implementations of Shor's algorithm, in contrast
to existing realizations which have often explicitly adapted the circuit to the
number being factorized.
| [
{
"version": "v1",
"created": "Thu, 13 Mar 2025 03:39:13 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 14:31:05 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Patoary",
"Abu Musa",
""
],
[
"Vikram",
"Amit",
""
],
[
"Galitski",
"Victor",
""
]
] |
2503.10428 | Dibyakanti Kumar | Dibyakanti Kumar, Samyak Jha, Anirbit Mukherjee | Langevin Monte-Carlo Provably Learns Depth Two Neural Nets at Any Size
and Data | null | null | null | null | cs.LG math.FA math.PR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work, we will establish that the Langevin Monte-Carlo algorithm can
learn depth-2 neural nets of any size and for any data and we give
non-asymptotic convergence rates for it. We achieve this via showing that under
Total Variation distance and q-Renyi divergence, the iterates of Langevin Monte
Carlo converge to the Gibbs distribution of Frobenius norm regularized losses
for any of these nets, when using smooth activations and in both classification
and regression settings. Most critically, the amount of regularization needed
for our results is independent of the size of the net. This result combines
several recent observations, like our previous papers showing that two-layer
neural loss functions can always be regularized by a certain constant amount
such that they satisfy the Villani conditions, and thus their Gibbs measures
satisfy a Poincare inequality.
| [
{
"version": "v1",
"created": "Thu, 13 Mar 2025 14:50:33 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 15:57:34 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Kumar",
"Dibyakanti",
""
],
[
"Jha",
"Samyak",
""
],
[
"Mukherjee",
"Anirbit",
""
]
] |
2503.10745 | Alexander Swerdlow | Ayush Jain, Alexander Swerdlow, Yuzhou Wang, Sergio Arnaud, Ada
Martin, Alexander Sax, Franziska Meier, Katerina Fragkiadaki | Unifying 2D and 3D Vision-Language Understanding | The first two authors contributed equally | null | null | null | cs.CV cs.AI cs.RO | http://creativecommons.org/licenses/by/4.0/ | Progress in 3D vision-language learning has been hindered by the scarcity of
large-scale 3D datasets. We introduce UniVLG, a unified architecture for 2D and
3D vision-language understanding that bridges the gap between existing
2D-centric models and the rich 3D sensory data available in embodied systems.
Our approach initializes most model weights from pre-trained 2D models and
trains on both 2D and 3D vision-language data. We propose a novel
language-conditioned mask decoder shared across 2D and 3D modalities to ground
objects effectively in both RGB and RGB-D images, outperforming box-based
approaches. To further reduce the domain gap between 2D and 3D, we incorporate
2D-to-3D lifting strategies, enabling UniVLG to utilize 2D data to enhance 3D
performance. With these innovations, our model achieves state-of-the-art
performance across multiple 3D vision-language grounding tasks, demonstrating
the potential of transferring advances from 2D vision-language learning to the
data-constrained 3D domain. Furthermore, co-training on both 2D and 3D data
enhances performance across modalities without sacrificing 2D capabilities. By
removing the reliance on 3D mesh reconstruction and ground-truth object
proposals, UniVLG sets a new standard for realistic, embodied-aligned
evaluation. Code and additional visualizations are available at
https://univlg.github.io .
| [
{
"version": "v1",
"created": "Thu, 13 Mar 2025 17:56:22 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 16:24:10 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Jain",
"Ayush",
""
],
[
"Swerdlow",
"Alexander",
""
],
[
"Wang",
"Yuzhou",
""
],
[
"Arnaud",
"Sergio",
""
],
[
"Martin",
"Ada",
""
],
[
"Sax",
"Alexander",
""
],
[
"Meier",
"Franziska",
""
],
[
"Fragkiadaki",
"Katerina",
""
]
] |
2503.10788 | Leo Noirot M | L\'eo Noirot, C\'ecile X. Yu, Jos\'e C. Abadillo-Uriel, \'Etienne
Dumur, Heimanu Niebojewski, Benoit Bertrand, Romain Maurand, Simon Zihlmann | Coherence of a hole spin flopping-mode qubit in a circuit quantum
electrodynamics environment | null | null | null | null | cond-mat.mes-hall quant-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | The entanglement of microwave photons and spin qubits in silicon represents a
pivotal step forward for quantum information processing utilizing semiconductor
quantum dots. Such hybrid spin circuit quantum electrodynamics (cQED) has been
achieved by granting a substantial electric dipole moment to a spin by
de-localizing it in a double quantum dot under spin-orbit interaction, thereby
forming a flopping-mode (FM) spin qubit. Despite its promise, the coherence
properties demonstrated to date remain insufficient to envision FM spin qubits
as practical single qubits. Here, we present a FM hole spin qubit in a silicon
nanowire coupled to a high-impedance niobium nitride microwave resonator for
readout. We report Rabi frequencies exceeding 100 MHz and coherence times in
the microsecond range, resulting in a high single gate quality factor of 380.
This establishes FM spin qubits as fast and reliable qubits. Moreover, using
the large frequency tunability of the FM qubit, we reveal for the first time
that photonic effects predominantly limit coherence, with radiative decay being
the main relaxation channel and photon shot-noise inducing dephasing. These
results highlight that optimized microwave engineering can unlock the potential
of FM spin qubits in hybrid cQED architectures, offering a scalable and robust
platform for fast and coherent spin qubits with strong coupling to microwave
photons.
| [
{
"version": "v1",
"created": "Thu, 13 Mar 2025 18:26:59 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Noirot",
"Léo",
""
],
[
"Yu",
"Cécile X.",
""
],
[
"Abadillo-Uriel",
"José C.",
""
],
[
"Dumur",
"Étienne",
""
],
[
"Niebojewski",
"Heimanu",
""
],
[
"Bertrand",
"Benoit",
""
],
[
"Maurand",
"Romain",
""
],
[
"Zihlmann",
"Simon",
""
]
] |
2503.10825 | Pavel Fileviez Perez | Hridoy Debnath, Pavel Fileviez Perez | On Gauge Theories of Neutrino Masses and Dark Matter | null | null | null | null | hep-ph astro-ph.CO hep-ex | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We discuss the predictions in the simplest theory for neutrino masses based
on the spontaneous breaking of local lepton number. This theory provides a
simple theoretical framework to understand the possible relation between the
origin of neutrino masses and the nature of the dark matter. In this theory,
one of the fields needed for anomaly cancellation is a dark matter candidate
and the local lepton number is broken at the low scale. We discuss in great
detail the dark matter properties showing the allowed parameter space by the
relic density bounds and the predictions for direct detection. The predictions
for gamma and neutrino lines from dark matter annihilation are investigated. In
the case of Dirac neutrinos, the bound on the effective number of relativistic
degrees of freedom plays an important role and the predictions for gamma lines
could be tested in the near future. We discuss the predictions in the case of
Majorana neutrinos where the dark matter candidate has extra annihilation
channels and compare all the predictions to the case with Dirac neutrinos.
| [
{
"version": "v1",
"created": "Thu, 13 Mar 2025 19:15:37 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Debnath",
"Hridoy",
""
],
[
"Perez",
"Pavel Fileviez",
""
]
] |
2503.10841 | Xiaoxiao Tang | Xiao-Xiao Tang, Peng Wang, Wei Wang, Ming-Jie Sheng, Hao-Ran Yu,
Haojie Xu | Cosmic filament spin -- I: a comparative study in observation | 14 pages, 6 figures and 1 tables, accepted for publication in ApJ | null | null | null | astro-ph.CO | http://creativecommons.org/licenses/by/4.0/ | In the cosmic web, filaments play a crucial role in connecting walls to
clusters and also act as an important stage for galaxy formation and evolution.
Recent observational studies claim that filaments have spin. In this study, we
examined the potential impact of diversity in filament identification
algorithms and galaxy survey datasets on the quantification of filament spin.
The results of this study demonstrate qualitative agreement with previous
research, suggesting that a reliable filament spin signal is detectable when
the viewing angle of filament spine larger than 80 degrees under a rough
estimation. The detected filament spin signal is intricately linked to the
viewing angle, dynamic temperature, etc. The quantitative difference of
filament spin signal among samples is slightly dependent on the filament
identification algorithms, while the value is relatively greater dependent on
the redshift space distortion effect in the galaxy sample.
| [
{
"version": "v1",
"created": "Thu, 13 Mar 2025 19:47:29 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 02:14:09 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Tang",
"Xiao-Xiao",
""
],
[
"Wang",
"Peng",
""
],
[
"Wang",
"Wei",
""
],
[
"Sheng",
"Ming-Jie",
""
],
[
"Yu",
"Hao-Ran",
""
],
[
"Xu",
"Haojie",
""
]
] |
2503.10844 | Xiaoxiao Tang | Peng Wang, Xiao-Xiao Tang, Hao-Da Wang, Noam I. Libeskind, Elmo
Tempel, Wei Wang, Youcai Zhang, Ming-Jie Sheng, Hao-Ran Yu, Haojie Xu | Cosmic filament spin -- II: filament spin and its impact on galaxy
spin-filament alignment in a cosmological simulation | 11 pages, 7 figures, accepted for publication in ApJ | null | null | null | astro-ph.CO | http://creativecommons.org/licenses/by/4.0/ | Observational studies have reported that cosmic filaments on the megaparsec
scale exhibit rotational motion. Subsequent simulation studies have shown
qualitative agreement with these findings, but quantitative discrepancies
remain due to differences in data and methods, which require verification. To
address this issue, we adopt the same methodology as used in the observations
to identify filament spin from the galaxy distribution constructed from a
hydrodynamic simulation. Using the same approach to measure filament spin, we
find that the simulation results closely match the observational findings, with
only minor discrepancies arising from slight differences in the fraction of
filaments classified as dynamically cold or hot based on their dynamic
temperature. Additionally, an analysis of how filament spin affects the galaxy
spin-filament correlation shows that filaments with strong spin signals and
dynamically cold have a greater impact on the galaxy spin-filament correlation
than those with weaker spin signals and dynamically hot filaments. These
results not only provide further evidence that cosmic filaments exhibit spin,
but also highlight the importance of this rotation in the acquisition of
angular momentum by individual galaxies. Future studies exploring the influence
of filament spin on galaxy spin may shed light on the physical origins of
filaments and the angular momentum of galaxies.
| [
{
"version": "v1",
"created": "Thu, 13 Mar 2025 19:53:00 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 02:27:45 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Wang",
"Peng",
""
],
[
"Tang",
"Xiao-Xiao",
""
],
[
"Wang",
"Hao-Da",
""
],
[
"Libeskind",
"Noam I.",
""
],
[
"Tempel",
"Elmo",
""
],
[
"Wang",
"Wei",
""
],
[
"Zhang",
"Youcai",
""
],
[
"Sheng",
"Ming-Jie",
""
],
[
"Yu",
"Hao-Ran",
""
],
[
"Xu",
"Haojie",
""
]
] |
2503.10999 | Moka Nishigaki | Moka Nishigaki, Peter Behroozi, Masami Ouchi, Hong Guo, Rachel S.
Somerville, Anna R. Gallazzi, Kimihiko Nakajima, Kuria Watanabe | ChemicalUniverseMachine I: Uncovering the Cosmic Evolution of Metals in
the Galaxy-ISM-CGM Ecosystem | 27 pages, 17 figures, accepted for publication in ApJ | null | null | null | astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | We present an empirical chemical evolution model that explains the
distribution of metals in the interstellar medium (ISM) and the circumgalactic
medium (CGM) of galaxies based on the UniverseMachine and
NeutralUniverseMachine models in the framework of $\Lambda$CDM structure
formation. We parameterize the fractions of outflowing metals returned and
mixed into the multi-phase ISM of the star-forming regions ($f_{\rm H2}$) and
into the neutral gas regions ($f_{\rm HI}$); metal production, transfer, and
dilution are caused by star formation, galaxy mergers, and gas inflow from the
inter-galactic medium, respectively, with rates determined by the
(Neutral)UniverseMachine models. Using a Markov Chain Monte Carlo algorithm, we
explore the posterior distributions of metal return and mixing consistent with
observed mass-metallicity relations in HII regions (at $0<z<5$), HI damped
Lyman-alpha systems (at $1<z<4$), and the CGM (at $z=0$). We find that the
fraction of metals present in the ISM, $f_{\rm H2}+f_{\rm HI}$, increases with
halo mass from $\sim20$\% at $10^{10}M_\odot$ to $\sim80$\% at
$10^{13}M_\odot$. These fractions increase mildly at higher redshifts, to
$\sim30$\% at $10^{10}M_\odot$ and $\sim80$\% at $10^{13}M_\odot$ at $z=5$.
Interestingly, there is no significant redshift evolution of $f_{\rm H2}+f_{\rm
HI}$ at fixed circular velocity, suggesting that metal distribution between the
ISM and CGM is universally determined by the halo potential well depth. CGM
metal enrichment is thus slow in high-$z$ halos with deep potential wells.
While $f_{\rm H2}$ monotonically increases with halo mass, $f_{\rm HI}$ peaks
at $\sim10^{12}-10^{13} M_\odot$, suggesting that reinfall may be inefficient
in larger-mass halos.
| [
{
"version": "v1",
"created": "Fri, 14 Mar 2025 01:53:52 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 15:25:02 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Nishigaki",
"Moka",
""
],
[
"Behroozi",
"Peter",
""
],
[
"Ouchi",
"Masami",
""
],
[
"Guo",
"Hong",
""
],
[
"Somerville",
"Rachel S.",
""
],
[
"Gallazzi",
"Anna R.",
""
],
[
"Nakajima",
"Kimihiko",
""
],
[
"Watanabe",
"Kuria",
""
]
] |
2503.11022 | Hai Huang | Hai Huang, Ziteng Xu, and Zhaoyu Zhang | Towards Efficient PCSEL Design: A Data-Driven Approach for Design
Insights | null | null | null | null | physics.optics | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a data-driven design approach for photonic crystals to achieve
high efficiency in photonic crystal surface-emitting lasers (PCSELs). By
discretizing the photonic crystal structure into a grid, we enable the
generation of arbitrary lattice designs. Multiple fully connected layers
combined with a position embedding module extract essential features from the
photonic crystal designs, while coupled-wave theory (CWT) is used to evaluate
the efficiency (based on the ratio of surface-emitting to edge-emitting
resonant) and quality factor Q. We introduce the Neural Networks (NNs) model to
evaluate the structures, and to find a better performance design according to
the evaluation result. The model achieves high prediction accuracy, with
Pearson correlation coefficients of 0.780 for SEE and 0.887 for the
log-transformed Q. Additionally, we perform Shapley value analysis to identify
the most important Fourier coefficients, providing insights into the factors
that impact the performance of PCSEL designs. Our work speeds up the design
process and offers valuable guidance for optimizing high-performance PCSELs,
supporting the development of fully photonic design automation (PDA).
| [
{
"version": "v1",
"created": "Fri, 14 Mar 2025 02:40:30 GMT"
},
{
"version": "v2",
"created": "Tue, 18 Mar 2025 10:14:24 GMT"
},
{
"version": "v3",
"created": "Thu, 20 Mar 2025 06:27:45 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Huang",
"Hai",
""
],
[
"Xu",
"Ziteng",
""
],
[
"Zhang",
"Zhaoyu",
""
]
] |
2503.11031 | Anirban Chandra | Anirban Chandra, Marius Koch, Suraj Pawar, Aniruddha Panda, Kamyar
Azizzadenesheli, Jeroen Snippe, Faruk O. Alpak, Farah Hariri, Clement
Etienam, Pandu Devarakota, Anima Anandkumar, Detlef Hohl | Fourier Neural Operator based surrogates for $CO_2$ storage in realistic
geologies | null | null | null | null | physics.comp-ph cs.AI physics.geo-ph | http://creativecommons.org/licenses/by/4.0/ | This study aims to develop surrogate models for accelerating decision making
processes associated with carbon capture and storage (CCS) technologies.
Selection of sub-surface $CO_2$ storage sites often necessitates expensive and
involved simulations of $CO_2$ flow fields. Here, we develop a Fourier Neural
Operator (FNO) based model for real-time, high-resolution simulation of $CO_2$
plume migration. The model is trained on a comprehensive dataset generated from
realistic subsurface parameters and offers $O(10^5)$ computational acceleration
with minimal sacrifice in prediction accuracy. We also explore super-resolution
experiments to improve the computational cost of training the FNO based models.
Additionally, we present various strategies for improving the reliability of
predictions from the model, which is crucial while assessing actual geological
sites. This novel framework, based on NVIDIA's Modulus library, will allow
rapid screening of sites for CCS. The discussed workflows and strategies can be
applied to other energy solutions like geothermal reservoir modeling and
hydrogen storage. Our work scales scientific machine learning models to
realistic 3D systems that are more consistent with real-life subsurface
aquifers/reservoirs, paving the way for next-generation digital twins for
subsurface CCS applications.
| [
{
"version": "v1",
"created": "Fri, 14 Mar 2025 02:58:24 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 15:44:45 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Chandra",
"Anirban",
""
],
[
"Koch",
"Marius",
""
],
[
"Pawar",
"Suraj",
""
],
[
"Panda",
"Aniruddha",
""
],
[
"Azizzadenesheli",
"Kamyar",
""
],
[
"Snippe",
"Jeroen",
""
],
[
"Alpak",
"Faruk O.",
""
],
[
"Hariri",
"Farah",
""
],
[
"Etienam",
"Clement",
""
],
[
"Devarakota",
"Pandu",
""
],
[
"Anandkumar",
"Anima",
""
],
[
"Hohl",
"Detlef",
""
]
] |
2503.11032 | Lilin Zhang | Lilin Zhang, Chengpei Wu, Ning Yang | Weakly Supervised Contrastive Adversarial Training for Learning Robust
Features from Semi-supervised Data | This paper has been accepted by CVPR 2025 | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Existing adversarial training (AT) methods often suffer from incomplete
perturbation, meaning that not all non-robust features are perturbed when
generating adversarial examples (AEs). This results in residual correlations
between non-robust features and labels, leading to suboptimal learning of
robust features. However, achieving complete perturbation, i.e., perturbing as
many non-robust features as possible, is challenging due to the difficulty in
distinguishing robust and non-robust features and the sparsity of labeled data.
To address these challenges, we propose a novel approach called Weakly
Supervised Contrastive Adversarial Training (WSCAT). WSCAT ensures complete
perturbation for improved learning of robust features by disrupting
correlations between non-robust features and labels through complete AE
generation over partially labeled data, grounded in information theory.
Extensive theoretical analysis and comprehensive experiments on widely adopted
benchmarks validate the superiority of WSCAT. Our code is available at
https://github.com/zhang-lilin/WSCAT.
| [
{
"version": "v1",
"created": "Fri, 14 Mar 2025 03:01:10 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 09:17:56 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Zhang",
"Lilin",
""
],
[
"Wu",
"Chengpei",
""
],
[
"Yang",
"Ning",
""
]
] |
2503.11249 | Viet-Hoang Tran | Viet-Hoang Tran, Thanh T. Chu, Khoi N.M. Nguyen, Trang Pham, Tam Le,
Tan M. Nguyen | Spherical Tree-Sliced Wasserstein Distance | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Sliced Optimal Transport (OT) simplifies the OT problem in high-dimensional
spaces by projecting supports of input measures onto one-dimensional lines and
then exploiting the closed-form expression of the univariate OT to reduce the
computational burden of OT. Recently, the Tree-Sliced method has been
introduced to replace these lines with more intricate structures, known as tree
systems. This approach enhances the ability to capture topological information
of integration domains in Sliced OT while maintaining low computational cost.
Inspired by this approach, in this paper, we present an adaptation of tree
systems on OT problems for measures supported on a sphere. As a counterpart to
the Radon transform variant on tree systems, we propose a novel spherical Radon
transform with a new integration domain called spherical trees. By leveraging
this transform and exploiting the spherical tree structures, we derive
closed-form expressions for OT problems on the sphere. Consequently, we obtain
an efficient metric for measures on the sphere, named Spherical Tree-Sliced
Wasserstein (STSW) distance. We provide an extensive theoretical analysis to
demonstrate the topology of spherical trees and the well-definedness and
injectivity of our Radon transform variant, which leads to an orthogonally
invariant distance between spherical measures. Finally, we conduct a wide range
of numerical experiments, including gradient flows and self-supervised
learning, to assess the performance of our proposed metric, comparing it to
recent benchmarks.
| [
{
"version": "v1",
"created": "Fri, 14 Mar 2025 10:00:13 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 11:04:51 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Tran",
"Viet-Hoang",
""
],
[
"Chu",
"Thanh T.",
""
],
[
"Nguyen",
"Khoi N. M.",
""
],
[
"Pham",
"Trang",
""
],
[
"Le",
"Tam",
""
],
[
"Nguyen",
"Tan M.",
""
]
] |
2503.11623 | Rhea Hoyer | Rhea Hoyer, P. Peter Stavropoulos, Aleksandar Razpopov, Roser
Valent\'i, Libor \v{S}mejkal, Alexander Mook | Altermagnetic splitting of magnons in hematite ($\alpha$-Fe$_2$O$_3$) | 22 pages, 17 figures | null | null | null | cond-mat.str-el cond-mat.mtrl-sci | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We develop a four-sublattice spin-wave theory for the $g$-wave altermagnet
candidate hematite ($\alpha$-Fe$_2$O$_3$), considering both its easy-axis phase
below and its weak ferromagnetic phase above the Morin temperature. A key
question is whether the defining altermagnetic feature - magnon spin splitting
(also called chirality or polarization splitting) due to nonrelativistic
time-reversal symmetry breaking - remains intact when relativistic corrections,
which contribute to hematite's magnetic order, are included. Using a detailed
symmetry analysis supported by density functional theory, we show that
capturing the magnon splitting within a Heisenberg model requires exchange
interactions extending at least to the 13th neighbor. We find an altermagnetic
band splitting of approximately 2 meV, which contrasts with the total band
width of about 100 meV. To evaluate the experimental observability of this
splitting, we analyze relativistic corrections to the magnon spectrum in both
magnetic phases. We show that spin-orbit coupling - manifesting as
magnetocrystalline anisotropies and the Dzyaloshinskii-Moriya interaction (DMI)
- does not obscure the key altermagnetic features. These findings indicate that
inelastic neutron scattering can directly probe altermagnetic magnon splitting
in hematite. We also discuss implications for magnon transport, particularly
magnonic contributions to the thermal Hall effect (which requires spin-orbit
coupling) and to spin splitter effects (which do not). Notably, we predict a
third-order nonlinear magnon spin splitter effect. This result suggests that
the $g$-wave magnon spin splitting in hematite enables transverse heat-to-spin
conversion without requiring an external magnetic field.
| [
{
"version": "v1",
"created": "Fri, 14 Mar 2025 17:44:45 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 09:28:48 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Hoyer",
"Rhea",
""
],
[
"Stavropoulos",
"P. Peter",
""
],
[
"Razpopov",
"Aleksandar",
""
],
[
"Valentí",
"Roser",
""
],
[
"Šmejkal",
"Libor",
""
],
[
"Mook",
"Alexander",
""
]
] |
2503.11690 | Anirban Chakraborti | Ashok Kumar, Abbinav Sankar Kailasam, Anish Rai, Sudeep Shukla,
Sourish Das and Anirban Chakraborti | The Impact of Meteorological Factors on Crop Price Volatility in India:
Case studies of Soybean and Brinjal | 15 pages, 9 figures | null | null | null | stat.AP econ.GN physics.soc-ph q-fin.EC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Climate is an evolving complex system with dynamic interactions and
non-linear feedback mechanisms, shaping environmental and socio-economic
outcomes. Crop production is highly sensitive to such climatic fluctuations.
This paper studies the price volatility of agricultural crops as influenced by
meteorological variables (and many other environmental, social and governance
factors), which is a critical challenge in sustainable finance, agricultural
planning, and policy-making. As case studies, we choose the two Indian states
of Madhya Pradesh (for Soybean) and Odisha (for Brinjal). We employ an
Exponential Generalized Autoregressive Conditional Heteroskedasticity (EGARCH)
model to estimate the conditional volatility of the log returns of crop prices
from 2012 to 2024. This study further explores the cross-correlations between
volatility and the meteorological variables. Further, a Granger-causality test
is carried out to analyze the causal effect of meteorological variables on the
price volatility. Finally, the Seasonal Auto-Regressive Integrated Moving
Average with Exogenous Regressors (SARIMAX) and Long Short-Term Memory (LSTM)
models are implemented as simple machine learning models of price volatility
with meteorological factors as exogenous variables. We believe that this will
illustrate the usefulness of simple machine learning models in agricultural
finance, and help the farmers to make informed decisions by considering climate
patterns and making beneficial decisions with regard to crop rotation or
allocations. In general, incorporating meteorological factors to assess
agricultural performance could help to understand and reduce price volatility
and possibly lead to economic stability.
| [
{
"version": "v1",
"created": "Thu, 6 Mar 2025 18:48:04 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 17:40:13 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Kumar",
"Ashok",
""
],
[
"Kailasam",
"Abbinav Sankar",
""
],
[
"Rai",
"Anish",
""
],
[
"Shukla",
"Sudeep",
""
],
[
"Das",
"Sourish",
""
],
[
"Chakraborti",
"Anirban",
""
]
] |
2503.11749 | A. Torres-Hernandez | Anthony Torres-Hernandez | Proposal for the Application of Fractional Operators in Polynomial
Regression Models to Enhance the Determination Coefficient $R^2$ on Unseen
Data | null | null | null | null | stat.ME cs.NA math.NA math.ST physics.comp-ph physics.data-an stat.TH | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Since polynomial regression models are generally quite reliable for data with
a linear trend, it is important to note that, in some cases, they may encounter
overfitting issues during the training phase, which could result in negative
values of the coefficient of determination $R^2$ for unseen data. For this
reason, this work proposes the partial implementation of fractional operators
in polynomial regression models to generate a fractional regression model. The
goal of this proposal is to attempt to mitigate overfitting, which could
improve the value of the coefficient of determination for unseen data, compared
to the polynomial model, under the assumption that this would contribute to
generating predictive models with better performance. The methodology for
constructing these fractional regression models is detailed, and examples
applicable to both Riemann-Liouville and Caputo fractional operators are
presented.
| [
{
"version": "v1",
"created": "Fri, 14 Mar 2025 17:54:39 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 10:05:27 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Torres-Hernandez",
"Anthony",
""
]
] |
2503.11989 | Dharani Chandra | Dharani Chandra | Applications of Large Language Model Reasoning in Feature Generation | I just updated the format of the references in the paper | null | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Large Language Models (LLMs) have revolutionized natural language processing
through their state of art reasoning capabilities. This paper explores the
convergence of LLM reasoning techniques and feature generation for machine
learning tasks. We examine four key reasoning approaches: Chain of Thought,
Tree of Thoughts, Retrieval-Augmented Generation, and Thought Space
Exploration. Our analysis reveals how these approaches can be used to identify
effective feature generation rules without having to manually specify search
spaces. The paper categorizes LLM-based feature generation methods across
various domains including finance, healthcare, and text analytics. LLMs can
extract key information from clinical notes and radiology reports in
healthcare, by enabling more efficient data utilization. In finance, LLMs
facilitate text generation, summarization, and entity extraction from complex
documents. We analyze evaluation methodologies for assessing feature quality
and downstream performance, with particular attention to OCTree's decision tree
reasoning approach that provides language-based feedback for iterative
improvements. Current challenges include hallucination, computational
efficiency, and domain adaptation. As of March 2025, emerging approaches
include inference-time compute scaling, reinforcement learning, and supervised
fine-tuning with model distillation. Future directions point toward multimodal
feature generation, self-improving systems, and neuro-symbolic approaches. This
paper provides a detailed overview of an emerging field that promises to
automate and enhance feature engineering through language model reasoning.
| [
{
"version": "v1",
"created": "Sat, 15 Mar 2025 04:18:01 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 02:18:33 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Chandra",
"Dharani",
""
]
] |
2503.12277 | Quanyu Tang | Zheng Li, Quanyu Tang | On a conjecture of Erd\H{o}s and Graham about the Sylvester's sequence | 23 pages; v2 generalizes the previous results; v3 fixes some
typographical errors and adds several remarks | null | null | null | math.NT math.CA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Let $\{u_n\}$ be the Sylvester's sequence (sequence A000058 in the OEIS), and
let $ a_1 < a_2 < \cdots $ be any other positive integer sequence satisfying $
\sum_{k=1}^\infty \frac{1}{a_k} = 1 $. In this paper, we solve a conjecture of
Erd\H{o}s and Graham, which asks whether $$ \liminf_{n\to\infty}
a_n^{\frac{1}{2^n}} < \lim_{n\to\infty} u_n^{\frac{1}{2^n}} = c_0 =
1.264085\ldots. $$ We prove this conjecture using a constructive approach.
Furthermore, assuming that the unproven claim of Erd\H{o}s and Graham that "all
rationals have eventually greedy best Egyptian underapproximations" holds, we
establish a generalization of this conjecture using a non-constructive
approach. [This paper solves Problem 315 on Bloom's website "Erd\H{o}s
problems".]
| [
{
"version": "v1",
"created": "Sat, 15 Mar 2025 22:22:47 GMT"
},
{
"version": "v2",
"created": "Tue, 18 Mar 2025 07:20:42 GMT"
},
{
"version": "v3",
"created": "Thu, 20 Mar 2025 17:21:37 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Li",
"Zheng",
""
],
[
"Tang",
"Quanyu",
""
]
] |
2503.12461 | Fanhu Zeng | Fanhu Zeng, Hao Tang, Yihua Shao, Siyu Chen, Ling Shao, Yan Wang | MambaIC: State Space Models for High-Performance Learned Image
Compression | Accepted to CVPR 2025 | null | null | null | cs.CV eess.IV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A high-performance image compression algorithm is crucial for real-time
information transmission across numerous fields. Despite rapid progress in
image compression, computational inefficiency and poor redundancy modeling
still pose significant bottlenecks, limiting practical applications. Inspired
by the effectiveness of state space models (SSMs) in capturing long-range
dependencies, we leverage SSMs to address computational inefficiency in
existing methods and improve image compression from multiple perspectives. In
this paper, we integrate the advantages of SSMs for better
efficiency-performance trade-off and propose an enhanced image compression
approach through refined context modeling, which we term MambaIC. Specifically,
we explore context modeling to adaptively refine the representation of hidden
states. Additionally, we introduce window-based local attention into
channel-spatial entropy modeling to reduce potential spatial redundancy during
compression, thereby increasing efficiency. Comprehensive qualitative and
quantitative results validate the effectiveness and efficiency of our approach,
particularly for high-resolution image compression. Code is released at
https://github.com/AuroraZengfh/MambaIC.
| [
{
"version": "v1",
"created": "Sun, 16 Mar 2025 11:32:34 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 02:27:03 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Zeng",
"Fanhu",
""
],
[
"Tang",
"Hao",
""
],
[
"Shao",
"Yihua",
""
],
[
"Chen",
"Siyu",
""
],
[
"Shao",
"Ling",
""
],
[
"Wang",
"Yan",
""
]
] |
2503.12468 | Dingping Li | Yingze Su, Ruitao Xiao, Junnian Xiong, Hui Li, Huaqing Huang and
Dingping Li | Nonpertubative Many-Body Theory for the Two-Dimensional Hubbard Model at
Low Temperature: From Weak to Strong Coupling Regimes | 42 pages, 7 figures | null | null | null | cond-mat.str-el physics.atom-ph physics.comp-ph | http://creativecommons.org/licenses/by/4.0/ | In the theoretical study of two-dimensional systems, difficulties emerge as
that quantum phase transitions at zero temperature cause low- and
high-temperature scenarios to belong to different branches, while the
Mermin-Wagner theorem prohibits continuous symmetry breaking at finite
temperatures, excluding a Landau phase transition marked by a critical
temperature $T_c$. In many-body theory, fundamental symmetries like
Ward-Takahashi identity (WTI), Fluctuation-Dissipation theorem (FDT), and
Pauli-repulsive principle must be satisfied, yet widely-used approximate
theories struggle to meet them simultaneously. We introduce a symmetrization
theory that, by using spurious broken phases from approximate theories, can
naturally generate different low- and high-temperature branches. Employing the
GW method beyond mean-field and the covariance scheme, which strictly satisfy
WTI and FDT, we numerically show that the violation of Pauli-repulsive
principle is significantly less than that in mean-field, and further restore
Pauli-repulsive principle without breaking WTI and FDT. By calculating
symmetrized one-body Green's function and two-body correlation function and
comparing them with accurate results from Determinant Quantum Monte Carlo
(DQMC), we demonstrate its numerical accuracy at low temperatures and strong
coupling. This symmetrization theory can be easily applied to other
two-dimensional systems at low temperatures.
| [
{
"version": "v1",
"created": "Sun, 16 Mar 2025 11:49:03 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Su",
"Yingze",
""
],
[
"Xiao",
"Ruitao",
""
],
[
"Xiong",
"Junnian",
""
],
[
"Li",
"Hui",
""
],
[
"Huang",
"Huaqing",
""
],
[
"Li",
"Dingping",
""
]
] |
2503.12478 | Zhiyu Liang | Zhiyu Liang, Dongrui Cai, Chenyuan Zhang, Zheng Liang, Chen Liang, Bo
Zheng, Shi Qiu, Jin Wang, Hongzhi Wang | KDSelector: A Knowledge-Enhanced and Data-Efficient Model Selector
Learning Framework for Time Series Anomaly Detection | This paper has been accepted by SIGMOD 2025 | null | null | null | cs.LG cs.AI cs.DB | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Model selection has been raised as an essential problem in the area of time
series anomaly detection (TSAD), because there is no single best TSAD model for
the highly heterogeneous time series in real-world applications. However,
despite the success of existing model selection solutions that train a
classification model (especially neural network, NN) using historical data as a
selector to predict the correct TSAD model for each series, the NN-based
selector learning methods used by existing solutions do not make full use of
the knowledge in the historical data and require iterating over all training
samples, which limits the accuracy and training speed of the selector. To
address these limitations, we propose KDSelector, a novel knowledge-enhanced
and data-efficient framework for learning the NN-based TSAD model selector, of
which three key components are specifically designed to integrate available
knowledge into the selector and dynamically prune less important and redundant
samples during the learning. We develop a TSAD model selection system with
KDSelector as the internal, to demonstrate how users improve the accuracy and
training speed of their selectors by using KDSelector as a plug-and-play
module. Our demonstration video is hosted at https://youtu.be/2uqupDWvTF0.
| [
{
"version": "v1",
"created": "Sun, 16 Mar 2025 12:13:19 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 03:06:28 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Liang",
"Zhiyu",
""
],
[
"Cai",
"Dongrui",
""
],
[
"Zhang",
"Chenyuan",
""
],
[
"Liang",
"Zheng",
""
],
[
"Liang",
"Chen",
""
],
[
"Zheng",
"Bo",
""
],
[
"Qiu",
"Shi",
""
],
[
"Wang",
"Jin",
""
],
[
"Wang",
"Hongzhi",
""
]
] |
2503.12552 | Tianyu Li | Tianyu Li, Yihang Qiu, Zhenhua Wu, Carl Lindstr\"om, Peng Su, Matthias
Nie{\ss}ner, Hongyang Li | MTGS: Multi-Traversal Gaussian Splatting | null | null | null | null | cs.CV cs.GR | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Multi-traversal data, commonly collected through daily commutes or by
self-driving fleets, provides multiple viewpoints for scene reconstruction
within a road block. This data offers significant potential for high-quality
novel view synthesis, which is crucial for applications such as autonomous
vehicle simulators. However, inherent challenges in multi-traversal data often
result in suboptimal reconstruction quality, including variations in appearance
and the presence of dynamic objects. To address these issues, we propose
Multi-Traversal Gaussian Splatting (MTGS), a novel approach that reconstructs
high-quality driving scenes from arbitrarily collected multi-traversal data by
modeling a shared static geometry while separately handling dynamic elements
and appearance variations. Our method employs a multi-traversal dynamic scene
graph with a shared static node and traversal-specific dynamic nodes,
complemented by color correction nodes with learnable spherical harmonics
coefficient residuals. This approach enables high-fidelity novel view synthesis
and provides flexibility to navigate any viewpoint. We conduct extensive
experiments on a large-scale driving dataset, nuPlan, with multi-traversal
data. Our results demonstrate that MTGS improves LPIPS by 23.5% and geometry
accuracy by 46.3% compared to single-traversal baselines. The code and data
would be available to the public.
| [
{
"version": "v1",
"created": "Sun, 16 Mar 2025 15:46:12 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 08:09:23 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Li",
"Tianyu",
""
],
[
"Qiu",
"Yihang",
""
],
[
"Wu",
"Zhenhua",
""
],
[
"Lindström",
"Carl",
""
],
[
"Su",
"Peng",
""
],
[
"Nießner",
"Matthias",
""
],
[
"Li",
"Hongyang",
""
]
] |
2503.12741 | Zhao Zhang Joseph | Zhao Joseph Zhang, Kentaro Nagamine, Yuri Oku, Khee-Gan Lee, Keita
Fukushima, Kazuki Tomaru, Bing Zhang, Isabel Medlock and Daisuke Nagai | Probing the cosmic baryon distribution and the impact of AGN feedback
with FRBs in CROCODILE simulation | null | null | null | null | astro-ph.CO astro-ph.HE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate the Missing Baryon problem using Fast Radio Bursts (FRBs) to
trace cosmic baryons. We use the CROCODILE simulations with the GADGET3/4-OSAKA
SPH code, incorporating star formation, supernova (SN), and active galactic
nuclei (AGN) feedback. Light cones generated from large-scale structure
simulations allow us to compute gas density profiles and dispersion measures
(DMs) measurable by FRBs. Our results show that AGN feedback reduces central
gas densities in halos, reshaping the boundary between the circumgalactic
medium (CGM) and intergalactic medium (IGM). Zoom-in simulations further reveal
that AGN feedback significantly modulates foreground halo DM contributions
along different sightlines. By analyzing the DM--redshift (DM--z) relation and
comparing it to the Macquart relation, we constrain the diffuse baryon mass
fraction at z = 1 to fdiff = 0.865 (+0.101, -0.165) (fiducial) and fdiff =
0.856 (+0.101, -0.162) (NoBH), including IGM (fIGM) and halo (fHalos)
contributions. We also quantify the redshift evolution of fdiff and fIGM,
providing the fitting results. Our study provides a framework for understanding
baryon distribution across cosmic structures, FRB host galaxies, and the role
of AGN in shaping foreground DM contributions.
| [
{
"version": "v1",
"created": "Mon, 17 Mar 2025 02:17:32 GMT"
},
{
"version": "v2",
"created": "Thu, 20 Mar 2025 03:30:06 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Zhang",
"Zhao Joseph",
""
],
[
"Nagamine",
"Kentaro",
""
],
[
"Oku",
"Yuri",
""
],
[
"Lee",
"Khee-Gan",
""
],
[
"Fukushima",
"Keita",
""
],
[
"Tomaru",
"Kazuki",
""
],
[
"Zhang",
"Bing",
""
],
[
"Medlock",
"Isabel",
""
],
[
"Nagai",
"Daisuke",
""
]
] |
2503.12742 | Denis Osin | Ionu\c{t} Chifan, Adriana Fern\'andez Quero, Denis Osin and Hui Tan | W$^*$-superrigidity for property (T) groups with infinite center | null | null | null | null | math.OA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose to study a natural version of Connes' Rigidity Conjecture that
involves property (T) groups with infinite center. Utilizing techniques at the
intersection of von Neumann algebras and geometric group theory, we establish
several cases where this conjecture holds. In particular, we provide the first
example of a W$^*$-superrigid property (T) group with infinite center. In the
course of proving our main results, we also generalize the main
W$^*$-superrigidity result from \cite{cios22} to twisted group factors.
| [
{
"version": "v1",
"created": "Mon, 17 Mar 2025 02:20:45 GMT"
},
{
"version": "v2",
"created": "Wed, 19 Mar 2025 20:58:47 GMT"
}
] | 2025-03-21T00:00:00 | [
[
"Chifan",
"Ionuţ",
""
],
[
"Quero",
"Adriana Fernández",
""
],
[
"Osin",
"Denis",
""
],
[
"Tan",
"Hui",
""
]
] |
Subsets and Splits