Searcharxiv⌕ Search

arXiv subjects

Xidan Song

Publications and source records attributed to Xidan Song.

10 recordsLinked to original sources

Beyond Accuracy: A Geometric Stability Analysis of Large Language Models in Chess Evaluation

Large language models (LLMs) are usually evaluated in chess by how closely their position evaluations agree with a strong engine, but such agreement does not show that a model evaluates equivalent positions consistently. We apply metamorphic relations to chess evaluation: a 180-degree rotation of pawnless positions, left-right mirroring, color swapping, position pairs one move apart, conversion between Forsyth-Edwards Notation and a natural-language description, and illegal positions. Six LLMs (Claude Sonnet 4.5, DeepSeek Chat, Gemini 2.5 Flash, GPT-5.1, Grok 4.1 Fast and Kimi K2 Turbo) evaluated seven test sets of 500 items each, with Stockfish 16.1 as the reference. All six models failed on rotation: at most 10.6% of the 500 rotated pairs received evaluations within 300 centipawns (cp) of each other, and the models often declared the legal rotated positions illegal. On the other four transformations, stability ranked the models as accuracy did except for the first two places: GPT-5.1 had the lowest mean stability error (186.17 cp) and Claude Sonnet 4.5 the lowest accuracy error (330.41 cp). Rejection rates for illegal positions ranged from 44.2% to 96.0%. Stability should be reported with the share of consistent pairs, because means over answered pairs concealed most rotation failures.

cs.AI↗

The norm, not the labels: training and evaluating finite-element surrogates for stress analysis in the energy norm

Finite-element surrogates predict the stresses of a new design in one evaluation, but are usually trained on solutions with a displacement loss and compared by it, although engineering decisions rest on stresses. Application in engineering: stress analysis of perforated plates and solids with cavities, parametric families in which holes concentrate stress. Contribution to artificial intelligence: training and evaluating such surrogates in the energy norm, with or without solutions, and screening their predictions without solutions. For linear elastostatics, minimising the assembled discrete potential energy is exactly supervised regression in this norm: the energy gap of a prediction is the strain energy of its stress error. A displacement loss under-weights the stiff modes that carry the strains. In pre-registered comparisons, changing only the norm of the supervised loss cut a three-dimensional transformer's median energy-norm error 7.5-fold and its median von Mises error 6.1-fold; in two dimensions it more than halved the von Mises error, also on held-out instances. In three dimensions, the transformer trained on the energy alone came within a factor of 1.3 of the energy-norm-supervised one in median energy-norm and von Mises error. It was more accurate than the displacement-supervised one on all five metrics in every seed. Given the stiffness matrix, a Ritz projection was more accurate than the label-free transformer, in the seed means, on every metric but peak stress. No network transferred to a substantially finer mesh. The energy ranks predictions of a load case, and its sign flags those worse than the zero field.

cs.CE↗

Arithmetic of qudit dimensions governs erasure correction on the Fano plane and stabilizer-state entanglement

Qudit codes often inherit the parameters of qubit codes, but not necessarily their correctable erasures. We show that a code encoding one qudit of dimension D into seven such qudits that corrects every erasure Steane's code corrects exists exactly when D is a power of 2; for the qutrit color code, exactly when D is odd. This needs no stabilizer assumption: these codes are, up to local unitaries, stabilizer codes of qubits, respectively prime-power qudits. The two patterns differ in one line, which completes the Fano plane only where 1+1=0. As ideal secret sharing schemes, these codes exist exactly where classical ones do: entanglement does not help, unlike for Euler's 36 officers. The same arithmetic makes the entropy cones of qubit and odd-dimensional stabilizer states incomparable from six parties on. At five, the stabilizer cone is strictly smaller than the quantum linear rank cone, answering an open question.

cs.IT↗

Ordered Ruzsa-Szemeredi Numbers at Matching Size Two

Bondy and Szwarcfiter defined $\mathrm{ex}^*(n,F)$ as the largest number of edges in an $n$-vertex graph whose edge set partitions into induced copies of $F$; for $F=2K_2$ the deficiency $\binom{n}{2}-\mathrm{ex}^*(n,2K_2)$ is $Θ(n^{3/2})$. We study the ordered relaxation at fixed matching size, in which each part need only be induced in the union of itself with the parts that follow it; write $\mathrm{ORS}_n(r)$ for the largest number of parts, so that $r\,\mathrm{ORS}_n(r)$ is the ordered analogue of $\mathrm{ex}^*(n,rK_2)$. Our main tool is a characterisation valid for every $r$: an ordered decomposition into induced $r$-matchings is a sequence of steps that start from $K_n$ and repeatedly delete a perfect matching from $2r$ vertices currently spanning a clique. Reading a decomposition backwards turns a condition about the ordering into a reachability question that an exhaustive search can settle. For $r=2$ we determine $\mathrm{ORS}_n(2)$ exactly at orders five through nineteen, where it takes the values $1,3,5,8,11,14,19,23,28,34,40,47,54,62,70$, and we confine $\mathrm{ORS}_{20}(2)$ to $\{78,79\}$. The counting bound $\lfloor n(n-4)/4\rfloor$ is attained at orders five through nine and at eleven, and missed by exactly one part at every other order below twenty, so order eleven is an isolated exception, not a parity effect. Across this range the ordered deficiency equals $\frac32n+O(1)$, and along powers of two a dyadic construction keeps it below $O(n\log n)$; whether it is linear for all $n$ is our main open question. The structural results are formalised in Lean 4, and the searches are certified by fail-closed sweeps and an independent checker.

math.CO↗

Discrete energy as an exact label-free training objective for finite-element surrogates

Supervised training of finite-element (FE) surrogate models requires reference solutions, and each reference solution is obtained by solving the system that the surrogate is intended to replace. The assembled discrete potential energy provides a training signal that requires no reference solution. This note records, with proofs, the identities that make this signal exact for linear elastostatics: the difference between the energy of a prediction and the energy of the reference solution equals one half of the squared stiffness-norm error, and the gradient of the energy equals the stiffness-weighted error. Label-free discrete-energy minimisation and supervised regression in the stiffness norm therefore have the same unique minimiser and identical gradients at every point. Around this central result, the note states a conditioning lemma that bounds the displacement error by the energy gap, a modewise contraction identity that explains why the Euclidean displacement error is an unsuitable primary metric, the Chebyshev bound that governs conjugate-gradient post-processing of surrogate predictions, and a conditional latent-separation proposition for joint-embedding predictive architecture (JEPA) pretraining on a shared stiffness operator, with an explicit numerical counterexample that delimits its scope. Every claim with numeric content is implemented as an executable falsification check; the checks were executed twice, on synthetic test problems and on a probe set of 16 instances from the validation split of a pre-registered experimental run, and every inequality holds, with the measured tightness reported. A closing section explains why the construction does not extend to elastodynamics through direct minimisation of the action functional, and which time-discrete formulation restores exactness.

cs.CE↗

QNNRepair: Quantized Neural Network Repair

We present QNNRepair, the first method in the literature for repairing quantized neural networks (QNNs). QNNRepair aims to improve the accuracy of a neural network model after quantization. It accepts the full-precision and weight-quantized neural networks and a repair dataset of passing and failing tests. At first, QNNRepair applies a software fault localization method to identify the neurons that cause performance degradation during neural network quantization. Then, it formulates the repair problem into a linear programming problem of solving neuron weights parameters, which corrects the QNN's performance on failing tests while not compromising its performance on passing tests. We evaluate QNNRepair with widely used neural network architectures such as MobileNetV2, ResNet, and VGGNet on popular datasets, including high-resolution images. We also compare QNNRepair with the state-of-the-art data-free quantization method SQuant. According to the experiment results, we conclude that QNNRepair is effective in improving the quantized model's performance in most cases. Its repaired models have 24% higher accuracy than SQuant's in the independent validation set, especially for the ImageNet dataset.

cs.LG↗

AIREPAIR: A Repair Platform for Neural Networks

We present AIREPAIR, a platform for repairing neural networks. It features the integration of existing network repair tools. Based on AIREPAIR, one can run different repair methods on the same model, thus enabling the fair comparison of different repair techniques. We evaluate AIREPAIR with three state-of-the-art repair tools on popular deep-learning datasets and models. Our evaluation confirms the utility of AIREPAIR, by comparing and analyzing the results from different repair techniques. A demonstration is available at https://youtu.be/UkKw5neeWhw.

cs.NE↗

CEG4N: Counter-Example Guided Neural Network Quantization Refinement

Neural networks are essential components of learning-based software systems. However, their high compute, memory, and power requirements make using them in low resources domains challenging. For this reason, neural networks are often quantized before deployment. Existing quantization techniques tend to degrade the network accuracy. We propose Counter-Example Guided Neural Network Quantization Refinement (CEG4N). This technique combines search-based quantization and equivalence verification: the former minimizes the computational requirements, while the latter guarantees that the network's output does not change after quantization. We evaluate CEG4N~on a diverse set of benchmarks, including large and small networks. Our technique successfully quantizes the networks in our evaluation while producing models with up to 72% better accuracy than state-of-the-art techniques.

cs.LG↗

QNNVerifier: A Tool for Verifying Neural Networks using SMT-Based Model Checking

QNNVerifier is the first open-source tool for verifying implementations of neural networks that takes into account the finite word-length (i.e. quantization) of their operands. The novel support for quantization is achieved by employing state-of-the-art software model checking (SMC) techniques. It translates the implementation of neural networks to a decidable fragment of first-order logic based on satisfiability modulo theories (SMT). The effects of fixed- and floating-point operations are represented through direct implementations given a hardware-determined precision. Furthermore, QNNVerifier allows to specify bespoke safety properties and verify the resulting model with different verification strategies (incremental and k-induction) and SMT solvers. Finally, QNNVerifier is the first tool that combines invariant inference via interval analysis and discretization of non-linear activation functions to speed up the verification of neural networks by orders of magnitude. A video presentation of QNNVerifier is available at https://youtu.be/7jMgOL41zTY

cs.AI↗

Verifying Quantized Neural Networks using SMT-Based Model Checking

Artificial Neural Networks (ANNs) are being deployed for an increasing number of safety-critical applications, including autonomous cars and medical diagnosis. However, concerns about their reliability have been raised due to their black-box nature and apparent fragility to adversarial attacks. These concerns are amplified when ANNs are deployed on restricted system, which limit the precision of mathematical operations and thus introduce additional quantization errors. Here, we develop and evaluate a novel symbolic verification framework using software model checking (SMC) and satisfiability modulo theories (SMT) to check for vulnerabilities in ANNs. More specifically, we propose several ANN-related optimizations for SMC, including invariant inference via interval analysis, slicing, expression simplifications, and discretization of non-linear activation functions. With this verification framework, we can provide formal guarantees on the safe behavior of ANNs implemented both in floating- and fixed-point arithmetic. In this regard, our verification approach was able to verify and produce adversarial examples for $52$ test cases spanning image classification and general machine learning applications. Furthermore, for small- to medium-sized ANN, our approach completes most of its verification runs in minutes. Moreover, in contrast to most state-of-the-art methods, our approach is not restricted to specific choices regarding activation functions and non-quantized representations. Our experiments show that our approach can analyze larger ANN implementations and substantially reduce the verification time compared to state-of-the-art techniques that use SMT solving.

cs.LG↗