arXiv subjects

Vijay Ganesh

Publications and source records attributed to Vijay Ganesh.

At least 37 records · Page 2Linked to original sources

arXiv2026

AlphaMapleSAT: An MCTS-based Cube-and-Conquer SAT Solver for Hard Combinatorial Problems

This paper introduces AlphaMapleSAT, a Cube-and-Conquer (CnC) parallel SAT solver that integrates Monte Carlo Tree Search (MCTS) with deductive feedback to efficiently solve challenging combinatorial SAT problems. Traditional lookahead cubing methods, used by solvers such as March, limit their search depth to reduce overhead often resulting in suboptimal partitions. By contrast, AlphaMapleSAT performs a deeper MCTS search guided by deductive rewards from SAT solvers. This approach enables informed exploration of the cubing space while keeping cubing costs low. We demonstrate the efficacy of our technique via extensive evaluations against the widely used and established March cubing solver on three well-known challenging combinatorial benchmarks, including the minimum Kochen-Specker (KS) problem from quantum mechanics, the Murty-Simon Conjecture, and the Ramsey problems from extremal graph theory. We compare AlphaMapleSAT against March using different types of conquering solvers such as SAT Modulo Symmetries (SMS) and SAT+CAS, both built on top of the CaDiCaL SAT solver. We show that in all cases, there is a speedup in elapsed real time (wall clock time) ranging from 1.61x to 7.57x on a 128 core machine for the above-mentioned problems. We also perform cube-level and parallel scaling analysis over 32, 64, and 128 cores, which shows that AlphaMapleSAT outperforms March on all these settings. Our results show that deductively-guided MCTS search technique for cubing in CnC solvers can significantly outperform March on hard combinatorial problems.

cs.AI↗

arXiv2025

FrontierCS: Evolving Challenges for Evolving Intelligence

We introduce FrontierCS, a benchmark of 156 open-ended problems across diverse areas of computer science, designed and reviewed by experts, including CS PhDs and top-tier competitive programming participants and problem setters. Unlike existing benchmarks that focus on tasks with known optimal solutions, FrontierCS targets problems where the optimal solution is unknown, but the quality of a solution can be objectively evaluated. Models solve these tasks by implementing executable programs rather than outputting a direct answer. FrontierCS includes algorithmic problems, which are often NP-hard variants of competitive programming problems with objective partial scoring, and research problems with the same property. For each problem we provide an expert reference solution and an automatic evaluator. Combining open-ended design, measurable progress, and expert curation, FrontierCS provides a benchmark at the frontier of computer-science difficulty. Empirically, we find that frontier reasoning models still lag far behind human experts on both the algorithmic and research tracks, that increasing reasoning budgets alone does not close this gap, and that models often over-optimize for generating merely workable code instead of discovering high-quality algorithms and system designs.

cs.LG↗

arXiv2025

O-Forge: An LLM + Computer Algebra Framework for Asymptotic Analysis

Large language models have recently demonstrated advanced capabilities in solving IMO and Putnam problems; yet their role in research mathematics has remained fairly limited. The key difficulty is verification: suggested proofs may look plausible, but cannot be trusted without rigorous checking. We present a framework, called LLM+CAS, and an associated tool, O-Forge, that couples frontier LLMs with a computer algebra systems (CAS) in an In-Context Symbolic Feedback loop to produce proofs that are both creative and symbolically verified. Our focus is on asymptotic inequalities, a topic that often involves difficult proofs and appropriate decomposition of the domain into the "right" subdomains. Many mathematicians, including Terry Tao, have suggested that using AI tools to find the right decompositions can be very useful for research-level asymptotic analysis. In this paper, we show that our framework LLM+CAS turns out to be remarkably effective at proposing such decompositions via a combination of a frontier LLM and a CAS. More precisely, we use an LLM to suggest domain decomposition, and a CAS (such as Mathematica) that provides a verification of each piece axiomatically. Using this loop, we answer a question posed by Terence Tao: whether LLMs coupled with a verifier can be used to help prove intricate asymptotic inequalities. More broadly, we show how AI can move beyond contest math towards research-level tools for professional mathematicians.

cs.AI↗

arXiv2025

Robustness of deep learning classification to adversarial input on GPUs: asynchronous parallel accumulation is a source of vulnerability

The ability of machine learning (ML) classification models to resist small, targeted input perturbations -- known as adversarial attacks -- is a key measure of their safety and reliability. We show that floating-point non-associativity (FPNA) coupled with asynchronous parallel programming on GPUs is sufficient to result in misclassification, without any perturbation to the input. Additionally, we show that standard adversarial robustness results may be overestimated up to 4.6 when not considering machine-level details. We develop a novel black-box attack using Bayesian optimization to discover external workloads that can change the instruction scheduling which bias the output of reductions on GPUs and reliably lead to misclassification. Motivated by these results, we present a new learnable permutation (LP) gradient-based approach to learning floating-point operation orderings that lead to misclassifications. The LP approach provides a worst-case estimate in a computationally efficient manner, avoiding the need to run identical experiments tens of thousands of times over a potentially large set of possible GPU states or architectures. Finally, using instrumentation-based testing, we investigate parallel reduction ordering across different GPU architectures under external background workloads, when utilizing multi-GPU virtualization, and when applying power capping. Our results demonstrate that parallel reduction ordering varies significantly across architectures under the first two conditions, substantially increasing the search space required to fully test the effects of this parallel scheduler-based vulnerability. These results and the methods developed here can help to include machine-level considerations into adversarial robustness assessments, which can make a difference in safety and mission critical applications.

cs.LG↗

arXiv2025

Verified Certificates via SAT and Computer Algebra Systems for the Ramsey $R(3, 8)$ and $R(3, 9)$ Problems

The Ramsey problem $R(3, k)$ seeks to determine the smallest value of $n$ such that any red/blue edge coloring of the complete graph on $n$ vertices must either contain a blue triangle (3-clique) or a red clique of size $k$. Despite its significance, many previous computational results for the Ramsey $R(3, k)$ problem such as $R(3, 8)$ and $R(3, 9)$ lack formal verification. To address this issue, we use the software MathCheck to generate certificates for Ramsey problems $R(3, 8)$ and $R(3, 9)$ (and symmetrically $R(8, 3)$ and $R(9, 3)$) by integrating a Boolean satisfiability (SAT) solver with a computer algebra system (CAS). Our SAT+CAS approach significantly outperforms traditional SAT-only methods, demonstrating an improvement of several orders of magnitude in runtime. For instance, our SAT+CAS approach solves $R(3, 8)$ (resp., $R(8, 3)$) sequentially in 59 hours (resp., in 11 hours), while a SAT-only approach using state-of-the-art CaDiCaL solver times out after 7 days. Additionally, in order to be able to scale to harder Ramsey problems $R(3, 9)$ and $R(9, 3)$ we further optimized our SAT+CAS tool using a parallelized cube-and-conquer approach. Our results provide the first independently verifiable certificates for these Ramsey numbers, ensuring both correctness and completeness of the exhaustive search process of our SAT+CAS tool.

cs.LO↗

arXiv2025

RLSF: Fine-tuning LLMs via Symbolic Feedback

Large Language Models (LLMs) have transformed AI but often struggle with tasks that require domain-specific reasoning and logical alignment. Traditional fine-tuning methods do not leverage the vast amount of symbolic domain-knowledge available to us via symbolic reasoning tools (e.g., provers), and are further limited by sparse rewards and unreliable reward models. We introduce Reinforcement Learning via Symbolic Feedback (RLSF), a novel fine-tuning paradigm where symbolic reasoning tools (e.g., solvers, provers, and algebra systems) provide fine-grained feedback to LLMs. RLSF uses poly-sized certificates (e.g., proofs) generated by symbolic tools to identify and correct errors in model outputs, offering token-level guidance without requiring differentiable reasoning systems. This paradigm bridges the gap between symbolic reasoning and LLM fine-tuning, enabling precise alignment with domain-specific constraints while addressing key limitations of traditional reward signals. Via extensive evaluations, we show that our RLSF-based fine-tuning of LLMs outperforms traditional approaches on five different applications (that have some associated logical or domain constraints), namely, program synthesis from natural language pseudo-code to programming language, three chemistry tasks, and solving the Game of 24. A key takeaway is that fine-tuning via RLSF enables relatively smaller LLMs to significantly outperform closed-source models that are orders of magnitude larger.

cs.CL↗

arXiv2025

Abstractions-of-Thought: Intermediate Representations for LLM Reasoning in Hardware Design

Large language models (LLMs) have achieved impressive proficiency on logic and programming tasks, often rivaling expert-level performance. However, generating functionally correct hardware description language (HDL) code from natural language specifications remains challenging, primarily in data-scarce domains. Therefore, we present Abstractions-of-Thought (AoT) - a training-free, inference-only prompting framework to mitigate misinterpretations and reasoning pitfalls of LLMs through a series of task-based abstractions within the prompting procedure, assisting in the transition from high-level to low-level representations of hardware. Furthermore, AoT consists of the following stages: (1) an LLM-based classification of hardware design patterns, (2) a structured intermediate representation (IR) to separate functional decomposition from code syntax, and (3) a line-by-line pseudocode solution enabling a more direct mapping to the final Verilog implementation. Experimental results on the VerilogEval benchmark depict that AoT demonstrates improvements in functionality when applied to large non-reasoning models (such as GPT-4o, outperforming all baseline techniques (including 1-shot, Chain-of-Thought, and Tree-of-Thought) while significantly reducing the generated tokens by 1.8-5.2x compared to popular Tree-of-Thought prompting.

cs.PL↗

arXiv2025

Can Transformers Reason Logically? A Study in SAT Solving

We formally study the logical reasoning capabilities of decoder-only Transformers in the context of the boolean satisfiability (SAT) problem. First, we prove by construction that decoder-only Transformers can decide 3-SAT, in a non-uniform model of computation, using backtracking and deduction via Chain-of-Thought (CoT). %We prove its correctness by showing trace equivalence to the well-known DPLL SAT-solving algorithm. Second, we implement our construction as a PyTorch model with a tool (PARAT) that we designed to empirically demonstrate its correctness and investigate its properties. Third, rather than \textit{programming} a transformer to reason, we evaluate empirically whether it can be \textit{trained} to do so by learning directly from algorithmic traces (``reasoning paths'') from our theoretical construction. The trained models demonstrate strong out-of-distribution generalization on problem sizes seen during training but has limited length generalization, which is consistent with the implications of our theoretical result

cs.LG↗

arXiv2024

CoTran: An LLM-based Code Translator using Reinforcement Learning with Feedback from Compiler and Symbolic Execution

In this paper, we present an LLM-based code translation method and an associated tool called CoTran, that translates whole-programs from one high-level programming language to another. Existing LLM-based code translation methods lack training to ensure that the translated code reliably compiles or bears substantial functional equivalence to the input code. In our work, we fine-tune an LLM using reinforcement learning, incorporating compiler feedback, and symbolic execution (symexec)-based testing feedback to assess functional equivalence between the input and output programs. The idea is to guide an LLM during fine-tuning, via compiler and symexec-based testing feedback, by letting it know how far it is from producing perfect translations. We conduct extensive experiments comparing CoTran with 14 other code translation tools, including human-written transpilers, LLM-based translation tools, and ChatGPT. Using a benchmark of over \num{57000} code pairs in Java and Python, we demonstrate that CoTran outperforms the other tools on relevant metrics such as compilation accuracy (CompAcc) and functional equivalence accuracy (FEqAcc). For example, in Python-to-Java translation, CoTran achieves 48.68% FEqAcc and 76.98% CompAcc, whereas the nearest competing tool (PLBART-base) gets 38.26% and 75.77% respectively. Additionally, CoTran, built on top of CodeT5, improves FEqAcc by +14.89% and CompAcc by +8.14% for Python-to-Java (resp., +12.94% and +4.30% for Java-to-Python).

cs.PL↗

arXiv2024

AdChain: Decentralized Header Bidding

Due to the involvement of multiple intermediaries without trusted parties, lack of proper regulations, and a complicated supply chain, ad impression discrepancy affects online advertising. This issue causes up to $82 billion annual revenue loss for honest parties. The loss can be significantly reduced with a precise and trusted decentralized mechanism. This paper presents AdChain, a decentralized, distributed, and verifiable solution that detects and minimizes online advertisement impression discrepancies. AdChain establishes trust by employing multiple independent agents to receive and record log-level data, along with a consensus protocol to validate each ad data. AdChain is scalable, efficient, and compatible with the current infrastructure. Our experimental evaluation, using over half a million ad data points, identifies system parameters that achieve 98% accuracy, reducing the ad discrepancy rate from 20% to 2%. Our cost analysis shows that active nodes on AdChain can generate profits comparable to miners on major blockchain networks like Bitcoin.

cs.CR↗

arXiv2024

CountChain: A Decentralized Oracle Network for Counting Systems

Blockchain integration in industries like online advertising is hindered by its connectivity limitations to off-chain data. These industries heavily rely on precise counting systems for collecting and analyzing off-chain data. This requires mechanisms, often called oracles, to feed off-chain data into smart contracts. However, current oracle solutions are ill-suited for counting systems since the oracles do not know when to expect the data, posing a significant challenge. To address this, we present CountChain, a decentralized oracle network for counting systems. In CountChain, data is received by all oracle nodes, and any node can submit a proposition request. Each proposition contains enough data to evaluate the occurrence of an event. Only randomly selected nodes participate in a game to evaluate the truthfulness of each proposition by providing proof and some stake. Finally, the propositions with the outcome of True increment the counter in a smart contract. Thus, instead of a contract calling oracles for data, in CountChain, the oracles call a smart contract when the data is available. Furthermore, we present a formal analysis and experimental evaluation of the system's parameters on over half a million data points to obtain optimal system parameters. In such conditions, our game-theoretical analysis demonstrates that a Nash equilibrium exists wherein all rational parties participate with honesty.

cs.CR↗

arXiv2024

Layered and Staged Monte Carlo Tree Search for SMT Strategy Synthesis

Modern SMT solvers, such as Z3, offer user-controllable strategies, enabling users to tailor solving strategies for their unique set of instances, thus dramatically enhancing solver performance for their use case. However, this approach of strategy customization presents a significant challenge: handcrafting an optimized strategy for a class of SMT instances remains a complex and demanding task for both solver developers and users alike. In this paper, we address this problem of automatic SMT strategy synthesis via a novel Monte Carlo Tree Search (MCTS) based method. Our method treats strategy synthesis as a sequential decision-making process, whose search tree corresponds to the strategy space, and employs MCTS to navigate this vast search space. The key innovations that enable our method to identify effective strategies, while keeping costs low, are the ideas of layered and staged MCTS search. These novel heuristics allow for a deeper and more efficient exploration of the strategy space, enabling us to synthesize more effective strategies than the default ones in state-of-the-art (SOTA) SMT solvers. We implement our method, dubbed Z3alpha, as part of the Z3 SMT solver. Through extensive evaluations across six important SMT logics, Z3alpha demonstrates superior performance compared to the SOTA synthesis tool FastSMT, the default Z3 solver, and the CVC5 solver on most benchmarks. Remarkably, on a challenging QF_BV benchmark set, Z3alpha solves 42.7% more instances than the default strategy in the Z3 SMT solver.

cs.AI↗

arXiv2024

A Reinforcement Learning based Reset Policy for CDCL SAT Solvers

Restart policy is an important technique used in modern Conflict-Driven Clause Learning (CDCL) solvers, wherein some parts of the solver state are erased at certain intervals during the run of the solver. In most solvers, variable activities are preserved across restart boundaries, resulting in solvers continuing to search parts of the assignment tree that are not far from the one immediately prior to a restart. To enable the solver to search possibly "distant" parts of the assignment tree, we study the effect of resets, a variant of restarts which not only erases the assignment trail, but also randomizes the activity scores of the variables of the input formula after reset, thus potentially enabling a better global exploration of the search space. In this paper, we model the problem of whether to trigger reset as a multi-armed bandit (MAB) problem, and propose two reinforcement learning (RL) based adaptive reset policies using the Upper Confidence Bound (UCB) and Thompson sampling algorithms. These two algorithms balance the exploration-exploitation tradeoff by adaptively choosing arms (reset vs. no reset) based on their estimated rewards during the solver's run. We implement our reset policies in four baseline SOTA CDCL solvers and compare the baselines against the reset versions on Satcoin benchmarks and SAT Competition instances. Our results show that RL-based reset versions outperform the corresponding baseline solvers on both Satcoin and the SAT competition instances, suggesting that our RL policy helps to dynamically and profitably adapt the reset frequency for any given input instance. We also introduce the concept of a partial reset, where at least a constant number of variable activities are retained across reset boundaries. Building on previous results, we show that there is an exponential separation between O(1) vs. $Ω(n)$-length partial resets.

cs.LO↗

arXiv2024

A SAT Solver and Computer Algebra Attack on the Minimum Kochen-Specker Problem

One of the fundamental results in quantum foundations is the Kochen-Specker (KS) theorem, which states that any theory whose predictions agree with quantum mechanics must be contextual, i.e., a quantum observation cannot be understood as revealing a pre-existing value. The theorem hinges on the existence of a mathematical object called a KS vector system. While many KS vector systems are known, the problem of finding the minimum KS vector system in three dimensions (3D) has remained stubbornly open for over 55 years. To address the minimum KS problem, we present a new verifiable proof-producing method based on a combination of a Boolean satisfiability (SAT) solver and a computer algebra system (CAS) that uses an isomorph-free orderly generation technique that is very effective in pruning away large parts of the search space. Our method shows that a KS system in 3D must contain at least 24 vectors. We show that our sequential and parallel Cube-and-Conquer (CnC) SAT+CAS methods are significantly faster than SAT-only, CAS-only, and a prior CAS-based method of Uijlen and Westerbaan. Further, while our parallel pipeline is somewhat slower than the parallel CnC version of the recently introduced Satisfiability Modulo Theories (SMS) method, this is in part due to the overhead of proof generation. Finally, we provide the first computer-verifiable proof certificate of a lower bound to the KS problem with a size of 40.3 TiB in order 23.

quant-ph↗

arXiv2024

BertRLFuzzer: A BERT and Reinforcement Learning Based Fuzzer

We present a novel tool BertRLFuzzer, a BERT and Reinforcement Learning (RL) based fuzzer aimed at finding security vulnerabilities for Web applications. BertRLFuzzer works as follows: given a set of seed inputs, the fuzzer performs grammar-adhering and attack-provoking mutation operations on them to generate candidate attack vectors. The key insight of BertRLFuzzer is the use of RL with a BERT model as an agent to guide the fuzzer to efficiently learn grammar-adhering and attack-provoking mutation operators. In order to establish the efficacy of BertRLFuzzer we compare it against a total of 13 black box and white box fuzzers over a benchmark of 9 victim websites with over 16K LOC. We observed a significant improvement relative to the nearest competing tool in terms of time to first attack (54% less), new vulnerabilities found (17 new vulnerabilities), and attack rate (4.4% more attack vectors generated).

cs.SE↗

arXiv2023

Limits of CDCL Learning via Merge Resolution

In their seminal work, Atserias et al. and independently Pipatsrisawat and Darwiche in 2009 showed that CDCL solvers can simulate resolution proofs with polynomial overhead. However, previous work does not address the tightness of the simulation, i.e., the question of how large this overhead needs to be. In this paper, we address this question by focusing on an important property of proofs generated by CDCL solvers that employ standard learning schemes, namely that the derivation of a learned clause has at least one inference where a literal appears in both premises (aka, a merge literal). Specifically, we show that proofs of this kind can simulate resolution proofs with at most a linear overhead, but there also exist formulas where such overhead is necessary or, more precisely, that there exist formulas with resolution proofs of linear length that require quadratic CDCL proofs.

cs.CC↗

arXiv2023

CGDTest: A Constrained Gradient Descent Algorithm for Testing Neural Networks

In this paper, we propose a new Deep Neural Network (DNN) testing algorithm called the Constrained Gradient Descent (CGD) method, and an implementation we call CGDTest aimed at exposing security and robustness issues such as adversarial robustness and bias in DNNs. Our CGD algorithm is a gradient-descent (GD) method, with the twist that the user can also specify logical properties that characterize the kinds of inputs that the user may want. This functionality sets CGDTest apart from other similar DNN testing tools since it allows users to specify logical constraints to test DNNs not only for $\ell_p$ ball-based adversarial robustness but, more importantly, includes richer properties such as disguised and flow adversarial constraints, as well as adversarial robustness in the NLP domain. We showcase the utility and power of CGDTest via extensive experimentation in the context of vision and NLP domains, comparing against 32 state-of-the-art methods over these diverse domains. Our results indicate that CGDTest outperforms state-of-the-art testing tools for $\ell_p$ ball-based adversarial robustness, and is significantly superior in testing for other adversarial robustness, with improvements in PAR2 scores of over 1500% in some cases over the next best tool. Our evaluation shows that our CGD method outperforms competing methods we compared against in terms of expressibility (i.e., a rich constraint language and concomitant tool support to express a wide variety of properties), scalability (i.e., can be applied to very large real-world models with up to 138 million parameters), and generality (i.e., can be used to test a plethora of model architectures).

cs.LG↗

arXiv2023

Learning Modulo Theories

Recent techniques that integrate \emph{solver layers} into Deep Neural Networks (DNNs) have shown promise in bridging a long-standing gap between inductive learning and symbolic reasoning techniques. In this paper we present a set of techniques for integrating \emph{Satisfiability Modulo Theories} (SMT) solvers into the forward and backward passes of a deep network layer, called SMTLayer. Using this approach, one can encode rich domain knowledge into the network in the form of mathematical formulas. In the forward pass, the solver uses symbols produced by prior layers, along with these formulas, to construct inferences; in the backward pass, the solver informs updates to the network, driving it towards representations that are compatible with the solver's theory. Notably, the solver need not be differentiable. We implement \layername as a Pytorch module, and our empirical results show that it leads to models that \emph{1)} require fewer training samples than conventional models, \emph{2)} that are robust to certain types of covariate shift, and \emph{3)} that ultimately learn representations that are consistent with symbolic knowledge, and thus naturally interpretable.

cs.LG↗