TCSAlgBench: Benchmarking Automated Proving for Research-Level Theoretical Computer Science
Paper • 2609.35606 • Published
id stringlengths 16 71 | title stringlengths 19 78 | text stringlengths 1.44k 10.9k | paper_title stringlengths 20 121 | arxiv_id stringlengths 10 10 | arxiv_url stringlengths 32 32 | source_license stringclasses 2
values | source_license_url stringclasses 2
values | source_path stringlengths 34 147 |
|---|---|---|---|---|---|---|---|---|
2512.12960/corollary-separation | Strict separation of 3-query RLDCs and LDCs | # Research Challenge: Strict separation of 3-query RLDCs and LDCs
> **Source paper:** 3-Query RLDCs are Strictly Stronger than 3-Query LDCs
> **Reference:** [arXiv:2512.12960](https://arxiv.org/abs/2512.12960) — Corollary separation
> **Task:** Prove the theorem stated below. This document is self-contained: all d... | 3-Query RLDCs are Strictly Stronger than 3-Query LDCs | 2512.12960 | https://arxiv.org/abs/2512.12960 | CC0 1.0 | http://creativecommons.org/publicdomain/zero/1.0/ | 3-query-rldcs-are-strictly-stronger-than-3-query-ldcs/corollary-separation.md |
2512.12960/thm-RLDC-intro | Near-quadratic 3-query RLDC construction | # Research Challenge: Near-quadratic 3-query RLDC construction
> **Source paper:** 3-Query RLDCs are Strictly Stronger than 3-Query LDCs
> **Reference:** [arXiv:2512.12960](https://arxiv.org/abs/2512.12960) — Thm RLDC.intro
> **Task:** Prove the theorem stated below. This document is self-contained: all definition... | 3-Query RLDCs are Strictly Stronger than 3-Query LDCs | 2512.12960 | https://arxiv.org/abs/2512.12960 | CC0 1.0 | http://creativecommons.org/publicdomain/zero/1.0/ | 3-query-rldcs-are-strictly-stronger-than-3-query-ldcs/thm-RLDC-intro.md |
2512.12960/thm-ldc-lb | Cubic lower bound for 3-query LDCs | # Research Challenge: Cubic lower bound for 3-query LDCs
> **Source paper:** 3-Query RLDCs are Strictly Stronger than 3-Query LDCs
> **Reference:** [arXiv:2512.12960](https://arxiv.org/abs/2512.12960) — Thm ldc.lb
> **Task:** Prove the theorem stated below. This document is self-contained: all definitions and nota... | 3-Query RLDCs are Strictly Stronger than 3-Query LDCs | 2512.12960 | https://arxiv.org/abs/2512.12960 | CC0 1.0 | http://creativecommons.org/publicdomain/zero/1.0/ | 3-query-rldcs-are-strictly-stronger-than-3-query-ldcs/thm-ldc-lb.md |
2509.11399/theorem-hardness | Streaming hardness above the LP threshold | # Research Challenge: Streaming hardness above the LP threshold
> **Source paper:** A Dichotomy Theorem for Multi-Pass Streaming CSPs
> **Reference:** [arXiv:2509.11399](https://arxiv.org/abs/2509.11399) — Theorem hardness
> **Task:** Prove the theorem stated below. This document is self-contained: all definitions... | A Dichotomy Theorem for Multi-Pass Streaming CSPs | 2509.11399 | https://arxiv.org/abs/2509.11399 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | a-dichotomy-theorem-for-multi-pass-streaming-csps/theorem-hardness.md |
2509.11399/theorem-main | Dichotomy for multi-pass streaming CSPs | # Research Challenge: Dichotomy for multi-pass streaming CSPs
> **Source paper:** A Dichotomy Theorem for Multi-Pass Streaming CSPs
> **Reference:** [arXiv:2509.11399](https://arxiv.org/abs/2509.11399) — Theorem main
> **Task:** Prove the theorem stated below. This document is self-contained: all definitions and n... | A Dichotomy Theorem for Multi-Pass Streaming CSPs | 2509.11399 | https://arxiv.org/abs/2509.11399 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | a-dichotomy-theorem-for-multi-pass-streaming-csps/theorem-main.md |
2509.11399/theorem-streaming-solve-lp | Streaming algorithm approximating the LP value | # Research Challenge: Streaming algorithm approximating the LP value
> **Source paper:** A Dichotomy Theorem for Multi-Pass Streaming CSPs
> **Reference:** [arXiv:2509.11399](https://arxiv.org/abs/2509.11399) — Theorem streaming.solve.lp
> **Task:** Prove the theorem stated below. This document is self-contained: ... | A Dichotomy Theorem for Multi-Pass Streaming CSPs | 2509.11399 | https://arxiv.org/abs/2509.11399 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | a-dichotomy-theorem-for-multi-pass-streaming-csps/theorem-streaming-solve-lp.md |
2510.08542/rapid | Rapid mixing of the Gibbs sampler at high temperature | # Research Challenge: Rapid mixing of the Gibbs sampler at high temperature
> **Source paper:** A Dobrushin condition for quantum Markov chains: Rapid mixing and conditional mutual information at high temperature
> **Reference:** [arXiv:2510.08542](https://arxiv.org/abs/2510.08542) — Rapid
> **Task:** Prove the th... | A Dobrushin condition for quantum Markov chains: Rapid mixing and conditional mutual information at high temperature | 2510.08542 | https://arxiv.org/abs/2510.08542 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | a-dobrushin-condition-for-quantum-markov-chains-rapid-mixing-and-conditional-mutual-information-at-high-temperature/rapid.md |
2510.08542/thm-cmi-main | Decay of conditional mutual information for high-temperature Gibbs states | # Research Challenge: Decay of conditional mutual information for high-temperature Gibbs states
> **Source paper:** A Dobrushin condition for quantum Markov chains: Rapid mixing and conditional mutual information at high temperature
> **Reference:** [arXiv:2510.08542](https://arxiv.org/abs/2510.08542) — Thm cmi.main... | A Dobrushin condition for quantum Markov chains: Rapid mixing and conditional mutual information at high temperature | 2510.08542 | https://arxiv.org/abs/2510.08542 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | a-dobrushin-condition-for-quantum-markov-chains-rapid-mixing-and-conditional-mutual-information-at-high-temperature/thm-cmi-main.md |
2510.08542/thm-dobrushin-implies-mixing | Quantum Dobrushin condition implies rapid mixing | # Research Challenge: Quantum Dobrushin condition implies rapid mixing
> **Source paper:** A Dobrushin condition for quantum Markov chains: Rapid mixing and conditional mutual information at high temperature
> **Reference:** [arXiv:2510.08542](https://arxiv.org/abs/2510.08542) — Thm dobrushin.implies.mixing
> **Ta... | A Dobrushin condition for quantum Markov chains: Rapid mixing and conditional mutual information at high temperature | 2510.08542 | https://arxiv.org/abs/2510.08542 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | a-dobrushin-condition-for-quantum-markov-chains-rapid-mixing-and-conditional-mutual-information-at-high-temperature/thm-dobrushin-implies-mixing.md |
2605.00797/theorem-main | Deterministic fully dynamic maximal matching in n^{1/2+o(1)} time | # Research Challenge: Deterministic fully dynamic maximal matching in n^{1/2+o(1)} time
> **Source paper:** A Faster Deterministic Algorithm for Fully Dynamic Maximal Matching
> **Reference:** [arXiv:2605.00797](https://arxiv.org/abs/2605.00797) — Theorem main
> **Task:** Prove the theorem stated below. This docum... | A Faster Deterministic Algorithm for Fully Dynamic Maximal Matching | 2605.00797 | https://arxiv.org/abs/2605.00797 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | a-faster-deterministic-algorithm-for-fully-dynamic-maximal-matching/theorem-main.md |
2603.00770/thm-gaussian-general | Lower bound for sparse Gaussian mean detection | # Research Challenge: Lower bound for sparse Gaussian mean detection
> **Source paper:** A Unified Approach to Memory-Sample Tradeoffs for Detecting Planted Structures
> **Reference:** [arXiv:2603.00770](https://arxiv.org/abs/2603.00770) — Thm gaussian.general
> **Task:** Prove the theorem stated below. This docum... | A Unified Approach to Memory-Sample Tradeoffs for Detecting Planted Structures | 2603.00770 | https://arxiv.org/abs/2603.00770 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | a-unified-approach-to-memory-sample-tradeoffs-for-detecting-planted-structures/thm-gaussian-general.md |
2603.00770/thm-main | Lower bound for planted biclique streaming detection | # Research Challenge: Lower bound for planted biclique streaming detection
> **Source paper:** A Unified Approach to Memory-Sample Tradeoffs for Detecting Planted Structures
> **Reference:** [arXiv:2603.00770](https://arxiv.org/abs/2603.00770) — Thm main
> **Task:** Prove the theorem stated below. This document is... | A Unified Approach to Memory-Sample Tradeoffs for Detecting Planted Structures | 2603.00770 | https://arxiv.org/abs/2603.00770 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | a-unified-approach-to-memory-sample-tradeoffs-for-detecting-planted-structures/thm-main.md |
2603.00770/thm-micgeneral | General memory-sample lower bound framework | # Research Challenge: General memory-sample lower bound framework
> **Source paper:** A Unified Approach to Memory-Sample Tradeoffs for Detecting Planted Structures
> **Reference:** [arXiv:2603.00770](https://arxiv.org/abs/2603.00770) — Thm micgeneral
> **Task:** Prove the theorem stated below. This document is se... | A Unified Approach to Memory-Sample Tradeoffs for Detecting Planted Structures | 2603.00770 | https://arxiv.org/abs/2603.00770 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | a-unified-approach-to-memory-sample-tradeoffs-for-detecting-planted-structures/thm-micgeneral.md |
2603.00770/thm-pca-general | Lower bound for sparse PCA detection | # Research Challenge: Lower bound for sparse PCA detection
> **Source paper:** A Unified Approach to Memory-Sample Tradeoffs for Detecting Planted Structures
> **Reference:** [arXiv:2603.00770](https://arxiv.org/abs/2603.00770) — Thm pca.general
> **Task:** Prove the theorem stated below. This document is self-con... | A Unified Approach to Memory-Sample Tradeoffs for Detecting Planted Structures | 2603.00770 | https://arxiv.org/abs/2603.00770 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | a-unified-approach-to-memory-sample-tradeoffs-for-detecting-planted-structures/thm-pca-general.md |
2606.17260/theorem-1 | Linear convergence of HFopt-avg under quadratic growth | # Research Challenge: Linear convergence of HFopt-avg under quadratic growth
> **Source paper:** Accelerated Convex Optimization via Hamiltonian Dynamics with Deterministic Integration Time
> **Reference:** [arXiv:2606.17260](https://arxiv.org/abs/2606.17260) — Theorem 1
> **Task:** Prove the theorem stated below.... | Accelerated Convex Optimization via Hamiltonian Dynamics with Deterministic Integration Time | 2606.17260 | https://arxiv.org/abs/2606.17260 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | accelerated-convex-optimization-via-hamiltonian-dynamics-with-deterministic-integration-time/theorem-1.md |
2606.17260/theorem-2 | Accelerated convergence of HFopt-avg for convex f | # Research Challenge: Accelerated convergence of HFopt-avg for convex f
> **Source paper:** Accelerated Convex Optimization via Hamiltonian Dynamics with Deterministic Integration Time
> **Reference:** [arXiv:2606.17260](https://arxiv.org/abs/2606.17260) — Theorem 2
> **Task:** Prove the theorem stated below. This... | Accelerated Convex Optimization via Hamiltonian Dynamics with Deterministic Integration Time | 2606.17260 | https://arxiv.org/abs/2606.17260 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | accelerated-convex-optimization-via-hamiltonian-dynamics-with-deterministic-integration-time/theorem-2.md |
2606.17260/theorem-3 | Linear convergence of discretized dHFA under quadratic growth | # Research Challenge: Linear convergence of discretized dHFA under quadratic growth
> **Source paper:** Accelerated Convex Optimization via Hamiltonian Dynamics with Deterministic Integration Time
> **Reference:** [arXiv:2606.17260](https://arxiv.org/abs/2606.17260) — Theorem 3
> **Task:** Prove the theorem stated... | Accelerated Convex Optimization via Hamiltonian Dynamics with Deterministic Integration Time | 2606.17260 | https://arxiv.org/abs/2606.17260 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | accelerated-convex-optimization-via-hamiltonian-dynamics-with-deterministic-integration-time/theorem-3.md |
2606.17260/theorem-4 | Accelerated convergence of discretized dHFA for convex f | # Research Challenge: Accelerated convergence of discretized dHFA for convex f
> **Source paper:** Accelerated Convex Optimization via Hamiltonian Dynamics with Deterministic Integration Time
> **Reference:** [arXiv:2606.17260](https://arxiv.org/abs/2606.17260) — Theorem 4
> **Task:** Prove the theorem stated belo... | Accelerated Convex Optimization via Hamiltonian Dynamics with Deterministic Integration Time | 2606.17260 | https://arxiv.org/abs/2606.17260 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | accelerated-convex-optimization-via-hamiltonian-dynamics-with-deterministic-integration-time/theorem-4.md |
2509.20848/theorem-3 | Query complexity of learning decision stumps | # Research Challenge: Query complexity of learning decision stumps
> **Source paper:** Actively Learning Halfspaces without Synthetic Data
> **Reference:** [arXiv:2509.20848](https://arxiv.org/abs/2509.20848) — Theorem 3
> **Task:** Prove the theorem stated below. This document is self-contained: all definitions a... | Actively Learning Halfspaces without Synthetic Data | 2509.20848 | https://arxiv.org/abs/2509.20848 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | actively-learning-halfspaces-without-synthetic-data/theorem-3.md |
2509.20848/theorem-agnostic-formal | Tolerant agnostic learner for known-direction halfspaces | # Research Challenge: Tolerant agnostic learner for known-direction halfspaces
> **Source paper:** Actively Learning Halfspaces without Synthetic Data
> **Reference:** [arXiv:2509.20848](https://arxiv.org/abs/2509.20848) — Theorem agnostic.formal
> **Task:** Prove the theorem stated below. This document is self-co... | Actively Learning Halfspaces without Synthetic Data | 2509.20848 | https://arxiv.org/abs/2509.20848 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | actively-learning-halfspaces-without-synthetic-data/theorem-agnostic-formal.md |
2509.20848/theorem-d-direction-hs | Active learning halfspaces with known directions | # Research Challenge: Active learning halfspaces with known directions
> **Source paper:** Actively Learning Halfspaces without Synthetic Data
> **Reference:** [arXiv:2509.20848](https://arxiv.org/abs/2509.20848) — Theorem d.direction.hs
> **Task:** Prove the theorem stated below. This document is self-contained: ... | Actively Learning Halfspaces without Synthetic Data | 2509.20848 | https://arxiv.org/abs/2509.20848 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | actively-learning-halfspaces-without-synthetic-data/theorem-d-direction-hs.md |
2602.23448/lemma-subroutine-main | Fast augmenting-chain subroutine for degree reduction | # Research Challenge: Fast augmenting-chain subroutine for degree reduction
> **Source paper:** Additive One Approximation for Minimum Degree Spanning Tree: Breaking the $O(mn)$ Time Barrier
> **Reference:** [arXiv:2602.23448](https://arxiv.org/abs/2602.23448) — Lemma subroutine.main
> **Task:** Prove the theorem ... | Additive One Approximation for Minimum Degree Spanning Tree: Breaking the $O(mn)$ Time Barrier | 2602.23448 | https://arxiv.org/abs/2602.23448 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | additive-one-approximation-for-minimum-degree-spanning-tree-breaking-the-o-mn-time-barrier/lemma-subroutine-main.md |
2505.21892/theorem-4-1 | Almost-linear convergence of quantized transition diffusion | # Research Challenge: Almost-linear convergence of quantized transition diffusion
> **Source paper:** Almost Linear Convergence under Minimal Score Assumptions: Quantized Transition Diffusion
> **Reference:** [arXiv:2505.21892](https://arxiv.org/abs/2505.21892) — Theorem 4.1
> **Task:** Prove the theorem stated be... | Almost Linear Convergence under Minimal Score Assumptions: Quantized Transition Diffusion | 2505.21892 | https://arxiv.org/abs/2505.21892 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | almost-linear-convergence-under-minimal-score-assumptions-quantized-transition-diffusion/theorem-4-1.md |
2603.27795/theorem-1 | Optimal (1+ε)-approximation for stochastic vertex cover | # Research Challenge: Optimal (1+ε)-approximation for stochastic vertex cover
> **Source paper:** An Optimal Algorithm for Stochastic Vertex Cover
> **Reference:** [arXiv:2603.27795](https://arxiv.org/abs/2603.27795) — Theorem 1
> **Task:** Prove the theorem stated below. This document is self-contained: all defin... | An Optimal Algorithm for Stochastic Vertex Cover | 2603.27795 | https://arxiv.org/abs/2603.27795 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | an-optimal-algorithm-for-stochastic-vertex-cover/theorem-1.md |
2603.29702/theorem-lcs-main | Subquadratic (1-eps)-approximation for LCS | # Research Challenge: Subquadratic (1-eps)-approximation for LCS
> **Source paper:** Approximation Schemes for Edit Distance and LCS in Quasi-Strongly Subquadratic Time
> **Reference:** [arXiv:2603.29702](https://arxiv.org/abs/2603.29702) — Theorem lcs.main
> **Task:** Prove the theorem stated below. This document... | Approximation Schemes for Edit Distance and LCS in Quasi-Strongly Subquadratic Time | 2603.29702 | https://arxiv.org/abs/2603.29702 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | approximation-schemes-for-edit-distance-and-lcs-in-quasi-strongly-subquadratic-time/theorem-lcs-main.md |
2603.29702/theorem-main | Subquadratic (1+eps)-approximation for edit distance | # Research Challenge: Subquadratic (1+eps)-approximation for edit distance
> **Source paper:** Approximation Schemes for Edit Distance and LCS in Quasi-Strongly Subquadratic Time
> **Reference:** [arXiv:2603.29702](https://arxiv.org/abs/2603.29702) — Theorem main
> **Task:** Prove the theorem stated below. This do... | Approximation Schemes for Edit Distance and LCS in Quasi-Strongly Subquadratic Time | 2603.29702 | https://arxiv.org/abs/2603.29702 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | approximation-schemes-for-edit-distance-and-lcs-in-quasi-strongly-subquadratic-time/theorem-main.md |
2511.13659/thm-informal-hecke | Hecke equidistribution for rank-r module lattices | # Research Challenge: Hecke equidistribution for rank-r module lattices
> **Source paper:** Average hardness of SIVP for module lattices of fixed rank
> **Reference:** [arXiv:2511.13659](https://arxiv.org/abs/2511.13659) — Thm informal.hecke
> **Task:** Prove the theorem stated below. This document is self-contain... | Average hardness of SIVP for module lattices of fixed rank | 2511.13659 | https://arxiv.org/abs/2511.13659 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | average-hardness-of-sivp-for-module-lattices-of-fixed-rank/thm-informal-hecke.md |
2511.13659/thm-main | Average-case hardness of fixed-rank module SIVP | # Research Challenge: Average-case hardness of fixed-rank module SIVP
> **Source paper:** Average hardness of SIVP for module lattices of fixed rank
> **Reference:** [arXiv:2511.13659](https://arxiv.org/abs/2511.13659) — Thm main
> **Task:** Prove the theorem stated below. This document is self-contained: all defi... | Average hardness of SIVP for module lattices of fixed rank | 2511.13659 | https://arxiv.org/abs/2511.13659 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | average-hardness-of-sivp-for-module-lattices-of-fixed-rank/thm-main.md |
2509.20697/thm-Search_sqncp-qsdp_equivalence | Search QNCP and QSDP are equivalent | # Research Challenge: Search QNCP and QSDP are equivalent
> **Source paper:** Average-Case Complexity of Quantum Stabilizer Decoding
> **Reference:** [arXiv:2509.20697](https://arxiv.org/abs/2509.20697) — Thm Search_sqncp.qsdp_equivalence
> **Task:** Prove the theorem stated below. This document is self-contained:... | Average-Case Complexity of Quantum Stabilizer Decoding | 2509.20697 | https://arxiv.org/abs/2509.20697 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | average-case-complexity-of-quantum-stabilizer-decoding/thm-Search_sqncp-qsdp_equivalence.md |
2509.20697/thm-classical-representation-equivalent | Quantum LSN equivalent to classical LSN | # Research Challenge: Quantum LSN equivalent to classical LSN
> **Source paper:** Average-Case Complexity of Quantum Stabilizer Decoding
> **Reference:** [arXiv:2509.20697](https://arxiv.org/abs/2509.20697) — Thm classical.representation.equivalent
> **Task:** Prove the theorem stated below. This document is self-... | Average-Case Complexity of Quantum Stabilizer Decoding | 2509.20697 | https://arxiv.org/abs/2509.20697 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | average-case-complexity-of-quantum-stabilizer-decoding/thm-classical-representation-equivalent.md |
2509.20697/thm-decision-to-search-lsn | Search-to-decision reduction for LSN | # Research Challenge: Search-to-decision reduction for LSN
> **Source paper:** Average-Case Complexity of Quantum Stabilizer Decoding
> **Reference:** [arXiv:2509.20697](https://arxiv.org/abs/2509.20697) — Thm decision.to.search.lsn
> **Task:** Prove the theorem stated below. This document is self-contained: all d... | Average-Case Complexity of Quantum Stabilizer Decoding | 2509.20697 | https://arxiv.org/abs/2509.20697 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | average-case-complexity-of-quantum-stabilizer-decoding/thm-decision-to-search-lsn.md |
2507.15173/thm-alg-main | Recovering dense edges from Markov dynamics | # Research Challenge: Recovering dense edges from Markov dynamics
> **Source paper:** Better Models and Algorithms for Learning Ising Models from Dynamics
> **Reference:** [arXiv:2507.15173](https://arxiv.org/abs/2507.15173) — Thm alg.main
> **Task:** Prove the theorem stated below. This document is self-contained... | Better Models and Algorithms for Learning Ising Models from Dynamics | 2507.15173 | https://arxiv.org/abs/2507.15173 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | better-models-and-algorithms-for-learning-ising-models-from-dynamics/thm-alg-main.md |
2507.15173/thm-main-match | Recovering the residual matching of edges | # Research Challenge: Recovering the residual matching of edges
> **Source paper:** Better Models and Algorithms for Learning Ising Models from Dynamics
> **Reference:** [arXiv:2507.15173](https://arxiv.org/abs/2507.15173) — Thm main.match
> **Task:** Prove the theorem stated below. This document is self-contained... | Better Models and Algorithms for Learning Ising Models from Dynamics | 2507.15173 | https://arxiv.org/abs/2507.15173 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | better-models-and-algorithms-for-learning-ising-models-from-dynamics/thm-main-match.md |
2507.15173/thm-main-paramter | Estimating Ising coupling parameters from dynamics | # Research Challenge: Estimating Ising coupling parameters from dynamics
> **Source paper:** Better Models and Algorithms for Learning Ising Models from Dynamics
> **Reference:** [arXiv:2507.15173](https://arxiv.org/abs/2507.15173) — Thm main.paramter
> **Task:** Prove the theorem stated below. This document is se... | Better Models and Algorithms for Learning Ising Models from Dynamics | 2507.15173 | https://arxiv.org/abs/2507.15173 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | better-models-and-algorithms-for-learning-ising-models-from-dynamics/thm-main-paramter.md |
2510.21613/cor-easytouse | Explicit shadow path length bound | # Research Challenge: Explicit shadow path length bound
> **Source paper:** Beyond Smoothed Analysis: Analyzing the Simplex Method by the Book
> **Reference:** [arXiv:2510.21613](https://arxiv.org/abs/2510.21613) — Cor easytouse
> **Task:** Prove the theorem stated below. This document is self-contained: all defin... | Beyond Smoothed Analysis: Analyzing the Simplex Method by the Book | 2510.21613 | https://arxiv.org/abs/2510.21613 | CC BY 4.0 | http://creativecommons.org/licenses/by/4.0/ | beyond-smoothed-analysis-analyzing-the-simplex-method-by-the-book/cor-easytouse.md |
End of preview. Expand in Data Studio
TCSAlgBench
Research-level proof discovery in theoretical computer science.
TCSAlgBench evaluates whether language models can develop rigorous mathematical arguments for algorithm design and sample- or runtime-complexity guarantees. It covers upper bounds that require constructing and analyzing algorithms, as well as lower bounds that establish computational limitations.
Each challenge asks for a natural-language proof of a theorem, with the definitions, assumptions, notation, and problem setting needed to interpret it. The construction pipeline uses expert-designed rules to complete context and preserve quantitative guarantees. It withholds target proofs and, when finding an algorithm is part of the task, its construction. The pipeline supports fresh, versioned batches from newly released research.
Paper: TCSAlgBench: Benchmarking Automated Proving for Research-Level Theoretical Computer Science
Public dataset
This release contains 166 theorem-level challenges from 57 STOC and COLT 2026 papers, derived from arXiv versions released under CC BY 4.0 or CC0. Each challenge includes its source attribution and the context needed to work on the proof-discovery task.
Dataset contents
data/test.jsonl: one row per challenge, including the full Markdown text and source metadata.<paper-folder>/*.md: the original challenge documents, with source attribution and theorem context.<paper-folder>/paper.json: the paper title, arXiv identifier and URL, original license, and list of challenge files.index.json: an aggregate index of all 57 source papers.
Dataset viewer and loading
The test split contains 166 rows, one per challenge, from 57 papers.
It is explicitly configured to read data/test.jsonl.
Each row includes the complete original challenge Markdown in text, a unique
id, a challenge title, the paper_title, arxiv_id, arxiv_url,
source_license, source_license_url, and the original source_path.
The paper folders remain available in their original format.
from datasets import load_dataset
dataset = load_dataset("cyang98/TCSAlgBENCH", split="test")
print(len(dataset)) # 166
print(dataset[0]["text"])
Licensing
The dataset compilation and original contributions by the dataset authors are licensed under Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0), to the extent the authors hold the relevant rights. See LICENSE and the license summary.
The source-paper excerpts retain their original CC BY 4.0 or CC0 1.0
terms. The dataset-level license does not replace or restrict the rights granted
under those source licenses. Source license URLs and attribution are retained in
paper.json, index.json, and the challenge headers. The CC BY 4.0 text is also
provided in licenses/CC-BY-4.0.txt.
Source-paper licenses
Only papers whose arXiv submissions carry a permissive license are included:
- CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/) — 55 papers
- CC0 1.0 (https://creativecommons.org/publicdomain/zero/1.0/) — 2 papers
License information was retrieved from arXiv OAI-PMH metadata (https://oaipmh.arxiv.org/) on 2026-08-13. Papers under the arXiv nonexclusive-distribution license or any CC NC/ND/SA variant were excluded. All source papers have LaTeX source available on arXiv. Challenge files retain source attribution in their headers, with license details in the accompanying metadata.
Citation
If you use this dataset, please cite the paper and retain the source-paper attribution supplied with each challenge.
@misc{yang2026tcsalgbench,
title = {TCSAlgBench: Benchmarking Automated Proving for Research-Level Theoretical Computer Science},
author = {Chutong Yang and Xiyuan Zhang and Yu Huang and Boran Han and Soonho Kong and Shuai Zhang and Vihang Prakash Patil and Zhen Han and Michael Bohlke-Schneider and Bernie Wang},
year = {2026},
eprint = {2609.35606},
archivePrefix = {arXiv},
primaryClass = {cs.AI},
url = {https://arxiv.org/abs/2609.35606}
}
- Downloads last month
- 413