A Survey on LLM Symbolic Reasoning

The official GitHub page for the survey paper "A Survey on LLM Symbolic Reasoning". And this survey has been accepted by AAAI 2026 Bridge--Logical and Symbolic Reasoning in Language Models.

2. Theorem Proving (TP)

2.1 ATP (Automated Theorem Proving)

2.1.1 Direct

1. 2025_arXiv_Reinforced Large Language Model is A Formal Theorem Prover.

   [arXiv] [GitHub]

2. 2025_arXiv_Steering LLMs for Formal Theorem Proving.

   [arXiv]

3. 2024_Nature_AlphaGeometry_Solving Olympiad Geometry without Human Demonstrations.

   [Nature] [GitHub]

4. 2025_ICLR_LIPS_Proving Olympiad Inequalities by Synergizing LLMs and Symbolic Reasoning.

   [ICLR] [GitHub]

5. 2025_arXiv_HybridProver_HybridProver: Augmenting Theorem Proving with LLM-Driven Proof Synthesis and Refinement.

   [arXiv]

2.1.2 Decomposed

1. 2025_Nature_AlphaProof_Olympial-Level Formal Mathematical Reasoning with Reinforcement Learning.

   [Nature]

2. 2025_arXiv_APOLLO_APOLLO: Automated LLM and Lean Collaboration for Advanced Formal Reasoning.

   [arXiv] [GitHub]

3. 2025_EMNLP_DREAM_Towards Advanced Mathematical Reasoning for LLMs via First-Order Logic Theorem Proving.

   [EMNLP]

4. 2025_arXiv_DeepSeek-Porver-V2_DeepSeek-Porver-V2: Advancing Formal Mathematical Reasoning via Reinforcement Learning for Subgoal Decomposition.

   [arXiv]

5. 2026_ICML_WZ-LLM_Automated Formal Proofs of Combinatorial Identities via Wilf–Zeilberger Guidance and LLMs.

   [arXiv]

2.2 ITP (Interactive Theorem Proving)

1. 2024_arXiv_Lean Copilot_Lean Copilot: Large Language Models as Copilots for Theorem Proving in Lean.

   [arXiv]

2. 2024_EMNLP_BC-Prover_BC-Prover: Backward Chaining Prover for Formal Theorem Proving.

   [EMNLP]

3. Satisfiability Solving (SAT)

3.1 Logical Inference Verification

1. 2023_EMNLP_LINC_LINC: A Neurosymbolic Approach for Logical Reasoning by Combining Language Models with First-Order Logic Provers.

   [EMNLP] [arXiv] [GitHub]

2. 2023_EMNLP_Logic-LM_Logic-LM: Empowering Large Language Models with Symbolic Solvers for Faithful Logical Reasoning

   [EMNLP] [arXiv] [GitHub]

3. 2024_ICLR_DTV_Don't Trust: Verify-Grounding LLM Quantitative Reasoning with Autoformalization.

   [ICLR] [arXiv] [GitHub]

4. 2024_NeurIPS_ALT_Enhancing Reasoning Capabilities of LLMs via Principled Synthetic Logic Corpus.

   [NeruIPS] [arXiv]

5. 2025_ICLR_CLOVER_Divide and Translate: Compositional First-Order Lgoic Translation and Verification for Complex Logical Reasoning.

   [ICLR] [OpenReview] [arXiv]

6. 2025_ACL_Aristotle_Aristotle: Mastering Logical Reasoning with A LogicComplete Decompose-Search-Resolve Framework.

   [ACL]

7. 2026_ICLR_MAD-Logic_MAD-Logic: Multi-Agent Debate Enhances Symbolic Translation and Reasoning.

   [ICLR]

3.2 Compound Constraint Solving

1. 2023_NeurIPS_SATLM_SATLM: Satisfiability-Aided Language Models Using Declarative Prompting.

   [NeurIPS] [arXiv] [GitHub]

2. 2025_NeurlPS_HAR&CoPA_Bootstrapping Hierarchical Autoregressive Formal Reasoner with Chain-of-Proxy-Autoformalization.

   [NeruIPS] [GitHub]

3. 2025_arXiv_Loop-Invariant-Generation_Loop-Invariant-Generation: A Hybrid Fraemwork of Reasoning Optimised LLMs and SMT Solvers.

   [arXiv]

4. Consistency Checking

4.1 Internal Self-Consistency

1. 2021_EMNLP_BeliefBank_BeliefBank: Adding Memory to a Pre-Trained Language Model for a Systematic Notion of Belief.

   [EMNLP] [arXiv]

2. 2022_EMNLP_ConCoRD_Enhancing Self-Consistency and Performance of Pre-Trained Language Models through Natural Language Inference.

   [EMNLP]

3. 2022_EMNLP_Maieutic Prompting_Maieutic Prompting: Logically Consistent Reasoning with Recursive Explanations.

   [EMNLP] [arXiv]

4. 2025_ICML_REPAIR_Aligning with Logic: Measuring, Evaluating and Improving Logical Preference Consistency in Large Language Models.

   [ICML]

5. 2026_ICLR Workshop_LogicVault_LogicVault: Persistent Symbolic Belief States for Cross-Query Logical Consistency in LLMs.

   [ICLR Workshop]

4.2 External Knowledge Consistency

1. 2023_EMNLP_REFLEX_Language Models with Rationality.

   [EMNLP] [OpenReview] [arXiv]

2. 2025_ICLR_LoCo-LMs_Logically Consistent Language Models via Neuro-Symbolic Integration.

   [OpenReview] [arXiv] [GitHUb]

3. 2025_ICLR_LLMQuery_Logical Consistency of Large Language Models in Fact-Checking.

   [OpenReview] [arXiv]

5. Planning and Searching

5.1 Planning for Actions

5.1.1 Ungrounded

1. 2023_arXiv_LLM+P_LLM+P: Empowering Large Language Models with Optimal Planning Proficiency.

   [arXiv] [GitHub]

2. 2023_NeurIPS_LLM-DM_Leveraging Pre-trained Large Language Models to Construct and Utilize World Models for Model-based Task Planning.

   [NeurIPS] [GitHub]

3. 2025_AAAI_Planning in the Dark: LLM-Symbolic Planning Pipeline Without Experts.

   [AAAI] [GitHub]

4. 2025_NAACL_PSALM_Language Models Can Infer Action Semantics for Symbolic Planners from Environment Feedback.

   [NAACL]

5. 2026_arXiv_L-ICL_Localizing and Correcting Errors for LLM-based Planners.

   [arXiv]

5.1.2 Grounded

1. 2025_ICML workshop_SPG_Sketch-Plan-Generalize: Learning and Planning with Neuro-Symbolic Programmatic Representations for Inductive Spatial Concepts.

   [OpenReview] [arXiv]

2. 2025_NeurIPS_InstructFlow_InstructFlow: Adaptive Symbolic Constraint-Guided Code Generation for Long-Horizon Planning.

   [OpenReview]

3. 2026_ICLR_NL-PDDL_Natural Language PDDL (NL-PDDL) for Open-world Goal-oriented Commonsense Regression Planning in Embodied AI.

   [ICLR]

4. 2026_ICLR_VIRF_Grounding Generative Planners In Verifiable Logic=A Hybrid Architecture For Trustworthy Embodied AI.

   [ICLR] [arXiv]

5.2 (MCTS-based) Searching for Reasoning Path

1. 2024_arXiv_HiAR-ICL_Beyond Examples: High-level Automated Reasoning Paradigm in In-Context Learning via MCTS.

   [arXiv] [GitHub]

2. 2025_EMNLP_Symbolic ReAct_Enhancing Logical Reasoning in Language Models via Symbolically-Guided Monte Carlo Process Supervision.

   [EMNLP] [arXiv]

3. 2026_AAAI_SPIRAL_SPIRAL: Symbolic LLM Planning via Grounded and Reflective Search.

   [AAAI]

6. Normalized Tabular Reasoning

6.1 Question Answering (QA)

1. 2024_EMNLP_NormTab_NormTab: Improving Symbolic Reasoning in LLMs Through Tabular Data Normalization.

   [EMNLP] [arXiv] [GitHub]

2. 2024_NAACL_TabSQLify_TabSQLify: Enhancing Reasoning Capabilities of LLMs Through Table Decomposition. [NAACL] [arXiv] [GitHub]

3. 2025_ACL_RelationalCoder_RelationalCoder: Rethinking Complex Tables via Programmatic Relational Transformation.

   [ACL] [GitHub]

4. 2025_SIGIR_TabFormer_Reasoning and Retrieval for Complex Semi-structured Tables via Reinforced Relational Data Transformation.

   [SIGIR]

5. 2025_NAACL_H-STAR_H-STAR: LLM-driven Hybrid SQL-Text Adaptive Reasoning on Tables.

   [NAACL]

6. 2026_arXiv_ASTRA_ASTRA: Adaptive Semantic Tree Reasoning Architecture for Complex Table Question Answering. [arXiv]

6.2 Fact Verification

1. 2024_TACL_TabVer_TabVer: Tabular Fact Verification with Natural Logic.

   [TACL]

2. 2025_ICLR_TIDE_Triples as the Key: Structuring Makes Decomposition and Verification Easier in LLM-based TableQA.

   [OpenReview]

3. 2025_SIGIR_TabFormer_Reasoning and Retrieval for Complex Semi-structured Tables via Reinforced Relational Data Transformation.

   [SIGIR]

4. 2026_EACL_Analyzing LLM Instruction Optimization for Tabular Fact Verification.

   [EACL] [arXiv]

6.3 Temporal Reasoning

1. 2025_arXiv_LLM-Symbolic_LLM-Symbolic Integration for Robust Temporal Tabular Reasoning.

   [arXiv]

2. 2025_arXiv_Evidence-Guided-Schema-Normalization-for-Temporal-Tabular-Reasoning.

   [arXiv]

7. Real-World Applications

7.1 Medical

1. 2023_arXiv_Coupling Symbolic Reasoning with Language Modeling for Efficient Longitudinal Understanding of Unstructured Electronic Medical Records.

   [arXiv]

2. 2023_BIBM_Integrating Automated Knowledge Extraction with Large Language Models for Explainable Medical Decision-Making.

   [IEEE]

3. 2024_BIBM_ArgMed-Agents_ArgMed-Agents: Explainable Clinical Decision Reasoning with LLM Discussion via Argumentation Schemes.

   [IEEE] [arXiv]

4. 2025_arXiv_Perceptual-CoT_From Metaphor to Mechanism: How LLMs Decode Traditional Chinese Medicine Symbolic Language for Modern Clinical Relevance.

   [arXiv]

5. 2026_AAAI_Concept-RuleNet_Concept-RuleNet: Grounded Multi-Agent Neurosymbolic Reasoning in Visioni Language Models.

   [arXiv]

7.2 Law (Legal Reasoning)

1. 2025_AI and Law_An LLMs-based Neuro-Symbolic Legal Judgement Prediction Framework for Civil Cases.

   [Spring]

2. 2025_arXiv_Towards Robust Legal Reasoning: Harnessing Logical LLMs in Law.

   [arXiv]

3. 2025_CIKM_SOLAR_On Verifiable Legal Reasoning_On Verifiable Legal Reasoning: A Multi-Agent Framework with Formalized Knowledge Representations.

   [ACM CIKM] [arXiv]

7.3 LLM Safety (Attack and Defense)

1. 2024_NeurIPS Workshop_MathPrompt_Jailbreaking Large Language Models with Symbolic Mathematics.

   [NeurIPS]

2. 2025_arXiv_LogiBreak_Logic Jailbreak: Efficiently Unlocking LLM Safety Restrictions Through Formal Logical Expression.

   [arXiv]

3. 2025_ICLR_R2-Guard_R2-Guard: Robust Reasoning Enhanced LLM Guardrail via Knowledge-Enhanced Logical Reasoning.

   [OpenReview] [arXiv] [GitHub]

7.4 Hardware Design

1. 2025_NeurIPS_SymRTLO_SymRTLO: Enhancing RTL Code Optimization with LLMs and Neuron-Inspired Symbolic Reasoning.

   [OpenReview] [arXiv] [GitHub]

2. 2025_arXiv_AssertionForge_AssertionForge: Enhancing Formal Verification Assertion Generation with Structured Representation os Specifications and RTL.

   [arXiv]

3. 2025_arXiv_FLAG_FLAG: Formal and LLM-assisted SVA Generation for Formal Specifications of On-Chip Communication Protocols.

   [arXiv]

7.5 Program Analysis

1. 2025_ACM on Programming Langauges_AutoBug_Large Language Model Powered Symbolic Execution.

   [ACM] [arXiv]

2. 2025_EMNLP_ConstraintLLM_ConstraintLLM: A Neuro-Symbolic Framework for Industrial-Level Constraint Programming.

   [EMNLP] [GitHub]

3. 2025_arXiv_WARP_Worst-Case Symbolic Constraints Analysis and Generalisation with Large Language Models.

   [arXiv]

Dedicated Benchmark and Environment

Dedicated Benchmark

1. 2024_ACL_NeuBAROCO-dataset_Exploring Reasoning Biases in Large Language Models Through Syllogism= Insights from the NeuBAROCO Dataset.

   [ACL]

2. 2022_arXiv_FOLIO-dataset_FOLIO: Natural Language Reasoning with First-Order Logic.

   [EMNLP] [GitHub]

3. 2025_ICLR_LFC-dataset_Logical Consistencyh of Large Language Models in Fact-Checking.

   [ICLR]

4. 2025_ICLR_ProverGen (ProverQA-dataset)_Large Language Models Meet Symbolic Provers for Logical Reasoning Evaluation.

   [ICLR] [arXiv]

Dedicated Environment

1. 2023_NeurIPS_LeanDojo_LeanDojo: Theorem Proving with Retrieval-Augmented Language Models.

   [NeurIPS] [GitHub]

2. 2025_arXiv_Reasoning Core_Reasoning Core: A Scalable RL Environment for LLM Symbolic Reasoning.

   [arXiv] [GitHub]

Related Survey

1. 2023_ACL_Survey_Towards Reasoning in Large Language Models: A Survey.

   [ACL] [GitHub]

2. 2024_COLM_Survey_A Survey on Deep Learning for Theorem Proving

   [OpenReview] [arXiv]

3. 2024_arXiv_Survey_Reasoning with Large Language Models: A Survey.

   [arXiv]

4. 2025_IJCAI_Survey_Empowering LLMs with Logical Reasoning: A Comprehensive Survey.

   [IJCAI] [arXiv]

5. 2025_IJCAI_Survey_Neuro-Symbolic Artificial Intelligence: Towards Improving the Reasoning Abilities of Large Language Models.

   [IJCAI] [arXiv] [GitHub]

6. 2025_arXiv_Survey_Logical Reasoning in Large Language Models: A Survey

   [arXiv]

7. 2025_arXiv_Survey_LLM Inference Enhanced by External Knowledge: A Survey.

   [arXiv] [GitHub]

8. 2025_OpenReview_Survey_A Survey on Enhancing Large Language Models with Symbolic Reasoning.

   [OpenReview]

9. 2025_CSUR_Survey_A Survey of Reasoning with Foundation Models.

   [ACM Computing Surveys] [arXiv]

📖 Citation

If you compare with, build on, or use aspects of this work, please cite the following:

@inproceedings{li2026survey,
  title={A Survey on LLM Symbolic Reasoning},
  author={Li, Jindong and Fu, Yali and Yang, Yang and Liu, Jiahong and Zhang, Hongce and Li, Haoxuan and Yue, Yutao and Yang, Menglin},
  booktitle={Logical and Symbolic Reasoning in Language Models@ AAAI 2026},
  year={2026}
}