Large Language Models as Topological Thinkers: A Benchmark on Graph Persistent Homology

Hao Li, Hao Wan, Yixue Huang, Yuzhou Chen, Yulia Gel, Hao Jiang
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:70549-70568, 2026.

Abstract

Large language models (LLMs) are increasingly used in scientific discovery, system modeling, and decision-making, prompting interest in their ability to reason over complex structured data. Existing benchmarks primarily focus on static or local graph reasoning, overlooking the high-order structures in real-world systems whose global properties evolve across multiple scales. We introduce LLM4PH, a benchmark that evaluates multi-scale structural reasoning through the lens of persistent homology (PH), a topological framework for tracking structural evolution. LLM4PH decomposes the PH pipeline into interpretable reasoning tasks spanning synthetic and real-world graphs, revealing that most models struggle with reasoning over structural transitions and persistence. Beyond task-level evaluation, we perform cross-task ablations on prompt encoding and transfer, explore post-training effects, and construct a compositional PH pipeline to assess end-to-end performance. Our results provide the first in-depth view of how well LLMs bridge discrete graph structures with continuous topological abstraction, and offer insights into their potential for structure-aware scientific reasoning.

Cite this Paper


BibTeX
@InProceedings{pmlr-v306-li26gg, title = {Large Language Models as Topological Thinkers: A Benchmark on Graph Persistent Homology}, author = {Li, Hao and Wan, Hao and Huang, Yixue and Chen, Yuzhou and Gel, Yulia and Jiang, Hao}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {70549--70568}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/li26gg/li26gg.pdf}, url = {https://proceedings.mlr.press/v306/li26gg.html}, abstract = {Large language models (LLMs) are increasingly used in scientific discovery, system modeling, and decision-making, prompting interest in their ability to reason over complex structured data. Existing benchmarks primarily focus on static or local graph reasoning, overlooking the high-order structures in real-world systems whose global properties evolve across multiple scales. We introduce LLM4PH, a benchmark that evaluates multi-scale structural reasoning through the lens of persistent homology (PH), a topological framework for tracking structural evolution. LLM4PH decomposes the PH pipeline into interpretable reasoning tasks spanning synthetic and real-world graphs, revealing that most models struggle with reasoning over structural transitions and persistence. Beyond task-level evaluation, we perform cross-task ablations on prompt encoding and transfer, explore post-training effects, and construct a compositional PH pipeline to assess end-to-end performance. Our results provide the first in-depth view of how well LLMs bridge discrete graph structures with continuous topological abstraction, and offer insights into their potential for structure-aware scientific reasoning.} }
Endnote
%0 Conference Paper %T Large Language Models as Topological Thinkers: A Benchmark on Graph Persistent Homology %A Hao Li %A Hao Wan %A Yixue Huang %A Yuzhou Chen %A Yulia Gel %A Hao Jiang %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-li26gg %I PMLR %P 70549--70568 %U https://proceedings.mlr.press/v306/li26gg.html %V 306 %X Large language models (LLMs) are increasingly used in scientific discovery, system modeling, and decision-making, prompting interest in their ability to reason over complex structured data. Existing benchmarks primarily focus on static or local graph reasoning, overlooking the high-order structures in real-world systems whose global properties evolve across multiple scales. We introduce LLM4PH, a benchmark that evaluates multi-scale structural reasoning through the lens of persistent homology (PH), a topological framework for tracking structural evolution. LLM4PH decomposes the PH pipeline into interpretable reasoning tasks spanning synthetic and real-world graphs, revealing that most models struggle with reasoning over structural transitions and persistence. Beyond task-level evaluation, we perform cross-task ablations on prompt encoding and transfer, explore post-training effects, and construct a compositional PH pipeline to assess end-to-end performance. Our results provide the first in-depth view of how well LLMs bridge discrete graph structures with continuous topological abstraction, and offer insights into their potential for structure-aware scientific reasoning.
APA
Li, H., Wan, H., Huang, Y., Chen, Y., Gel, Y. & Jiang, H.. (2026). Large Language Models as Topological Thinkers: A Benchmark on Graph Persistent Homology. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:70549-70568 Available from https://proceedings.mlr.press/v306/li26gg.html.

Related Material