Result 133, Theoretical computer science

The computational complexity of Weisfeiler–Leman refinement

Proves unconditional nΩ(k)n^{\Omega(k)} deterministic time lower bounds for joint and separate k-dimensional Weisfeiler–Leman equivalence, for sufficiently large fixed k in the specified sequential adjacency-matrix models. With dimension as input, joint equivalence is EXPTIME-complete even on subcubic graphs; deciding whether refinement identifies a graph is also EXPTIME-complete.

Lean formalization Algorithm or complexity result

The bigger picture

Why it matters

Graph refinement tries to tell networks apart by repeatedly updating structural labels. The manuscripts report limits on how fast this test can be decided, even for restricted graphs, clarifying the computational cost of greater distinguishing power.

What changes?

Weisfeiler-Leman refinement labels ordered groups of k vertices using their surrounding structure; equivalence means two graphs remain indistinguishable by these labels. For every sufficiently large fixed k, the manuscript reports an unconditional worst-case deterministic time lower bound of n raised to a constant times k, at every sufficiently large vertex count n. It covers joint and separate-coordinate updates, even on simple connected uncolored graphs of diameter at most two. Inputs are explicit adjacency matrices; the bound concerns specified sequential models.

What does that help mathematicians do?

With binary-encoded input dimension k at least two, joint equivalence is reported EXPTIME-complete, even on connected simple uncolored graphs of equal positive order and maximum degree three. Identification, meaning distinction from every nonisomorphic comparison graph, is also reported EXPTIME-complete for nonempty finite simple uncolored graphs and positive binary-encoded dimension. Neither sparse connectivity nor asking about one graph's uniqueness therefore avoids the full difficulty of deterministic exponential-time computation.

Are there practical applications?

The immediate value is foundational: these bounds constrain attempts to speed up graph comparison by refinement, rather than providing a faster algorithm. The unconditional bounds apply to multitape Turing machines and sequential logarithmic-word RAMs with fixed polynomial-bit-time instructions. They rule out polynomial runtimes whose exponent grows sublinearly with dimension in those models, not every possible computational approach.

This section was generated by GPT-6 Astra Medium. This explanation is based on the result summary and manuscript abstracts below. This context is separate from OpenAI's source text.

4 manuscripts

Parity lifts and bounded-treewidth witnesses for Weisfeiler–Leman equivalence

September 25, 2026 16 pages

For k ≥ 4, we construct two uncolored graphs that are k-dimensional Weisfeiler–Leman equivalent exactly when a prescribed finite-domain choice system has no compatible choice. The system has k+1k+1 domains for joint refinement and k for separate-coordinate refinement. A successful choice is detected after two joint rounds or one separate round. Applied to sparse satisfiability, the reduction gives fixed-dimension nΩ(k)n^{\Omega(k)} time exclusions under positive-rate ETH, even for deciding equality of these early histograms.

Cite (BibTeX)
@misc{OAI:Parity-lifts-and-bounded-treewidth-witnesses-for-Weisfeiler-Leman-equivalence-September-25-2026,
  author = {{OpenAI}},
  title = {{Parity lifts and bounded-treewidth witnesses for Weisfeiler--Leman equivalence}},
  howpublished = {OpenAI Math Release preprint
                  \href{https://github.com/openai/math/blob/main/preprints/Parity-lifts-and-bounded-treewidth-witnesses-for-Weisfeiler-Leman-equivalence-September-25-2026/paper.pdf}{OAI:Parity-lifts-and-bounded-treewidth-witnesses-for-Weisfeiler-Leman-equivalence-September-25-2026}},
  year = {2026}
}

The complexity of identifying a graph by Weisfeiler–Leman refinement

September 25, 2026 25 pages

We prove that deciding whether Weisfeiler–Leman refinement of an input dimension identifies a given graph is EXPTIME-complete. The input is a nonempty finite simple uncolored graph in adjacency-matrix form and a positive binary-encoded dimension. Identification quantifies over every comparison graph.

Cite (BibTeX)
@misc{OAI:The-complexity-of-identifying-a-graph-by-Weisfeiler-Leman-refinement-September-25-2026,
  author = {{OpenAI}},
  title = {{The complexity of identifying a graph by Weisfeiler--Leman refinement}},
  howpublished = {OpenAI Math Release preprint
                  \href{https://github.com/openai/math/blob/main/preprints/The-complexity-of-identifying-a-graph-by-Weisfeiler-Leman-refinement-September-25-2026/paper.pdf}{OAI:The-complexity-of-identifying-a-graph-by-Weisfeiler-Leman-refinement-September-25-2026}},
  year = {2026}
}

Unconditional time lower bounds for Weisfeiler–Leman equivalence

September 25, 2026 27 pages

For every sufficiently large fixed k, deciding whether two n-vertex graphs are k-Weisfeiler–Leman equivalent requires nΩ(k)n^{\Omega(k)} deterministic sequential time in the worst case. The bound holds at every sufficiently large graph order, even for simple connected uncolored graphs of diameter at most two, without a complexity assumption. Inputs are explicit adjacency matrices; the models are multitape Turing machines and sequential logarithmic-word RAMs with fixed polynomial-bit-time instructions. Both joint and separate replacement conventions are covered.

Cite (BibTeX)
@misc{OAI:Unconditional-time-lower-bounds-for-Weisfeiler-Leman-equivalence-September-25-2026,
  author = {{OpenAI}},
  title = {{Unconditional time lower bounds for Weisfeiler--Leman equivalence}},
  howpublished = {OpenAI Math Release preprint
                  \href{https://github.com/openai/math/blob/main/preprints/Unconditional-time-lower-bounds-for-Weisfeiler-Leman-equivalence-September-25-2026/paper.pdf}{OAI:Unconditional-time-lower-bounds-for-Weisfeiler-Leman-equivalence-September-25-2026}},
  year = {2026}
}

Variable-dimension Weisfeiler–Leman equivalence on general and subcubic graphs

September 25, 2026 39 pages

Deciding joint-update k-dimensional Weisfeiler–Leman equivalence is EXPTIME\mathsf{EXPTIME}-complete when the two graphs are given by explicit adjacency matrices and k ≥ 2 is encoded in binary. The result holds even for connected simple uncolored graphs of equal positive order and maximum degree at most three.

Cite (BibTeX)
@misc{OAI:Variable-dimension-Weisfeiler-Leman-equivalence-on-general-and-subcubic-graphs-September-25-2026,
  author = {{OpenAI}},
  title = {{Variable-dimension Weisfeiler--Leman equivalence on general and subcubic graphs}},
  howpublished = {OpenAI Math Release preprint
                  \href{https://github.com/openai/math/blob/main/preprints/Variable-dimension-Weisfeiler-Leman-equivalence-on-general-and-subcubic-graphs-September-25-2026/paper.pdf}{OAI:Variable-dimension-Weisfeiler-Leman-equivalence-on-general-and-subcubic-graphs-September-25-2026}},
  year = {2026}
}

Lean formalization

OpenAI's note on what the formalization covers, from lean/docs/133.md.

The computational complexity of Weisfeiler–Leman refinement

The following describes the scope of the Lean formalization related to the following accompanying paper(s):

Scope

The formalization gives the paper's parity-lift graph construction for every Weisfeiler–Leman dimension k≥4k\ge4, under both joint and separate conventions. From a finite choice system it constructs two equal-size uncolored graphs that are Weisfeiler–Leman equivalent exactly when the system has no successful compatible choice. The same equivalence holds for the pure variant.

A successful choice is detected by the specified early round: round two in the joint convention and round one in the separate convention. The graph-size bound is linear in the choice-system size with a factor depending on kk. The paper's conditional running-time exclusions are outside these selected construction statements.

The formalization proves that deciding whether Weisfeiler–Leman refinement of a supplied dimension identifies a graph is EXPTIME-complete under polynomial-time many-one reductions. The input is a nonempty finite simple uncolored graph encoded by its adjacency matrix together with a positive binary-encoded dimension. Identification quantifies over every comparison graph, and the statement includes both the exponential-time upper bound and hardness.

The formalization proves an unconditional deterministic time lower bound for deciding kk-dimensional Weisfeiler–Leman equivalence. For every sufficiently large fixed kk, every deciding algorithm in the stated multitape Turing-machine or sequential logarithmic-word RAM model requires at least nckn^{ck} worst-case time for all sufficiently large graph orders nn, for one absolute c>0c>0.

The bound covers both joint and separate replacement conventions and remains valid for explicit adjacency-matrix inputs restricted to simple connected uncolored graphs of diameter at most two.

The formalization proves EXPTIME-completeness of deciding joint-update kk-dimensional Weisfeiler–Leman equivalence when k≥2k\ge2 is part of the input in binary and the graphs are encoded by explicit adjacency matrices. It proves both the general result and the restriction to connected simple uncolored graphs of equal positive order and maximum degree at most three. Completeness uses polynomial-time many-one reductions.

Comparator links

Result Comparator statement
Parity-lift characterization of Weisfeiler–Leman equivalence ParityLifts.lean
EXPTIME-completeness of Weisfeiler–Leman graph identification WLIdentification.lean
Unconditional sequential time lower bounds for Weisfeiler–Leman equivalence WeisfeilerLeman.lean
EXPTIME-completeness of variable-dimension Weisfeiler–Leman equivalence VariableWL.lean

Data from github.com/openai/math at commit adc7f12, committed October 6, 2026 at 21:58 UTC, last checked for changes about 8 hours ago. Titles, subjects, summaries, abstracts and Lean notes are OpenAI's; page counts are read from the PDFs. The map, related results, search, kinds of results and the named-problem index are Emergent Mind's, built with text embeddings and an LLM, and may contain errors.

An Emergent Mind Labs project. Emergent Mind is not affiliated with OpenAI. None of these results has been peer reviewed. Cite the manuscripts themselves, using the BibTeX on each result's page.