Papers › Neural Sheaf Diffusion: A Topological Perspective on Heterophily and Oversmoothing in GNNs

Neural Sheaf Diffusion: A Topological Perspective on Heterophily and Oversmoothing in GNNs

9 Feb 2022arXiv:2202.04579archive 2025-07-28

Cristian Bodnar, Francesco Di Giovanni, Benjamin Paul Chamberlain, Pietro Liò, Michael M. Bronstein

Cellular sheaves equip graphs with a "geometrical" structure by assigning vector spaces and linear maps to nodes and edges. Graph Neural Networks (GNNs) implicitly assume a graph with a trivial underlying sheaf. This choice is reflected in the structure of the graph Laplacian operator, the properties of the associated diffusion equation, and the characteristics of the convolutional models that discretise this equation. In this paper, we use cellular sheaf theory to show that the underlying geometry of the graph is deeply linked with the performance of GNNs in heterophilic settings and their oversmoothing behaviour. By considering a hierarchy of increasingly general sheaves, we study how the ability of the sheaf diffusion process to achieve linear separation of the classes in the infinite time limit expands. At the same time, we prove that when the sheaf is non-trivial, discretised parametric diffusion processes have greater control than GNNs over their asymptotic behaviour. On the practical side, we study how sheaves can be learned from data. The resulting sheaf diffusion models have many desirable properties that address the limitations of classical graph diffusion equations (and corresponding GNN models) and obtain competitive results in heterophilic settings. Overall, our work provides new connections between GNNs and algebraic topology and would be of interest to both fields.

PaperPDFCodeCode Syntology ranMethod as code

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.

Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, on Syntology's MCP service (how to connect):

Code

Syntology 1 code sample harvested from 1 repository linked to this paper: 0 ran · 1 unverified.

By repository: official repository: 1 sample from 1 repository, 0 ran (0 with no instrument failure; 0 where Syntology's instrument failed). The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

twitter-research/neural-sheaf-diffusion officialmentioned in papermentioned on GitHubpytorchSyntology: 1 sample, 0 ran · 1 unverified report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

1 sample harvested: 0 ran · 1 unverified. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

1unverified

Licence: 0 of the 1 sample are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from twitter-research/neural-sheaf-diffusion. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of six reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

twitter-research/neural-sheaf-diffusion official repository 1 sample, 0 ran · 1 unverified
SheafDiffusion models/sheaf_base.py unverified Apache-2.0 (permissive) · 2000cda4ded21558 · report

Method as code Syntology

What Syntology's graph records as this paper's claimed method, lifted into code: each assembly names its bricks (functions and classes), each with its own role and its own verification level. An assembly is not a check that the code implements the method correctly or reproduces the paper's results. Levels are each brick's own verification level in Syntology's graph (level 0: no recorded run), listed brick by brick and never combined into one level; “ran” means the brick executed on a synthesized input, not that its output is correct, and the instrument figure counts failures of Syntology's instrument, not of the code. Where a model read the code against the paper's claim, each assembly's label verdict is one model's judgement, kept in this page's JSON twin; it is not calibrated, so no card prints it.

1 assembly on this paper: 1 harvested, 0 generated.

harvested Assembly for Neural Sheaf Diffusion (graph method: Neural Sheaf Diffusion). 1 brick; individually on a fixture: 0 ran · 1 unverified. The assembly has not been run end-to-end.

  1. SheafDiffusion entry · level 0 · unverified · 2000cda4ded21558

Each brick ends with its code_sha256, Syntology's identity for that exact code; click one to copy the get_code call that returns it. For agents, on Syntology's MCP service: assembly(arxiv_id="2202.04579") returns these assemblies with each brick's role and wiring (how to connect).

Tasks

Node ClassificationNode Classification on Non-Homophilic (Heterophilic) Graphs

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Node Classification Actor O(d)-NSD Accuracy 37.81 ± 1.15 #14 of 62 Archive leaderboard report
Node Classification Actor Gen-NSD Accuracy 37.80 ± 1.22 #15 of 62 Archive leaderboard report
Node Classification Actor Diag-NSD Accuracy 37.79 ± 1.01 #16 of 62 Archive leaderboard report
Node Classification Chameleon Diag-NSD Accuracy 68.68 ± 1.73 #36 of 61 Archive leaderboard report
Node Classification Chameleon O(d)-NSD Accuracy 68.04 ± 1.58 #41 of 61 Archive leaderboard report
Node Classification Chameleon Gen-NSD Accuracy 67.93 ± 1.58 #42 of 61 Archive leaderboard report
Node Classification Citeseer (48%/32%/20% fixed splits) Diag-NSD 1:1 Accuracy 77.14 ± 1.85 #10 of 26 Archive leaderboard report
Node Classification Citeseer (48%/32%/20% fixed splits) O(d)-NSD 1:1 Accuracy 76.70 ± 1.57 #17 of 26 Archive leaderboard report
Node Classification Citeseer (48%/32%/20% fixed splits) Gen-NSD 1:1 Accuracy 76.32 ± 1.65 #19 of 26 Archive leaderboard report
Node Classification Cora (48%/32%/20% fixed splits) Gen-NSD 1:1 Accuracy 87.30 ± 1.15 #18 of 26 Archive leaderboard report
Node Classification Cora (48%/32%/20% fixed splits) Diag-NSD 1:1 Accuracy 87.14 ± 1.06 #19 of 26 Archive leaderboard report
Node Classification Cora (48%/32%/20% fixed splits) O(d)-NSD 1:1 Accuracy 86.90 ± 1.13 #21 of 26 Archive leaderboard report
Node Classification Cornell Diag-NSD Accuracy 86.49 ± 7.35 #8 of 60 Archive leaderboard report
Node Classification Cornell Gen-NSD Accuracy 85.68 ± 6.51 #18 of 60 Archive leaderboard report
Node Classification Cornell O(d)-NSD Accuracy 84.86 ± 4.71 #26 of 60 Archive leaderboard report
Node Classification PubMed (48%/32%/20% fixed splits) O(d)-NSD 1:1 Accuracy 89.49 ± 0.40 #10 of 26 Archive leaderboard report
Node Classification PubMed (48%/32%/20% fixed splits) Diag-NSD 1:1 Accuracy 89.42 ± 0.43 #11 of 26 Archive leaderboard report
Node Classification PubMed (48%/32%/20% fixed splits) Gen-NSD 1:1 Accuracy 89.33 ± 0.35 #12 of 26 Archive leaderboard report
Node Classification Squirrel O(d)-NSD Accuracy 56.34 ± 1.32 #33 of 59 Archive leaderboard report
Node Classification Squirrel Diag-NSD Accuracy 54.78 ± 1.81 #39 of 59 Archive leaderboard report
Node Classification Squirrel Gen-NSD Accuracy 53.17 ± 1.31 #41 of 59 Archive leaderboard report
Node Classification Texas O(d)-NSD Accuracy 85.95 ± 5.51 #25 of 62 Archive leaderboard report
Node Classification Texas Diag-NSD Accuracy 85.67 ± 6.95 #28 of 62 Archive leaderboard report
Node Classification Texas Gen-NSD Accuracy 82.97 ± 5.13 #45 of 62 Archive leaderboard report
Node Classification Wisconsin O(d)-NSD Accuracy 89.41 ± 4.74 #5 of 63 Archive leaderboard report
Node Classification Wisconsin Gen-NSD Accuracy 89.21 ± 3.84 #6 of 63 Archive leaderboard report
Node Classification Wisconsin Diag-NSD Accuracy 88.63 ± 2.75 #11 of 63 Archive leaderboard report
Node Classification on Non-Homophilic (Heterophilic) Graphs Chameleon (48%/32%/20% fixed splits) Diag-NSD 1:1 Accuracy 68.68 ± 1.73 #14 of 29 Archive leaderboard report
Node Classification on Non-Homophilic (Heterophilic) Graphs Chameleon (48%/32%/20% fixed splits) O(d)-NSD 1:1 Accuracy 68.04 ± 1.58 #17 of 29 Archive leaderboard report
Node Classification on Non-Homophilic (Heterophilic) Graphs Chameleon (48%/32%/20% fixed splits) Gen-NSD 1:1 Accuracy 67.93 ± 1.58 #18 of 29 Archive leaderboard report
Node Classification on Non-Homophilic (Heterophilic) Graphs Cornell (48%/32%/20% fixed splits) Diag-NSD 1:1 Accuracy 86.49 ± 7.35 #1 of 27 Archive leaderboard report
Node Classification on Non-Homophilic (Heterophilic) Graphs Cornell (48%/32%/20% fixed splits) Gen-NSD 1:1 Accuracy 85.68 ± 6.51 #7 of 27 Archive leaderboard report
Node Classification on Non-Homophilic (Heterophilic) Graphs Cornell (48%/32%/20% fixed splits) O(d) - NSD 1:1 Accuracy 84.86 ± 4.71 #13 of 27 Archive leaderboard report
Node Classification on Non-Homophilic (Heterophilic) Graphs Film(48%/32%/20% fixed splits) O(d)-NSD 1:1 Accuracy 37.81 ± 1.15 #2 of 26 Archive leaderboard report
Node Classification on Non-Homophilic (Heterophilic) Graphs Film(48%/32%/20% fixed splits) Gen-NSD 1:1 Accuracy 37.80 ± 1.22 #3 of 26 Archive leaderboard report
Node Classification on Non-Homophilic (Heterophilic) Graphs Film(48%/32%/20% fixed splits) Diag-NSD 1:1 Accuracy 37.79 ± 1.01 #4 of 26 Archive leaderboard report
Node Classification on Non-Homophilic (Heterophilic) Graphs Squirrel (48%/32%/20% fixed splits) O(d)-NSD 1:1 Accuracy 56.34 ± 1.32 #14 of 29 Archive leaderboard report
Node Classification on Non-Homophilic (Heterophilic) Graphs Squirrel (48%/32%/20% fixed splits) Diag-NSD 1:1 Accuracy 54.78 ± 1.81 #17 of 29 Archive leaderboard report
Node Classification on Non-Homophilic (Heterophilic) Graphs Squirrel (48%/32%/20% fixed splits) Gen-NSD 1:1 Accuracy 53.17 ± 1.31 #18 of 29 Archive leaderboard report
Node Classification on Non-Homophilic (Heterophilic) Graphs Texas (48%/32%/20% fixed splits) O(d)-NSD 1:1 Accuracy 85.95 ± 5.51 #7 of 26 Archive leaderboard report
Node Classification on Non-Homophilic (Heterophilic) Graphs Texas (48%/32%/20% fixed splits) Diag-NSD 1:1 Accuracy 85.67 ± 6.95 #8 of 26 Archive leaderboard report
Node Classification on Non-Homophilic (Heterophilic) Graphs Texas (48%/32%/20% fixed splits) Gen-NSD 1:1 Accuracy 82.97 ± 5.13  #16 of 26 Archive leaderboard report
Node Classification on Non-Homophilic (Heterophilic) Graphs Wisconsin (48%/32%/20% fixed splits) O(d)-NSD 1:1 Accuracy 89.41 ± 4.74 #1 of 26 Archive leaderboard report
Node Classification on Non-Homophilic (Heterophilic) Graphs Wisconsin (48%/32%/20% fixed splits) Gen-NSD 1:1 Accuracy 89.21 ± 3.84 #2 of 26 Archive leaderboard report
Node Classification on Non-Homophilic (Heterophilic) Graphs Wisconsin (48%/32%/20% fixed splits) Diag-NSD 1:1 Accuracy 88.63 ± 2.75 #3 of 26 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

Diffusion

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections