SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2212.02374 · 2022

On the Trade-off between Over-smoothing and Over-squashing in Deep Graph Neural Networks

arXiv · PDF · Open in the Atlas

Code that ran

5 of 9 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 5 have no contract verdict; 0 failures of our instrument; 4 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2212.02374.json · badge: /api/ran/2212.02374.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
jhonygiraldo/sjlr canonical (classifier label)
classifier basis not recorded on the link
5 ran · 4 no recorded run
FunctionStatusWhere it lives
balanced_forman_curvature Ran
ran on a synthesized input
jhonygiraldo/sjlr/models/BFC_curvature.py
code served (permissive licence) · get_code("dc6573716bd3389f")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 1, dtype family float, shape (5,) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# Create a simple undirected graph with 5 nodes
# Edge list: (0,1), (1,2), (2,3), (3,4), (0,2)
edge_index = torch.tensor([
    [0, 1, 2, 3, 0],
    [1, 2, 3, 4, 2]
], dtype=torch.long)

n = 5
is_undirected = True

ARGS = {
    "edge_index": edge_index,
    "n": n,
    "is_undirected": is_undirected
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 1, dtype family float, shape (5,) — shape (5,) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
compute_JLC_matrices Ran
ran on a synthesized input
jhonygiraldo/sjlr/models/JLC_curvature.py
code served (permissive licence) · get_code("589c9df78f56f0af")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# Create a small graph with 4 nodes and 5 edges
n = 4
n_edges = 5

# Create edge_index: shape [2, 5]
# Edges: (0,1), (1,2), (2,3), (3,0), (0,2)
edge_index = torch.LongTensor([
    [0, 1, 2, 3, 0],
    [1, 2, 3, 0, 2]
])

# Parameters for compute_JLC_matrices
pA = 0.1  # small probability for adding edges
force_undirected = False

ARGS = {
    "edge_index": edge_index,
    "n": n,
    "n_edges": n_edges,
    "pA": pA,
    "force_undirected": force_undirected
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch, torch_geometric — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
compute_balaced_forman_curvature Ran
ran on a synthesized input
jhonygiraldo/sjlr/models/BFC_curvature.py
code served (permissive licence) · get_code("2697497d87010a62")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 1, dtype family float, shape (1,) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# Create a small graph adjacency matrix
# 5 nodes, dense representation
num_nodes = 5
adj_dense = torch.tensor([
    [0, 1, 1, 0, 0],
    [1, 0, 1, 1, 0],
    [1, 1, 0, 1, 1],
    [0, 1, 1, 0, 1],
    [0, 0, 1, 1, 0]
], dtype=torch.long)

# Create sparse version
indices = torch.nonzero(adj_dense, as_tuple=False).t()
values = torch.ones(indices.shape[1], dtype=torch.long)
adj_sparse = torch.sparse_coo_tensor(indices, values, adj_dense.shape, dtype=torch.long)

# Select two connected nodes for the base edge
node_x = 0
node_y = 1

# Precompute neighbor sets
s_x = torch.where(adj_dense[node_x] > 0)[0]
s_y = torch.where(adj_dense[node_y] > 0)[0]

ARGS = {
    "adj": adj_dense,
    "node_x": node_x,
    "node_y": node_y,
    "is_adj_sparse": False,
    "s_x": s_x,
    "s_y": s_y
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 1, dtype family float, shape (1,) — shape (1,) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
compute_jost_liu_curvature Ran
ran on a synthesized input
jhonygiraldo/sjlr/models/JLC_curvature.py
code served (permissive licence) · get_code("962cbf0cde3c7b42")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 1, dtype family float, shape (1,) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# Create a small adjacency matrix (5x5 graph)
# Using a dense tensor for simplicity
adj_dense = torch.tensor([
    [0, 1, 1, 0, 1],
    [1, 0, 1, 1, 0],
    [1, 1, 0, 1, 1],
    [0, 1, 1, 0, 1],
    [1, 0, 1, 1, 0]
], dtype=torch.long)

# Create a sparse version for testing the sparse path
adj_sparse = adj_dense.to_sparse()

# Node indices to compute curvature between
node_x = 0
node_y = 1

# Neighbor sets (precomputed for efficiency, optional)
s_x = torch.tensor([1, 2, 4], dtype=torch.long)  # neighbors of node 0
s_y = torch.tensor([0, 2, 3], dtype=torch.long)  # neighbors of node 1

ARGS = {
    "adj": adj_dense,
    "node_x": node_x,
    "node_y": node_y,
    "is_adj_sparse": False,
    "s_x": s_x,
    "s_y": s_y,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 1, dtype family float, shape (1,) — shape (1,) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
jost_liu_curvature Ran
ran on a synthesized input
jhonygiraldo/sjlr/models/JLC_curvature.py
code served (permissive licence) · get_code("006b8162ce0068b8")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 1, dtype family float, shape (5,) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# Create a small graph with 4 nodes
n = 4

# Create edge_index for a small undirected graph
# Let's create edges: (0,1), (1,2), (2,3), (0,3), (1,3)
edge_index = torch.tensor([
    [0, 1, 2, 0, 1],
    [1, 2, 3, 3, 3]
], dtype=torch.long)

is_undirected = True

ARGS = {
    "edge_index": edge_index,
    "n": n,
    "is_undirected": is_undirected
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 1, dtype family float, shape (5,) — shape (5,) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_component Not yet run
unverified (harvested, no recorded run)
jhonygiraldo/sjlr/load_data/data.py
pointer only (licence: MIT); the stored copy get_code serves, from ssjcode/FSD-CAP, is pointer only (licence: NONE) · get_code("c68140584cc0e991")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from ssjcode/FSD-CAP, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

get_dataset Not yet run
unverified (harvested, no recorded run)
jhonygiraldo/sjlr/load_data/data.py
code served (permissive licence) · get_code("d2845200b34fc758")
How it ran

no invocation recorded.

get_largest_connected_component Not yet run
unverified (harvested, no recorded run)
jhonygiraldo/sjlr/load_data/data.py
pointer only (licence: MIT); the stored copy get_code serves, from ssjcode/FSD-CAP, is pointer only (licence: NONE) · get_code("b36878b25e3455a3")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from ssjcode/FSD-CAP, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

stochastic_discrete_ricci_flow_rewiring Not yet run
unverified (harvested, no recorded run)
jhonygiraldo/sjlr/models/BFC_curvature.py
code served (permissive licence) · get_code("423f80b3cce420e7")
How it ran

no invocation recorded.

Also present in other papers' code

Of 9 samples lifted from this paper's linked repositories, 9 carry a function identity and 0 do not. Of the 9 distinct functions with an identity, 2 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 0 the same month, 2 later) and 7 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
get_component 7 other papers
5 dated earlier, 2 later
1911.05485 (2019-11)
Diffusion Improves Graph Learning
2111.12128 (2021-11)
On the Unreasonable Effectiveness of Feature propagation in Learning on Graphs with Missing Node Features
2206.10991 (2022-06)
Understanding convolution on graphs via energies
and 4 more not named here
get_largest_connected_component 6 other papers
4 dated earlier, 2 later
1911.05485 (2019-11)
Diffusion Improves Graph Learning
2111.12128 (2021-11)
On the Unreasonable Effectiveness of Feature propagation in Learning on Graphs with Missing Node Features
2206.10991 (2022-06)
Understanding convolution on graphs via energies
and 3 more not named here

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Graph Neural Networks (GNNs) have succeeded in various computer science applications, yet deep GNNs underperform their shallow counterparts despite deep learning's success in other domains. Over-smoothing and over-squashing are key challenges when stacking graph convolutional layers, hindering deep representation learning and information propagation from distant nodes. Our work reveals that over-smoothing and over-squashing are intrinsically related to the spectral gap of the graph Laplacian, resulting in an inevitable trade-off between these two issues, as they cannot be alleviated simultaneously. To achieve a suitable compromise, we propose adding and removing edges as a viable approach. We introduce the Stochastic Jost and Liu Curvature Rewiring (SJLR) algorithm, which is computationally efficient and preserves fundamental properties compared to previous curvature-based methods. Unlike existing approaches, SJLR performs edge addition and removal during GNN training while maintaining the graph unchanged during testing. Comprehensive comparisons demonstrate SJLR's competitive performance in addressing over-smoothing and over-squashing.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2212.02374")
get_code_for_paper("2212.02374")
have("2212.02374")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2212.02374.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2212.02374.svg)](https://syntology.ai/paper/2212.02374)

Connect an agent — have() is free.