SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2003.11657 · 2020

Deep Graph Matching via Blackbox Differentiation of Combinatorial Solvers

arXiv · PDF · Open in the Atlas

Code that ran

10 of 11 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, 1 violated it and 5 have no contract verdict; 4 failures of our instrument; 1 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/2003.11657.json · badge: /api/ran/2003.11657.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
martius-lab/blackbox-backprop canonical (classifier label)
classifier basis: census
7 ran · 1 no recorded run
Thinklab-SJTU/ThinkMatch reimplementation (classifier label)
classifier basis: classifier_2025
3 ran · 0 no recorded run
FunctionStatusWhere it lives
concat_features Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
Thinklab-SJTU/ThinkMatch/models/BBGM/model.py
pointer only (licence: NOASSERTION) · get_code("bc40b9897e027ece")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, Thinklab-SJTU/ThinkMatch (NOASSERTION). 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. A run is recorded.

darker Ran
ran on a synthesized input
martius-lab/blackbox-backprop/visualization_utils.py
code served (permissive licence) · get_code("b9da24f59f9a500a")
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, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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)

# The function darker takes a color (tuple of r, g, b, a values) and a factor
# It multiplies r, g, b by the factor while keeping a unchanged
# color is expected to be a tuple or sequence with 4 elements

color = (1.0, 0.5, 0.8, 1.0)
factor = 0.7

ARGS = {
    "color": color,
    "factor": factor
}
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 tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
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
dijkstra Ran
ran on a synthesized input
martius-lab/blackbox-backprop/blackbox_backprop/shortest_path.py
code served (permissive licence) · get_code("79d845c43088342b")
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 ndarray, ndim 2, dtype family float, shape (4, 3) 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 numpy as np
from functools import partial

# Create a small test matrix for the dijkstra function
# The function expects a 2D numpy array where each element represents the cost of a node
# We'll create a small 4x3 grid with varying costs
matrix = np.array([
    [1.0, 2.0, 3.0],
    [4.0, 5.0, 6.0],
    [7.0, 8.0, 9.0],
    [10.0, 11.0, 12.0]
], dtype=np.float64)

ARGS = {
    "matrix": matrix
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family float, shape (4, 3) — shape (4, 3) 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
interpolate Ran
ran on a synthesized input
martius-lab/blackbox-backprop/visualization_utils.py
code served (permissive licence) · get_code("7a3fc706eff98f36")
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, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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

# Analyze the function: interpolate takes three parameters
# - color1: tuple/list of color values (iterable of numbers)
# - color2: tuple/list of color values (iterable of numbers)
# - factor: a number between 0 and 1 (represents interpolation factor)

# The function zips color1 and color2, so they need same length
# It performs arithmetic operations on the elements
# Returns a tuple of interpolated values

# Create valid color tuples (e.g., RGB values)
color1 = (255, 128, 64, 32)  # 4-element color tuple
color2 = (100, 200, 150, 50)  # 4-element color tuple
factor = 0.5  # interpolation factor between 0 and 1

ARGS = {
    "color1": color1,
    "color2": color2,
    "factor": factor
}
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 tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
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
lexico_iter Ran
ran; our contract draft was wrong, not the code
Thinklab-SJTU/ThinkMatch/models/BBGM/model.py
pointer only (licence: NOASSERTION) · get_code("a81be531136a1b00")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, Thinklab-SJTU/ThinkMatch (NOASSERTION). 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. A run is recorded.

maybe_parallelize Ran
ran on a synthesized input
martius-lab/blackbox-backprop/blackbox_backprop/utils.py
code served (permissive licence) · get_code("f4d8af262e615db7")
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 list, ndim 1, dtype family int; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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 sys

# Define a simple test function that will be parallelized
def test_function(x):
    """Simple function that takes an argument and returns a result"""
    if isinstance(x, dict):
        return x['value'] * 2
    else:
        return x * 2

# Create argument list with mixed types
arg_list = [
    {'value': 1},
    {'value': 2},
    {'value': 3},
    {'value': 4},
]

ARGS = {
    'function': test_function,
    'arg_list': arg_list,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family int — a list output has no array shape of its own; none is stated
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
normalize_over_channels Ran
ran, violated the contract we drafted
Thinklab-SJTU/ThinkMatch/models/BBGM/model.py
pointer only (licence: NOASSERTION) · get_code("f12ddebef47d400e")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, Thinklab-SJTU/ThinkMatch (NOASSERTION). 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. A run is recorded.

rank Ran
ran; our contract draft was wrong, not the code
martius-lab/blackbox-backprop/blackbox_backprop/ranking.py
pointer only (licence: MIT); the stored copy get_code serves, from BorealisAI/ranksim-imbalanced-regression, is pointer only (licence: NOASSERTION) · get_code("2e1e2e8c543195f7")
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 BorealisAI/ranksim-imbalanced-regression, and GitHub could not classify the licence of that repository (NOASSERTION). 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. A run is recorded.

rank_normalised Ran
ran; our contract draft was wrong, not the code
martius-lab/blackbox-backprop/blackbox_backprop/ranking.py
pointer only (licence: MIT); the stored copy get_code serves, from BorealisAI/ranksim-imbalanced-regression, is pointer only (licence: NOASSERTION) · get_code("ed49cae4f66fd45c")
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 BorealisAI/ranksim-imbalanced-regression, and GitHub could not classify the licence of that repository (NOASSERTION). 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. A run is recorded.

subtour Ran
ran on a synthesized input
martius-lab/blackbox-backprop/blackbox_backprop/travelling_salesman.py
code served (permissive licence) · get_code("c08e18b8412b9339")
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 list, ndim 1, dtype family int; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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

# Analyze the function:
# - n: an integer representing the number of nodes
# - edges: a list of tuples (x, y) representing directed edges between nodes

# The function builds a graph from edges and finds cycles, returning the shortest one.
# It uses: visited as a boolean list, selected as adjacency list, and traverses the graph.

# Construct minimal valid arguments:
n = 5
edges = [
    (0, 1),
    (1, 2),
    (2, 0),
    (3, 4),
    (4, 3)
]

ARGS = {
    "n": n,
    "edges": edges
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family int — a list output has no array shape of its own; none is stated
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
gurobi_tsp Not yet run
unverified (harvested, no recorded run)
martius-lab/blackbox-backprop/blackbox_backprop/travelling_salesman.py
code served (permissive licence) · get_code("8073153564d8de8e")
How it ran

no invocation recorded.

Also present in other papers' code

Of 11 samples lifted from this paper's linked repositories, 11 carry a function identity and 0 do not. Of the 11 distinct functions with an identity, 5 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 5 later) and 6 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
rank 4 other papers
4 later
2110.01445 (2021-10)
Robust and Decomposable Average Precision for Image Retrieval
2205.15236 (2022-05)
RankSim: Ranking Similarity Regularization for Deep Imbalanced Regression
2410.22124 (2024-10)
RankUp: Boosting Semi-Supervised Regression with an Auxiliary Ranking Classifier
and 1 more not named here
rank_normalised 4 other papers
4 later
2110.01445 (2021-10)
Robust and Decomposable Average Precision for Image Retrieval
2205.15236 (2022-05)
RankSim: Ranking Similarity Regularization for Deep Imbalanced Regression
2410.22124 (2024-10)
RankUp: Boosting Semi-Supervised Regression with an Auxiliary Ranking Classifier
and 1 more not named here
concat_features 1 other paper
1 later
2212.04085 (2022-12)
Graph Matching with Bi-level Noisy Correspondence
lexico_iter 1 other paper
1 later
2212.04085 (2022-12)
Graph Matching with Bi-level Noisy Correspondence
normalize_over_channels 1 other paper
1 later
2212.04085 (2022-12)
Graph Matching with Bi-level Noisy Correspondence

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

Building on recent progress at the intersection of combinatorial optimization and deep learning, we propose an end-to-end trainable architecture for deep graph matching that contains unmodified combinatorial solvers. Using the presence of heavily optimized combinatorial solvers together with some improvements in architecture design, we advance state-of-the-art on deep graph matching benchmarks for keypoint correspondence. In addition, we highlight the conceptual advantages of incorporating solvers into deep learning architectures, such as the possibility of post-processing with a strong multi-graph matching solver or the indifference to changes in the training setting. Finally, we propose two new challenging experimental setups. The code is available at https://github.com/martius-lab/blackbox-deep-graph-matching

For agents

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

get_harvested_code_for_paper("2003.11657")
get_code_for_paper("2003.11657")
have("2003.11657")

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

curl https://syntology.ai/api/ran/2003.11657.json

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

[![Syntology run record](https://syntology.ai/api/ran/2003.11657.svg)](https://syntology.ai/paper/2003.11657)

Connect an agent — have() is free.