SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2403.16680 · ICLR · 2024

Symmetric Basis Convolutions for Learning Lagrangian Fluid Mechanics

Nils Thuerey, Rene Winchenbach

arXiv · PDF · Open in the Atlas

Code that ran

12 of 19 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; 1 honoured the contract we drafted, 1 violated it and 2 have no contract verdict; 8 failures of our instrument; 7 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/2403.16680.json · badge: /api/ran/2403.16680.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
tum-pbs/SFBC canonical (classifier label)
classifier basis: census
8 ran · 6 no recorded run
tum-pbs/sfbc canonical (classifier label)
classifier basis: classifier_2025
4 ran · 1 no recorded run
FunctionStatusWhere it lives
buildMLP Ran
ran; our contract draft was wrong, not the code
tum-pbs/sfbc/src/BasisConvolution/convLayerv2.py
code served (permissive licence) · get_code("01098f7fb5af8aca")
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) and returned Sequential. 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
from torch import nn

torch.manual_seed(0)

ARGS = {
    "layers": [8, 16, 4],
    "inputFeatures": 5,
    "gain": 1 / np.sqrt(34),
    "useBias": False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential output the recorded ndim 1 and dtype family float describe that coercion, not the output, and are not 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
buildMLPwActivation Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
tum-pbs/sfbc/src/BasisConvolution/convNetv2.py
code served (permissive licence) · get_code("73dad65c3366c0d6")
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) and returned Sequential. 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
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    'layers': [8, 16, 4],
    'inputFeatures': 5,
    'gain': 1 / np.sqrt(34),
    'activation': 'gelu',
    'norm': True,
    'groups': [2, 2, 2],
    'preNorm': False,
    'postNorm': False,
    'noLinear': False,
    'bias': True
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential output the recorded ndim 1 and dtype family float describe that coercion, not the output, and are not 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
buildMLPwActivation Ran
ran on a synthesized input
tum-pbs/SFBC/src/BasisConvolution/convNetv2.py
code served (permissive licence) · get_code("4b944e56a24334ac")
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 Sequential. 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 torch.nn as nn
import numpy as np

torch.manual_seed(0)

ARGS = {
    'layers': [16, 8],
    'inputFeatures': 4,
    'gain': 1 / np.sqrt(34),
    'activation': 'gelu',
    'norm': False,
    'groups': 1,
    'preNorm': False,
    'postNorm': False,
    'noLinear': False,
    'bias': True,
}
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 Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential output the recorded ndim 1 and dtype family float describe that coercion, not the output, and are not 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
buildMLPwDict Ran
ran; our contract draft was wrong, not the code
tum-pbs/sfbc/src/BasisConvolution/convNetv2.py
code served (permissive licence) · get_code("e12dd94211676bc2")
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) and returned Sequential. 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
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    'properties': {
        'layout': [64, 128],
        'output': 32,
        'inputFeatures': 16,
        'channels': [8, 16, 4],
        'gain': 1 / np.sqrt(34),
        'activation': 'gelu',
        'norm': True,
        'preNorm': False,
        'postNorm': False,
        'noLinear': False,
        'bias': True
    }
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential output the recorded ndim 1 and dtype family float describe that coercion, not the output, and are not 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
buildMLPwDict Ran
ran; our contract draft was wrong, not the code
tum-pbs/SFBC/src/BasisConvolution/convNetv2.py
code served (permissive licence) · get_code("48f466cb501e970c")
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 Sequential. 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 torch.nn as nn
import numpy as np

torch.manual_seed(0)

ARGS = {
    "properties": {
        "layout": [64, 32],
        "output": 16,
        "inputFeatures": 8,
        "channels": 4,
        "gain": 1 / np.sqrt(34),
        "activation": "gelu",
        "norm": True,
        "preNorm": False,
        "postNorm": False,
        "noLinear": False,
        "bias": True
    }
}
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 Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential output the recorded ndim 1 and dtype family float describe that coercion, not the output, and are not 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
generateParticles Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
tum-pbs/SFBC/eval/performanceMeasurement.py
code served (permissive licence) · get_code("416ab51cf7beeb85")
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. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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

torch.manual_seed(0)

# The function generateParticles takes parameters:
# - nx: number of points in each dimension (int)
# - dim: dimensionality, defaults to 1 (int, one of 1, 2, or 3)

# Since the function uses device='cuda' internally, we need to handle this carefully.
# We'll pass CPU tensors but note that the function will try to create CUDA tensors.
# We should modify the device to CPU for testing, but since we can't modify the function,
# we'll just set up valid arguments that would work if CUDA were available.

# However, following the rules strictly: pass CPU-valid arguments and let the function
# fail on CUDA if it must. But we can work around by monkey-patching torch temporarily.

ARGS = {
    'nx': 4,
    'dim': 2
}
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 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
getActivationLayer Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
tum-pbs/sfbc/src/BasisConvolution/convNetv2.py
code served (permissive licence) · get_code("3233434225504ec6")
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) and returned ReLU. 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 torch.nn as nn

torch.manual_seed(0)

# The function takes a single parameter: function (str)
# It returns an activation layer based on the string input
# We need to test it with a valid activation function name

ARGS = {
    'function': 'relu'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ReLU — a ReLU output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a ReLU output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not 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
getActivationLayer Ran
ran; our contract draft was wrong, not the code
tum-pbs/SFBC/src/BasisConvolution/convNetv2.py
code served (permissive licence) · get_code("fa19f4aff56b0e77")
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 ReLU. 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 torch.nn as nn

torch.manual_seed(0)

# The function takes a single parameter: function (a string)
# It returns an activation layer based on the string input
# We need to provide valid activation function names

ARGS = {
    'function': 'relu'
}
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 ReLU — a ReLU output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a ReLU output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not 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
interpolant Ran
ran, honoured the contract we drafted
tum-pbs/SFBC/eval/noise/perlin.py
code served (permissive licence) · get_code("bab05c8bf9da0a70")
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 0, dtype family float, shape () 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

torch.manual_seed(0)

def interpolant(t):
    return t * t * t * (t * (t * 6 - 15) + 10)

# The function takes a single parameter 't' which is used in arithmetic operations
# It appears to be a smoothstep/Perlin interpolation function
# 't' should be a numeric value or tensor that supports multiplication and addition

# Create a small tensor as the input
t_value = torch.tensor(0.5)

ARGS = {
    "t": t_value
}
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 Tensor, ndim 0, dtype family float, shape () — shape () 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
pdf Ran
ran, violated the contract we drafted
tum-pbs/SFBC/datasets/generate_test_case_I.py
code served (permissive licence) · get_code("d607897d6ca51fd7")
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 1, dtype family float, shape (4,) 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

# The pdf function takes a single parameter x
# It converts x to a numpy array and performs operations on it
# We need to provide a valid input that can be converted to a numpy array

x_input = np.array([-2.0, -1.0, 0.5, 1.5])

ARGS = {
    'x': x_input
}
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 1, dtype family float, shape (4,) — shape (4,) 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
perlinNoise2D Ran
ran on a synthesized input
tum-pbs/SFBC/eval/noise/perlin.py
code served (permissive licence) · get_code("948426e08e8e096a")
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)

def interpolant(t):
    return t * t * t * (t * (t * 6 - 15) + 10)

ARGS = {
    'shape': (8, 6),
    'res': (4, 3),
    'tileable': (False, False),
    'interpolant': interpolant,
    'rng': np.random.default_rng(seed=42),
    'device': 'cpu',
    'dtype': torch.float32
}
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 — 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
perlinNoise3D Ran
ran; our contract draft was wrong, not the code
tum-pbs/SFBC/eval/noise/perlin.py
code served (permissive licence) · get_code("12f727770539acfe")
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)

def interpolant(t):
    return t * t * t * (t * (t * 6 - 15) + 10)

ARGS = {
    'shape': (8, 6, 4),
    'res': (2, 2, 2),
    'tileable': (False, False, False),
    'interpolant': interpolant,
    'rng': np.random.default_rng(seed=42),
    'device': 'cpu',
    'dtype': torch.float32
}
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 — 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
generateSimplex Not yet run
unverified (harvested, no recorded run)
tum-pbs/SFBC/eval/noise/generator.py
code served (permissive licence) · get_code("0acf386a76e369ea")
How it ran

no invocation recorded.

get_analytical_jacobian Not yet run
unverified (harvested, no recorded run)
tum-pbs/SFBC/eval/gradcheck.py
code served (permissive licence) · get_code("e43814b5b0bb50d9")
How it ran

no invocation recorded.

get_numerical_jacobian Not yet run
unverified (harvested, no recorded run)
tum-pbs/SFBC/eval/gradcheck.py
code served (permissive licence) · get_code("64a48db8e27bbcf2")
How it ran

no invocation recorded.

get_numerical_jacobian_wrt_specific_input Not yet run
unverified (harvested, no recorded run)
tum-pbs/SFBC/eval/gradcheck.py
code served (permissive licence) · get_code("301b0e6586be3e19")
How it ran

no invocation recorded.

initializeWeights2D Not yet run
unverified (harvested, no recorded run)
tum-pbs/sfbc/src/BasisConvolution/convLayerv2.py
code served (permissive licence) · get_code("004b54582bca041e")
How it ran

no invocation recorded.

neighSearch Not yet run
unverified (harvested, no recorded run)
tum-pbs/SFBC/eval/performanceMeasurement.py
code served (permissive licence) · get_code("99edd179c77682a7")
How it ran

no invocation recorded.

supportFromVolume Not yet run
unverified (harvested, no recorded run)
tum-pbs/SFBC/eval/performanceMeasurement.py
code served (permissive licence) · get_code("a59b6f0cab7277f5")
How it ran

no invocation recorded.

Also present in other papers' code

Of 19 samples lifted from this paper's linked repositories, 19 carry a function identity and 0 do not. Of the 16 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 16 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.

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

Learning physical simulations has been an essential and central aspect of many recent research efforts in machine learning, particularly for Navier-Stokes-based fluid mechanics. Classic numerical solvers have traditionally been computationally expensive and challenging to use in inverse problems, whereas Neural solvers aim to address both concerns through machine learning. We propose a general formulation for continuous convolutions using separable basis functions as a superset of existing methods and evaluate a large set of basis functions in the context of (a) a compressible 1D SPH simulation, (b) a weakly compressible 2D SPH simulation, and (c) an incompressible 2D SPH Simulation. We demonstrate that even and odd symmetries included in the basis functions are key aspects of stability and accuracy. Our broad evaluation shows that Fourier-based continuous convolutions outperform all other architectures regarding accuracy and generalization. Finally, using these Fourier-based networks, we show that prior inductive biases, such as window functions, are no longer necessary. An implementation of our approach, as well as complete datasets and solver implementations, is available at https://github.com/tum-pbs/SFBC.

For agents

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

get_harvested_code_for_paper("2403.16680")
get_code_for_paper("2403.16680")
have("2403.16680")

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

curl https://syntology.ai/api/ran/2403.16680.json

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

[![Syntology run record](https://syntology.ai/api/ran/2403.16680.svg)](https://syntology.ai/paper/2403.16680)

Connect an agent — have() is free.