SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2110.10249 · NeurIPS · 2022

Neural Stochastic PDEs: Resolution-Invariant Learning of Continuous Spatiotemporal Dynamics

Cristopher Salvi, Maud Lemercier, Andris Gerasimovičs

arXiv · PDF · Open in the Atlas

Code that ran

18 of 29 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, 0 violated it and 16 have no contract verdict; 1 failure of our instrument; 11 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/2110.10249.json · badge: /api/ran/2110.10249.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
crispitagorico/neural-spdes community (classifier label)
classifier basis not recorded on the link
15 ran · 8 no recorded run
crispitagorico/torchspde community (classifier label)
classifier basis not recorded on the link
3 ran · 3 no recorded run
FunctionStatusWhere it lives
compl_mat_vec_mul_1d Ran
ran on a synthesized input
crispitagorico/neural-spdes/torchspde/diffeq_solver.py
code served (permissive licence) · get_code("94effb2bae1b9106")
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 4, dtype family float, shape (2, 2, 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 torch
from functools import partial

torch.manual_seed(0)

# Parameters for compl_mat_vec_mul_1d
batch = 2
dim_x = 4
hidden_size = 3

# A: contains complex matrices of coefficients (2, dim_x, hidden_size, hidden_size)
# The 2 represents real and imaginary parts
A = torch.randn(2, dim_x, hidden_size, hidden_size)

# z: (batch, 2, dim_x, hidden_size)
# The 2 represents real and imaginary parts
z = torch.randn(batch, 2, dim_x, hidden_size)

ARGS = {
    "A": A,
    "z": z,
}
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 4, dtype family float, shape (2, 2, 4, 3) — shape (2, 2, 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
compl_mat_vec_mul_2d Ran
ran on a synthesized input
crispitagorico/neural-spdes/torchspde/diffeq_solver.py
code served (permissive licence) · get_code("c608b1fc1a12ef8c")
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 5, dtype family float, shape (2, 2, 4, 3, 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
from functools import partial

torch.manual_seed(0)

# A: (2, dim_x, dim_y, hidden_size, hidden_size)
# z: (batch, 2, dim_x, dim_y, hidden_size)

batch = 2
dim_x = 4
dim_y = 3
hidden_size = 5

# Create A with shape (2, dim_x, dim_y, hidden_size, hidden_size)
# A[0] and A[1] are the real and imaginary parts of complex matrices
A = torch.randn(2, dim_x, dim_y, hidden_size, hidden_size)

# Create z with shape (batch, 2, dim_x, dim_y, hidden_size)
# z[:, 0] is real part, z[:, 1] is imaginary part
z = torch.randn(batch, 2, dim_x, dim_y, hidden_size)

ARGS = {
    'A': A,
    'z': z
}
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 5, dtype family float, shape (2, 2, 4, 3, 5) — shape (2, 2, 4, 3, 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
compl_mul1d_spatial Ran
ran on a synthesized input
crispitagorico/neural-spdes/baselines/NCDEFNO_1D.py
code served (permissive licence) · get_code("0bf977a8ba73d4a0")
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 3, dtype family float, shape (2, 3, 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

torch.manual_seed(0)

# Create two tensors with shapes compatible for the einsum operation
# einsum pattern: 'aib, ijb -> ajb'
# a shape: (a, i, b)
# b shape: (i, j, b)
# output shape: (a, j, b)

# Let's use distinct dimensions:
# a dimension: 2 (batch size)
# i dimension: 4 (shared dimension)
# j dimension: 3 (output dimension)
# b dimension: 5 (feature dimension)

a = torch.randn(2, 4, 5)  # shape (a=2, i=4, b=5)
b = torch.randn(4, 3, 5)  # shape (i=4, j=3, b=5)

ARGS = {
    'a': a,
    'b': b,
}
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 3, dtype family float, shape (2, 3, 5) — shape (2, 3, 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
compl_mul1d_time Ran
ran on a synthesized input
crispitagorico/torchspde/torchspde/fixed_point_solver.py
code served (permissive licence) · get_code("78f8021598bf8e54")
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 4, dtype family float, shape (2, 3, 5, 6) 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)

# The function performs an einsum operation on two tensors
# func_fft has shape (b, i, x, t) where:
#   b = batch size
#   i = input feature dimension
#   x = spatial/feature dimension
#   t = time/frequency dimension
# kernel_tensor has shape (i, j, x, t) where:
#   i = input feature dimension
#   j = output feature dimension
#   x = spatial/feature dimension
#   t = time/frequency dimension
# Output has shape (b, j, x, t)

# Create small but distinct dimension sizes
batch_size = 2
input_features = 4
output_features = 3
spatial_dim = 5
time_dim = 6

func_fft = torch.randn(batch_size, input_features, spatial_dim, time_dim)
kernel_tensor = torch.randn(input_features, output_features, spatial_dim, time_dim)

ARGS = {
    'func_fft': func_fft,
    'kernel_tensor': kernel_tensor
}
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 4, dtype family float, shape (2, 3, 5, 6) — shape (2, 3, 5, 6) 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
compl_mul2d Ran
ran on a synthesized input
crispitagorico/torchspde/torchspde/fixed_point_solver.py
code served (permissive licence) · get_code("43acc1a91197c15d")
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 4, dtype family float, shape (2, 6, 3, 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

torch.manual_seed(0)

# The function uses einsum with 'bixt, ijxt -> bjxt'
# func_fft should have shape (b, i, x, t)
# kernel_tensor should have shape (i, j, x, t)

func_fft = torch.randn(2, 4, 3, 5)  # (batch=2, i=4, x=3, t=5)
kernel_tensor = torch.randn(4, 6, 3, 5)  # (i=4, j=6, x=3, t=5)

ARGS = {
    'func_fft': func_fft,
    'kernel_tensor': kernel_tensor
}
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 4, dtype family float, shape (2, 6, 3, 5) — shape (2, 6, 3, 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
compl_mul2d_spatial Ran
ran on a synthesized input
crispitagorico/neural-spdes/baselines/NCDEFNO_2D.py
code served (permissive licence) · get_code("d25f0fa34864d956")
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 4, dtype family float, shape (2, 6, 4, 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

torch.manual_seed(0)

# The function uses einsum with pattern 'aibc, ijbc -> ajbc'
# a has shape (a, i, b, c)
# b has shape (i, j, b, c)
# result has shape (a, j, b, c)

# Create small tensors with distinct dimensions
# a: (2, 3, 4, 5) - batch=2, i=3, b=4, c=5
a = torch.randn(2, 3, 4, 5)

# b: (3, 6, 4, 5) - i=3 (must match a's i), j=6, b=4 (must match a's b), c=5 (must match a's c)
b = torch.randn(3, 6, 4, 5)

ARGS = {
    'a': a,
    'b': b
}
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 4, dtype family float, shape (2, 6, 4, 5) — shape (2, 6, 4, 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
compl_mul3d Ran
ran on a synthesized input
crispitagorico/torchspde/torchspde/fixed_point_solver.py
code served (permissive licence) · get_code("6ed35ce88ea0e5a4")
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 5, dtype family float, shape (2, 7, 3, 5, 6) 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 torch.nn as nn

torch.manual_seed(0)

# The function performs einsum operation on two tensors:
# func_fft: shape (b, i, x, y, t)
# kernel_tensor: shape (i, j, x, y, t)
# output: shape (b, j, x, y, t)

# Create func_fft tensor with shape (batch=2, i=4, x=3, y=5, t=6)
func_fft = torch.randn(2, 4, 3, 5, 6)

# Create kernel_tensor with shape (i=4, j=7, x=3, y=5, t=6)
kernel_tensor = torch.randn(4, 7, 3, 5, 6)

ARGS = {
    'func_fft': func_fft,
    'kernel_tensor': kernel_tensor
}
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 5, dtype family float, shape (2, 7, 3, 5, 6) — shape (2, 7, 3, 5, 6) 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
dataloader_deeponet_1d_xi Ran
ran on a synthesized input
crispitagorico/neural-spdes/baselines/deepOnet.py
code served (permissive licence) · get_code("ed6e3890adf18ecf")
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)

class UnitGaussianNormalizer:
    def __init__(self, x):
        self.mean = x.mean(dim=0)
        self.std = x.std(dim=0) + 1e-8
    
    def encode(self, x):
        return (x - self.mean) / self.std
    
    def decode(self, x):
        return x * self.std + self.mean

# Create small tensors with distinct dimensions
# u: (batch, dim_x, T) where batch=3, dim_x=4, T=5
u = torch.randn(3, 4, 5)

# xi: (batch, dim_x, T) where batch=3, dim_x+1=5, T=5
# Note: xi has shape (batch, dim_x+1, T) because code does xi[:, :-1, ...]
xi = torch.randn(3, 5, 5)

ARGS = {
    'u': u,
    'xi': xi,
    'ntrain': 2,
    'ntest': 1,
    'T': 5,
    'sub_t': 1,
    'batch_size': 2,
    'dim_x': 4,
    'normalizer': False,
    'dataset': None,
}
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
dataloader_fno_1d_u0 Ran
ran on a synthesized input
crispitagorico/neural-spdes/baselines/FNO1D.py
code served (permissive licence) · get_code("085a0d28f9a3a9f3")
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

torch.manual_seed(0)

# Construct a small tensor for `u`
# Shape: (ntrain + ntest, spatial_dim, time_steps)
# ntrain=1000, ntest=200, so we need 1200 samples
# spatial_dim should be > 1 (since we slice :-1), let's use 5
# time_steps should be > T=51, let's use 102
u = torch.randn(1200, 5, 102)

ARGS = {
    'u': u,
    'ntrain': 1000,
    'ntest': 200,
    'T': 51,
    'sub_t': 1,
    'batch_size': 20,
    'dim_x': 128,
    'dataset': None,
}
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, 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
dataloader_fno_1d_xi Ran
ran on a synthesized input
crispitagorico/neural-spdes/baselines/FNO1D.py
code served (permissive licence) · get_code("187a227e60c70db7")
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

torch.manual_seed(0)

# Create minimal test tensors
# u shape: (ntrain + ntest, spatial_dim, time_steps)
# xi shape: (ntrain + ntest, spatial_dim, time_steps)
ntrain = 2
ntest = 1
total_samples = ntrain + ntest
spatial_dim = 4
time_steps = 6

u = torch.randn(total_samples, spatial_dim, time_steps)
xi = torch.randn(total_samples, spatial_dim, time_steps)

ARGS = {
    'u': u,
    'xi': xi,
    'ntrain': ntrain,
    'ntest': ntest,
    'T': 6,
    'sub_t': 1,
    'batch_size': 2,
    'dim_x': 4,
    'dataset': None,
}
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, 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
dataloader_fno_2d_u0 Ran
ran on a synthesized input
crispitagorico/neural-spdes/baselines/FNO2D.py
code served (permissive licence) · get_code("08cf6f83101d2749")
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

torch.manual_seed(0)

# The function expects u to be a tensor with shape [num_samples, height, width, time_steps]
# We need at least ntrain + ntest samples
# Let's create a small tensor: [1202, 16, 16, 51] to accommodate ntrain=1000 and ntest=200
u = torch.randn(1202, 16, 16, 51)

ARGS = {
    'u': u,
    'ntrain': 1000,
    'ntest': 200,
    'T': 51,
    'sub_t': 1,
    'sub_x': 4,
    'batch_size': 20,
    'dataset': None
}
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, 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
dataloader_nspde_1d Ran
ran on a synthesized input
crispitagorico/neural-spdes/utilities.py
code served (permissive licence) · get_code("e04f786b5b09a1a4")
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

torch.manual_seed(0)

# Create a small tensor for u with shape (ntrain + ntest, dim_x, time_steps)
# ntrain=1000, ntest=200, dim_x=128, T=51
# We'll use smaller sizes: 10 total samples, dim_x=4, time steps=8
u = torch.randn(12, 4, 8)  # 12 samples total, 4 spatial dims, 8 time steps

# Create xi tensor with same spatial-temporal structure
xi = torch.randn(12, 4, 8)

ARGS = {
    'u': u,
    'xi': xi,
    'ntrain': 10,
    'ntest': 2,
    'T': 5,
    'sub_t': 1,
    'batch_size': 2,
    'dim_x': 4,
    'dataset': None,
}
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, 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
dataloader_nspde_2d Ran
ran on a synthesized input
crispitagorico/neural-spdes/utilities.py
code served (permissive licence) · get_code("db77f51860f997f3")
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

torch.manual_seed(0)

# Create small sample tensors
# u shape: (ntrain + ntest, spatial_x, spatial_y, time_steps)
# With ntrain=1000, ntest=200, we need 1200 total samples
# Keep spatial dims small (4x4) and time steps modest (51)
u = torch.randn(1200, 4, 4, 51)

# xi has same structure as u
xi = torch.randn(1200, 4, 4, 51)

ARGS = {
    'u': u,
    'xi': xi,
    'ntrain': 1000,
    'ntest': 200,
    'T': 51,
    'sub_t': 1,
    'sub_x': 4,
    'batch_size': 20,
    'dataset': None,
}
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, 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
eval_fno_1d Ran
ran on a synthesized input
crispitagorico/neural-spdes/baselines/FNO1D.py
code served (permissive licence) · get_code("1c6595e56a3b5ae9")
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 float, 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
import torch.nn as nn

torch.manual_seed(0)

# Define a minimal FNO-like model
class MinimalFNO1D(nn.Module):
    def __init__(self):
        super().__init__()
        self.fc = nn.Linear(4, 5)
    
    def forward(self, x):
        # x shape: (batch_size, 4)
        # output shape: (batch_size, 5, 1) to match u_pred[..., 0] indexing
        out = self.fc(x)
        return out.unsqueeze(-1)

# Define a minimal loss function
myloss = nn.MSELoss()

# Create a minimal test dataset as a list of (xi, u) tuples
# Each xi has shape (batch_size, 4), each u has shape (batch_size, 5)
test_dataset = [
    (torch.randn(2, 4), torch.randn(2, 5)),
    (torch.randn(2, 4), torch.randn(2, 5)),
]

# Create a minimal data loader wrapper that mimics torch.utils.data.DataLoader
class MinimalDataLoader:
    def __init__(self, dataset):
        self.dataset = dataset
    
    def __iter__(self):
        return iter(self.dataset)

test_dl = MinimalDataLoader(test_dataset)

# Set up parameters
model = MinimalFNO1D()
batch_size = 2
device = 'cpu'

ARGS = {
    'model': model,
    'test_dl': test_dl,
    'myloss': myloss,
    'batch_size': batch_size,
    'device': device
}
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 float, 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
eval_ncde Ran
ran on a synthesized input
crispitagorico/neural-spdes/baselines/NCDE.py
code served (permissive licence) · get_code("39f9eef0e43e9a68")
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 float, 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
import torch.nn as nn

torch.manual_seed(0)

# Define a minimal model that accepts (u0_, xi_) and returns a tensor
class MinimalModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 4)
    
    def forward(self, u0_, xi_):
        # u0_ shape: (batch, ..., features)
        # xi_ shape: (batch, ..., features)
        # Return shape should be compatible with u_ in test_dl
        batch_size = u0_.shape[0]
        return self.linear(u0_ + xi_)

# Define a minimal normalizer with decode method
class MinimalNormalizer:
    def decode(self, tensor):
        return tensor * 2.0

# Create test data loader as a list of tuples
# Each tuple: (u0_, xi_, u_)
# u0_: (batch=2, time=3, features=4)
# xi_: (batch=2, time=3, features=4)
# u_: (batch=2, time=3, features=4)
test_data = [
    (
        torch.randn(2, 3, 4),
        torch.randn(2, 3, 4),
        torch.randn(2, 3, 4),
    ),
    (
        torch.randn(2, 3, 4),
        torch.randn(2, 3, 4),
        torch.randn(2, 3, 4),
    ),
]

# Create a minimal data loader wrapper
class MinimalDataLoader:
    def __init__(self, data):
        self.data = data
        self.dataset = type('Dataset', (), {'__len__': lambda self: sum(d[0].shape[0] for d in test_data)})()
    
    def __iter__(self):
        return iter(self.data)

test_dl = MinimalDataLoader(test_data)

# Define loss function
myloss = nn.MSELoss()

# Build ARGS dict
ARGS = {
    'model': MinimalModel(),
    'test_dl': test_dl,
    'myloss': myloss,
    'batch_size': 2,
    'device': 'cpu',
    'u_normalizer': MinimalNormalizer(),
}
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 float, 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
forward_iteration Ran
ran on a synthesized input
crispitagorico/neural-spdes/torchspde/root_finding_algorithms.py
code served (permissive licence) · get_code("f786d04157be4dec")
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 dict. 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)

# Define a simple function f that takes a tensor and returns a tensor
class SimpleFunction(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 4)
    
    def forward(self, x):
        # Apply a simple transformation that converges
        return 0.9 * self.linear(x)

# Create an instance of the function
f = SimpleFunction()

# Create initial tensor x0 with shape (2, 4)
x0 = torch.randn(2, 4)

# Define ARGS dictionary with all parameters
ARGS = {
    'f': f,
    'x0': x0,
    'threshold': 50,
    'eps': 0.01
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
line_search Ran
ran; our contract draft was wrong, not the code
crispitagorico/neural-spdes/torchspde/root_finding_algorithms.py
code served (permissive licence) · get_code("2e551f7f67056cb3")
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)

# The function `line_search` uses:
# - update: a tensor, used in arithmetic operations (x0 + s * update)
# - x0: a tensor, used in arithmetic and norm operations
# - g0: a tensor, used in norm operations
# - g: a callable that takes x_est (a tensor) and returns a gradient tensor
# - nstep: an integer (not used in function body)
# - on: a boolean flag

# Create small tensors
x0 = torch.randn(4)
update = torch.randn(4)
g0 = torch.randn(4)

# Create a simple gradient function
# g should take an estimated x and return a gradient tensor
def g(x_est):
    # Simple quadratic function gradient: g = 2*x
    return 2.0 * x_est

ARGS = {
    "update": update,
    "x0": x0,
    "g0": g0,
    "g": g,
    "nstep": 0,
    "on": 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 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
scalar_search_armijo Ran
ran, honoured the contract we drafted
crispitagorico/neural-spdes/torchspde/root_finding_algorithms.py
code served (permissive licence) · get_code("b2b5022f654ee4d3")
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 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

torch.manual_seed(0)

# The function expects:
# - phi: a callable that takes a scalar alpha and returns a scalar value
# - phi0: a scalar (initial function value)
# - derphi0: a scalar (initial derivative value)
# - c1: a scalar (default 0.0001)
# - alpha0: a scalar (default 1)
# - amin: a scalar (default 0)

# Create a simple quadratic function as phi
# phi(alpha) = (alpha - 0.5)^2 + 0.1
def phi(alpha):
    return (alpha - 0.5) ** 2 + 0.1

# Evaluate at alpha=1 to get phi0
phi0 = phi(1.0)

# Derivative of phi at alpha=1: d/d_alpha[(alpha - 0.5)^2 + 0.1] = 2(alpha - 0.5)
# At alpha=1: 2(1 - 0.5) = 1.0
derphi0 = 1.0

ARGS = {
    "phi": phi,
    "phi0": phi0,
    "derphi0": derphi0,
    "c1": 0.0001,
    "alpha0": 1.0,
    "amin": 0.0,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
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
dataloader_deeponet_1d_u0 Not yet run
unverified (harvested, no recorded run)
crispitagorico/neural-spdes/baselines/deepOnet.py
code served (permissive licence) · get_code("d2a5e667ead26109")
How it ran

no invocation recorded.

dataloader_deeponet_2d_xi Not yet run
unverified (harvested, no recorded run)
crispitagorico/neural-spdes/baselines/deepOnet.py
code served (permissive licence) · get_code("4bd6cbbe9a02195d")
How it ran

no invocation recorded.

dataloader_fno_2d_xi Not yet run
unverified (harvested, no recorded run)
crispitagorico/neural-spdes/baselines/FNO2D.py
code served (permissive licence) · get_code("0a4f8df0cdd397a5")
How it ran

no invocation recorded.

eval_ncdeinf_1d Not yet run
unverified (harvested, no recorded run)
crispitagorico/neural-spdes/baselines/NCDEFNO_1D.py
code served (permissive licence) · get_code("e3e8efa5e8efb67c")
How it ran

no invocation recorded.

eval_nrde_1d Not yet run
unverified (harvested, no recorded run)
crispitagorico/neural-spdes/baselines/NRDE.py
code served (permissive licence) · get_code("669c9bae6c4e3862")
How it ran

no invocation recorded.

eval_nspde Not yet run
unverified (harvested, no recorded run)
crispitagorico/neural-spdes/utilities.py
code served (permissive licence) · get_code("0eaa5e38d1dc5ea8")
How it ran

no invocation recorded.

grad_space Not yet run
unverified (harvested, no recorded run)
crispitagorico/torchspde/torchspde/gradients.py
code served (permissive licence) · get_code("d9a8a84a3606f245")
How it ran

no invocation recorded.

grad_space_perp Not yet run
unverified (harvested, no recorded run)
crispitagorico/torchspde/torchspde/gradients.py
code served (permissive licence) · get_code("0389361f6c547465")
How it ran

no invocation recorded.

grad_var Not yet run
unverified (harvested, no recorded run)
crispitagorico/torchspde/torchspde/gradients.py
code served (permissive licence) · get_code("3bfc2050cef8f89f")
How it ran

no invocation recorded.

train_fno_2d Not yet run
unverified (harvested, no recorded run)
crispitagorico/neural-spdes/baselines/FNO2D.py
code served (permissive licence) · get_code("edc4e5ae89c52d98")
How it ran

no invocation recorded.

train_ncdeinf_2d Not yet run
unverified (harvested, no recorded run)
crispitagorico/neural-spdes/baselines/NCDEFNO_2D.py
code served (permissive licence) · get_code("b4119fee460e85ef")
How it ran

no invocation recorded.

Also present in other papers' code

Of 29 samples lifted from this paper's linked repositories, 29 carry a function identity and 0 do not. Of the 29 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 27 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
scalar_search_armijo 13 other papers
4 dated earlier, 9 later · 4 only as an assembly brick
2006.08656 (2020-06)
Multiscale Deep Equilibrium Models
2106.00553 (2021-06)
SHINE: SHaring the INverse Estimate from the forward pass for bi-level optimization and implicit models
2106.08748 (2021-06)
Input Invex Neural Network
and 10 more not named here
line_search 11 other papers
3 dated earlier, 8 later · 3 only as an assembly brick
2006.08656 (2020-06)
Multiscale Deep Equilibrium Models
2106.08748 (2021-06)
Input Invex Neural Network
2109.14247 (2021-09)
Training Feedback Spiking Neural Networks by Implicit Differentiation on the Equilibrium State
and 8 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

Stochastic partial differential equations (SPDEs) are the mathematical tool of choice for modelling spatiotemporal PDE-dynamics under the influence of randomness. Based on the notion of mild solution of an SPDE, we introduce a novel neural architecture to learn solution operators of PDEs with (possibly stochastic) forcing from partially observed data. The proposed Neural SPDE model provides an extension to two popular classes of physics-inspired architectures. On the one hand, it extends Neural CDEs and variants -continuous-time analogues of RNNs -in that it is capable of processing incoming sequential information arriving at arbitrary spatial resolutions. On the other hand, it extends Neural Operators -generalizations of neural networks to model mappings between spaces of functions -in that it can parameterize solution operators of SPDEs depending simultaneously on the initial condition and a realization of the driving noise. By performing operations in the spectral domain, we show how a Neural SPDE can be evaluated in two ways, either by calling an ODE solver (emulating a spectral Galerkin scheme), or by solving a fixed point problem. Experiments on various semilinear SPDEs, including the stochastic Navier-Stokes equations, demonstrate how the Neural SPDE model is capable of learning complex spatiotemporal dynamics in a resolution-invariant way, with better accuracy and lighter training data requirements compared to alternative models, and up to 3 orders of magnitude faster than traditional solvers.

For agents

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

get_harvested_code_for_paper("2110.10249")
get_code_for_paper("2110.10249")
have("2110.10249")

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

curl https://syntology.ai/api/ran/2110.10249.json

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

[![Syntology run record](https://syntology.ai/api/ran/2110.10249.svg)](https://syntology.ai/paper/2110.10249)

Connect an agent — have() is free.