SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2309.17082 · 2023

Diffusion Models as Stochastic Quantization in Lattice Field Theory

arXiv · PDF · Open in the Atlas

Code that ran

13 of 15 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; 6 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 7 failures of our instrument; 2 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/2309.17082.json · badge: /api/ran/2309.17082.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
anguswlx/dmassq canonical (classifier label)
classifier basis: census
13 ran · 2 no recorded run
FunctionStatusWhere it lives
Euler_Maruyama_sampler Ran
ran; our contract draft was wrong, not the code
anguswlx/dmassq/sampler.py
code served (permissive licence) · get_code("98e8305408831b32")
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
import torch.nn as nn

torch.manual_seed(0)

# Define a minimal score model that accepts (x, t) and returns a tensor
class ScoreModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 4)
    
    def forward(self, x, t):
        # x shape: (batch_size, 1, L, L)
        # t shape: (batch_size,)
        # Return shape should match x for the score
        batch_size = x.shape[0]
        # Just return a tensor of same shape as x
        return torch.zeros_like(x)

# Define marginal_prob_std function
def marginal_prob_std(t):
    """Returns standard deviation for time t.
    
    Args:
        t: tensor of shape (batch_size,)
    
    Returns:
        tensor of shape (batch_size,) with std values
    """
    return torch.ones_like(t) * 0.5

# Define diffusion_coeff function
def diffusion_coeff(t):
    """Returns diffusion coefficient for time t.
    
    Args:
        t: tensor of shape (batch_size,)
    
    Returns:
        tensor of shape (batch_size,) with diffusion values
    """
    return torch.ones_like(t) * 0.3

score_model = ScoreModel()

ARGS = {
    'score_model': score_model,
    'marginal_prob_std': marginal_prob_std,
    'diffusion_coeff': diffusion_coeff,
    'batch_size': 2,
    'num_steps': 5,
    'device': 'cpu',
    'eps': 0.001,
    'L': 4
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, tqdm — 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
action Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
anguswlx/dmassq/mc.py
code served (permissive licence) · get_code("295164393b62f041")
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 (2,) 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

# Construct arguments for the action function
# The function takes:
# - phi: np.ndarray (appears to be 3D based on axis=(1,2) and np.roll operations)
# - k: scalar or array-like parameter
# - l: scalar or array-like parameter

# Based on the function body:
# - phi is rolled along axes 0, 1 suggesting it's at least 2D
# - axis=(1,2) in np.sum suggests phi is 3D
# - The operations are element-wise, so k and l are likely scalars

# Create a small 3D numpy array for phi (batch_size=2, height=4, width=3)
phi = np.random.randn(2, 4, 3).astype(np.float32)

# k and l are scalar parameters used in the computation
k = 0.5
l = 0.3

ARGS = {
    "phi": phi,
    "k": k,
    "l": l
}
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 (2,) — shape (2,) 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
diffusion_coeff Ran
ran; our contract draft was wrong, not the code
anguswlx/dmassq/sde.py
code served (permissive licence) · get_code("c1965c68ebd4e36a")
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 (2,) 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)

# Analyze the function signature and body:
# - t: A vector of time steps (used in exponentiation t ** t)
# - sigma: The sigma parameter (used in exponentiation sigma ** t)
# - device: A torch device (passed to torch.tensor)

# The function computes sigma ** t element-wise and creates a tensor on the device

t = torch.tensor([0.1, 0.5], device='cpu')
sigma = 2.0
device = 'cpu'

ARGS = {
    't': t,
    'sigma': sigma,
    '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 Tensor, ndim 1, dtype family float, shape (2,) — shape (2,) 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_abs_mag Ran
ran, honoured the contract we drafted
anguswlx/dmassq/observable.py
code served (permissive licence) · get_code("b010a164fe10282d")
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 numpy as np

# Read the function to understand what it needs:
# get_abs_mag takes cfgs: np.ndarray
# It computes the mean along all axes except the first (samples axis)
# Then calls jackknife on those means
# jackknife expects a 2D array where axis 0 is samples

# cfgs should be a multi-dimensional array where:
# - axis 0 represents different samples/configurations
# - axes 1+ represent spatial or feature dimensions
# After mean(axis=axis) where axis is all dims except 0, we get a 1D array per sample
# jackknife then treats these as samples

# Create a small 3D array: (samples=5, spatial_dim1=3, spatial_dim2=4)
# This will result in shape (5,) after the mean operation
cfgs = np.random.randn(5, 3, 4)

ARGS = {
    "cfgs": cfgs
}
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 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
get_action Ran
ran, honoured the contract we drafted
anguswlx/dmassq/mc.py
code served (permissive licence) · get_code("3e3c5437e484e594")
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 float32, 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 numpy as np

# Analyze the function:
# - phi: used with np.roll, np.sum, arithmetic operations -> needs to be a numpy array
# - k: used in multiplication with phi -> scalar (int or float)
# - l: used in arithmetic operations -> scalar (int or float)

# Create a small 2D numpy array for phi (2x4 to have distinct dimensions)
phi = np.random.randn(2, 4).astype(np.float32)

# Create scalar values for k and l
k = 0.5
l = 0.3

ARGS = {
    "phi": phi,
    "k": k,
    "l": l
}
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 float32, 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
get_drift Ran
ran, honoured the contract we drafted
anguswlx/dmassq/mc.py
code served (permissive licence) · get_code("11ac00a92305076c")
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

# The function get_drift takes three parameters:
# - phi: a numpy array that supports np.roll operations (2D array based on usage)
# - k: a scalar multiplier
# - l: a scalar parameter used in the computation

# Create a small 2D numpy array for phi (4x3 to have distinct dimensions)
phi = np.random.randn(4, 3)

# Create scalar values for k and l
k = 0.5
l = 0.3

ARGS = {
    'phi': phi,
    'k': k,
    'l': l
}
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
get_mag Ran
ran, honoured the contract we drafted
anguswlx/dmassq/observable.py
code served (permissive licence) · get_code("2fb8f7549fd6da95")
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 numpy as np

# Create sample configuration data
# cfgs shape: (num_samples, spatial_dim_1, spatial_dim_2, ...)
# The function will compute mean over all spatial dimensions for each sample,
# then jackknife over the samples
np.random.seed(0)

# Create a 3D array: (num_samples=5, height=4, width=3)
# This represents 5 configuration samples, each with a 4x3 spatial grid
cfgs = np.random.randn(5, 4, 3)

ARGS = {
    "cfgs": cfgs
}
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 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
grab Ran
ran; our contract draft was wrong, not the code
anguswlx/dmassq/plot_sample.py
code served (permissive licence) · get_code("a7d4fba206cac9de")
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 (2, 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 torch
import numpy as np

torch.manual_seed(0)

# Create a small tensor on CPU
var = torch.randn(2, 4)

ARGS = {
    'var': var
}
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 ndarray, ndim 2, dtype family float, shape (2, 4) — shape (2, 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
jackknife Ran
ran, honoured the contract we drafted
anguswlx/dmassq/observable.py
code served (permissive licence) · get_code("3eb7d08565f61611")
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 2, 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 numpy as np

# The function jackknife takes samples as a numpy array
# It performs leave-one-out jackknife resampling
# The array is indexed with axis=0 and axis=1, and operations are done on axis=0
# So we need a 2D array with shape (n_samples, n_features)

# Create a small sample array: 5 samples, 3 features
np.random.seed(0)
samples = np.random.randn(5, 3).astype(np.float32)

ARGS = {
    "samples": samples
}
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 tuple, ndim 2, 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
loss_fn Ran
ran; our contract draft was wrong, not the code
anguswlx/dmassq/loss.py
code served (permissive licence) · get_code("73ec3b826b113561")
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
import torch.nn as nn

torch.manual_seed(0)

# Create a minimal model that takes perturbed_x and time t as inputs
class ScoreModel(nn.Module):
    def __init__(self):
        super().__init__()
        # Model expects input shape (batch, channels, height, width)
        # and time t shape (batch,)
        self.linear = nn.Linear(4 * 4 * 4, 4 * 4 * 4)
    
    def forward(self, x, t):
        # x shape: (batch, 4, 4, 4)
        # t shape: (batch,)
        batch_size = x.shape[0]
        x_flat = x.view(batch_size, -1)
        score = self.linear(x_flat)
        score = score.view(batch_size, 4, 4, 4)
        return score

# Create the model
model = ScoreModel()

# Create training data: batch_size=2, channels=4, height=4, width=4
x = torch.randn(2, 4, 4, 4)

# Create marginal_prob_std function that takes time t and returns std
def marginal_prob_std(t):
    # t shape: (batch,)
    # returns shape: (batch,)
    return torch.sqrt(t)

ARGS = {
    "model": model,
    "x": x,
    "marginal_prob_std": marginal_prob_std,
    "eps": 1e-05
}
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
marginal_prob_std Ran
ran; our contract draft was wrong, not the code
anguswlx/dmassq/sde.py
code served (permissive licence) · get_code("725b17e24eba5db7")
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 (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
import numpy as np

torch.manual_seed(0)

# Analyze function usage:
# - t: used with torch.tensor() and passed to operations, should be array-like
# - sigma: used in arithmetic operations (** operator), should be a scalar float
# - device: passed to torch.tensor(), should be a torch.device object

ARGS = {
    "t": [0.1, 0.5, 0.9],
    "sigma": 25.0,
    "device": torch.device("cpu")
}
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 (3,) — shape (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
ode_likelihood Ran
ran; our contract draft was wrong, not the code
anguswlx/dmassq/likelihood.py
code served (permissive licence) · get_code("979ca6bfb895133d")
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
import torch.nn as nn
import numpy as np

torch.manual_seed(0)

# Define a minimal score model that satisfies the function's usage
class ScoreModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 4)
    
    def forward(self, x, t):
        # x shape: (batch, 1, 2, 2) for a 4D tensor
        # t shape: (batch,)
        # Returns score with same shape as x
        batch_size = x.shape[0]
        x_flat = x.reshape(batch_size, -1)
        score_flat = self.linear(x_flat)
        score = score_flat.reshape(x.shape)
        return score

# Define marginal_prob_std function
def marginal_prob_std(t):
    """Returns standard deviation as a tensor."""
    return torch.tensor(0.1 + 0.9 * t, dtype=torch.float32)

# Define diffusion_coeff function
def diffusion_coeff(t):
    """Returns diffusion coefficient as a tensor."""
    return torch.tensor(np.sqrt(2.0 * t), dtype=torch.float32)

# Create input tensor: batch_size=2, spatial dims 2x2, channels 1 -> (2, 1, 2, 2)
x = torch.randn(2, 1, 2, 2, dtype=torch.float32)

# Instantiate score model
score_model = ScoreModel()

# Construct ARGS dictionary
ARGS = {
    'x': x,
    'score_model': score_model,
    'marginal_prob_std': marginal_prob_std,
    'diffusion_coeff': diffusion_coeff,
    'batch_size': 2,
    'device': 'cpu',
    'eps': 1e-05
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy, 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
pc_sampler Ran
ran, honoured the contract we drafted
anguswlx/dmassq/sampler.py
code served (permissive licence) · get_code("912afada37c424b9")
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, 1, 4, 4) as a later run recorded it. 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, 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
import numpy as np

torch.manual_seed(0)

# Define a minimal score model that takes (x, t) and returns a score
class ScoreModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4 * 4, 4 * 4)
    
    def forward(self, x, t):
        # x shape: (batch_size, 1, L, L)
        # t shape: (batch_size,)
        batch_size = x.shape[0]
        x_flat = x.reshape(batch_size, -1)
        return self.linear(x_flat).reshape(batch_size, 1, 4, 4)

score_model = ScoreModel()

# Define marginal_prob_std function
def marginal_prob_std(t):
    # t shape: (batch_size,)
    # returns shape: (batch_size,)
    return torch.ones_like(t) * 0.5

# Define diffusion_coeff function
def diffusion_coeff(t):
    # t shape: (batch_size,)
    # returns shape: (batch_size,)
    return torch.ones_like(t) * 0.3

ARGS = {
    'score_model': score_model,
    'marginal_prob_std': marginal_prob_std,
    'diffusion_coeff': diffusion_coeff,
    'batch_size': 2,
    'num_steps': 5,
    'snr': 0.16,
    'device': 'cpu',
    'eps': 0.001,
    'L': 4
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch, tqdm — 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, 1, 4, 4) — shape (2, 1, 4, 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
ode_sampler Not yet run
unverified (harvested, no recorded run)
anguswlx/dmassq/sampler.py
code served (permissive licence) · get_code("87b6c1b754a69d7f")
How it ran

no invocation recorded.

prior_likelihood Not yet run
unverified (harvested, no recorded run)
anguswlx/dmassq/likelihood.py
code served (permissive licence) · get_code("5d1e48c8db4f7dcd")
How it ran

no invocation recorded.

Also present in other papers' code

Of 15 samples lifted from this paper's linked repositories, 15 carry a function identity and 0 do not. Of the 15 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 15 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

In this work, we establish a direct connection between generative diffusion models (DMs) and stochastic quantization (SQ). The DM is realized by approximating the reversal of a stochastic process dictated by the Langevin equation, generating samples from a prior distribution to effectively mimic the target distribution. Using numerical simulations, we demonstrate that the DM can serve as a global sampler for generating quantum lattice field configurations in two-dimensional $φ^4$ theory. We demonstrate that DMs can notably reduce autocorrelation times in the Markov chain, especially in the critical region where standard Markov Chain Monte-Carlo (MCMC) algorithms experience critical slowing down. The findings can potentially inspire further advancements in lattice field theory simulations, in particular in cases where it is expensive to generate large ensembles.

For agents

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

get_harvested_code_for_paper("2309.17082")
get_code_for_paper("2309.17082")
have("2309.17082")

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

curl https://syntology.ai/api/ran/2309.17082.json

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

[![Syntology run record](https://syntology.ai/api/ran/2309.17082.svg)](https://syntology.ai/paper/2309.17082)

Connect an agent — have() is free.