SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2205.13503 · NeurIPS · 2022

Multi-layer State Evolution Under Random Convolutional Design

Florent Krzakala, Lenka Zdeborová, Cédric Gerbelot, Mara Daniels, = F T -→ E, Xt ← - E, L F T -→, L Xt -→ E L-1, = Q⊤ -→ E, Mt ← - E 1 -Mt-

arXiv · PDF · Open in the Atlas

Code that ran

8 of 11 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 8 have no contract verdict; 0 failures of our instrument; 3 no recorded run), measured 2026-11-22; 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/2205.13503.json · badge: /api/ran/2205.13503.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
mdnls/conv-ml-amp canonical (classifier label)
classifier basis not recorded on the link
8 ran · 3 no recorded run
FunctionStatusWhere it lives
GET_DEVICE Ran
ran on a synthesized input
mdnls/conv-ml-amp/arr.py
code served (permissive licence) · get_code("b44440bce4c22599")
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 str. 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 os
import torch

# Construct ARGS dictionary with all parameters for GET_DEVICE
ARGS = {
    'device': 'default'
}
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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
GET_NUMERICAL_LIB Ran
ran on a synthesized input
mdnls/conv-ml-amp/arr.py
code served (permissive licence) · get_code("42d881c07c80dd4d")
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 module. 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 os
import numpy

# Clear any environment variable that might affect the function
if 'MLAMP_NO_CUPY' in os.environ:
    del os.environ['MLAMP_NO_CUPY']

# Construct arguments for GET_NUMERICAL_LIB
ARGS = {
    'device': 'default'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
cupy, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type module — a module output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a module 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
fast_1d_T_multiconv Ran
ran on a synthesized input
mdnls/conv-ml-amp/model/conv.py
code served (permissive licence) · get_code("3a8d1de6929d966f")
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 (3, 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

# Set seed for reproducibility
np.random.seed(0)

# Construct arguments for fast_1d_T_multiconv
# Parameters: z (D, q), conv_filter (D, P, k), q (scalar)

D = 2  # number of input channels
P = 3  # number of output channels
k = 5  # kernel size
q = 4  # sequence length

# z: shape (D, q) - input tensor
z = np.random.randn(D, q).astype(np.float64)

# conv_filter: shape (D, P, k) - convolution filter
conv_filter = np.random.randn(D, P, k).astype(np.float64)

# q: scalar - sequence length (same as dimension in z)
q_param = q

ARGS = {
    "z": z,
    "conv_filter": conv_filter,
    "q": q_param
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numba, 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 (3, 4) — shape (3, 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
fast_1d_multiconv Ran
ran on a synthesized input
mdnls/conv-ml-amp/model/conv.py
code served (permissive licence) · get_code("c9bcc8c31b34a966")
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 (3, 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

# Analyze the function:
# - z: 2D array, indexed as z[j, (l + s - (k - 1) // 2) % q]
#   shape (P, q) where P is the number of "channels" and q is the sequence length
# - conv_filter: 3D array with shape (D, P, k)
#   D = output dimension, P = input channels, k = kernel size
# - q: scalar, sequence length
# - result: output shape (D, q)

# Create small distinct dimensions:
# D = 3 (output channels)
# P = 2 (input channels)
# k = 5 (kernel size)
# q = 4 (sequence length)

D = 3
P = 2
k = 5
q = 4

# z: shape (P, q) = (2, 4)
z = np.random.randn(P, q).astype(np.float64)

# conv_filter: shape (D, P, k) = (3, 2, 5)
conv_filter = np.random.randn(D, P, k).astype(np.float64)

ARGS = {
    "z": z,
    "conv_filter": conv_filter,
    "q": q,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numba, 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 (3, 4) — shape (3, 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
gaussian_measure_2d Ran
ran on a synthesized input
mdnls/conv-ml-amp/model/piecewise_linear.py
code served (permissive licence) · get_code("56dab5e3d7b454d7")
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 numpy as np
import scipy.stats
from scipy.integrate import quad, dblquad

torch.manual_seed(0)

# The function gaussian_measure_2d takes:
# - m1: mean for first dimension (scalar)
# - s1: standard deviation for first dimension (scalar)
# - m2: mean for second dimension (scalar)
# - s2: standard deviation for second dimension (scalar)
# - f: a callable that takes two arguments and returns a scalar

# Create a simple callable function that takes two arguments
def test_function(x1, x2):
    """Simple test function that takes two scalar/array arguments"""
    return np.sin(x1) * np.cos(x2)

ARGS = {
    "m1": 0.5,
    "s1": 1.0,
    "m2": -0.3,
    "s2": 0.8,
    "f": test_function
}
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 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
norm_cdf Ran
ran on a synthesized input
mdnls/conv-ml-amp/model/piecewise_linear.py
code served (permissive licence) · get_code("d435e3b6197cfdae")
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 torch
import numpy as np
import scipy.stats

torch.manual_seed(0)

# The function norm_cdf takes a single parameter x and passes it to scipy.stats.norm.cdf
# scipy.stats.norm.cdf accepts array-like inputs (scalars, lists, numpy arrays, torch tensors)
# We'll provide a small numpy array as input

x = np.array([0.0, 1.0, -1.0, 2.0])

ARGS = {
    "x": x
}
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 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
norm_pdf Ran
ran on a synthesized input
mdnls/conv-ml-amp/model/piecewise_linear.py
code served (permissive licence) · get_code("d1c92e9c7050a5c0")
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 torch
import numpy as np
import scipy.stats

torch.manual_seed(0)

# The function norm_pdf takes a single parameter x and computes scipy.stats.norm.pdf(x)
# x can be a scalar, array-like, or tensor
# We'll provide a numpy array as the argument

x = np.array([0.0, 1.0, -1.0, 0.5])

ARGS = {
    'x': x
}
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 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
sigmoid Ran
ran on a synthesized input
mdnls/conv-ml-amp/model/factor.py
code served (permissive licence) · get_code("d1c0ec1834540e2e")
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
from scipy.special import softmax, expit

def sigmoid(x):
    return expit(x)

# Test with various input types that sigmoid can handle
x_numpy = np.array([0.0, 1.0, -1.0, 0.5])
x_list = [0.0, 1.0, -1.0, 0.5]
x_scalar = 0.5

ARGS = {
    "x": x_numpy
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — 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
C Not yet run
unverified (harvested, no recorded run)
mdnls/conv-ml-amp/model/factor.py
code served (permissive licence) · get_code("e5dbc61853970ade")
How it ran

no invocation recorded.

N Not yet run
unverified (harvested, no recorded run)
mdnls/conv-ml-amp/model/factor.py
code served (permissive licence) · get_code("c53f52f22d452998")
How it ran

no invocation recorded.

to_cpu Not yet run
unverified (harvested, no recorded run)
mdnls/conv-ml-amp/arr.py
code served (permissive licence) · get_code("5bf114c8a2834815")
How it ran

no invocation recorded.

Also present in other papers' code

Of 11 samples lifted from this paper's linked repositories, 11 carry a function identity and 0 do not. Of the 11 distinct functions with an identity, 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 11 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

Signal recovery under generative neural network priors has emerged as a promising direction in statistical inference and computational imaging. Theoretical analysis of reconstruction algorithms under generative priors is, however, challenging. For generative priors with fully connected layers and Gaussian i.i.d. weights, this was achieved by the multi-layer approximate message (ML-AMP) algorithm via a rigorous state evolution. However, practical generative priors are typically convolutional, allowing for computational benefits and inductive biases, and so the Gaussian i.i.d. weight assumption is very limiting. In this paper, we overcome this limitation and establish the state evolution of ML-AMP for random convolutional layers. We prove in particular that random convolutional layers belong to the same universality class as Gaussian matrices. Our proof technique is of an independent interest as it establishes a mapping between convolutional matrices and spatially coupled sensing matrices used in coding theory.

For agents

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

get_harvested_code_for_paper("2205.13503")
get_code_for_paper("2205.13503")
have("2205.13503")

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

curl https://syntology.ai/api/ran/2205.13503.json

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

[![Syntology run record](https://syntology.ai/api/ran/2205.13503.svg)](https://syntology.ai/paper/2205.13503)

Connect an agent — have() is free.