SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2304.03376 · 2023

Interpretable statistical representations of neural population dynamics and geometry

arXiv · PDF · Open in the Atlas

Code that ran

8 of 13 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 7 have no contract verdict; 1 failure of our instrument; 5 no recorded run), measured 2026-11-21; 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/2304.03376.json · badge: /api/ran/2304.03376.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
Dynamics-of-Neural-Systems-Lab/MARBLE canonical (classifier label)
classifier basis not recorded on the link
8 ran · 5 no recorded run
FunctionStatusWhere it lives
cknneighbors_graph Ran
ran on a synthesized input
Dynamics-of-Neural-Systems-Lab/MARBLE/MARBLE/lib/cknn.py
code served (permissive licence) · get_code("db40332fd5c64652")
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 csr_matrix. 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 numpy as np

# Create a small sample dataset
# Using distinct dimensions: 5 samples, 3 features
np.random.seed(0)
X = np.random.randn(5, 3)

ARGS = {
    'X': X,
    'n_neighbors': 2,
    'delta': 1.0,
    'metric': 'euclidean',
    't': 'inf',
    'include_self': False,
    'is_sparse': True,
    'return_instance': False,
}
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 csr_matrix — a csr_matrix output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a csr_matrix 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
cluster Ran
ran on a synthesized input
Dynamics-of-Neural-Systems-Lab/MARBLE/MARBLE/geometry.py
code served (permissive licence) · get_code("741d528192ec3611")
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 numpy as np
from sklearn.cluster import KMeans, MeanShift

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

# Create a small sample dataset (n=10 samples, dim=4 features)
x = np.random.randn(10, 4).astype(np.float32)

# Define arguments for the cluster function
ARGS = {
    'x': x,
    'cluster_typ': 'kmeans',
    'n_clusters': 3,
    'seed': 0
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn — 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
detach_from_gpu Ran
ran on a synthesized input
Dynamics-of-Neural-Systems-Lab/MARBLE/MARBLE/utils.py
code served (permissive licence) · get_code("2777663c0f2f4d47")
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)

# Create a minimal model
class MinimalModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 2)
    
    def forward(self, x):
        return self.linear(x)

# Create a data object with all required attributes
class Data:
    def __init__(self):
        self.x = torch.randn(2, 4)
        self.pos = torch.randn(2, 3)
        self.mask = torch.randint(0, 2, (2,)).bool()
        self.L = [torch.randn(2, 2), torch.randn(2, 2)]
        self.Lc = [torch.randn(2, 2), torch.randn(2, 2)]
        self.kernels = [torch.randn(2, 2), torch.randn(3, 3)]
        self.gauges = torch.randn(2, 5)

# Create adjacency list (list of [edge_index_0, edge_index_1, edge_attr])
adjs = [
    [torch.tensor([[0, 1], [1, 0]]), torch.tensor([[0, 1], [1, 0]]), torch.randn(2, 4)],
    [torch.tensor([[0, 1, 1], [1, 0, 1]]), torch.tensor([[0, 1, 1], [1, 0, 1]]), torch.randn(3, 4)]
]

model = MinimalModel()
data = Data()

ARGS = {
    'model': model,
    'data': data,
    'adjs': adjs
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
fun_vanderpol Ran
ran on a synthesized input
Dynamics-of-Neural-Systems-Lab/MARBLE/MARBLE/dynamics.py
code served (permissive licence) · get_code("e79cd1c9c93a4239")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

ARGS = {
    'par': {'mu': 1.0}
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
graph Ran
ran; our contract draft was wrong, not the code
Dynamics-of-Neural-Systems-Lab/MARBLE/MARBLE/plotting.py
code served (permissive licence) · get_code("065d0cb22322a697")
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 Axes. 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 matplotlib.pyplot as plt
import networkx as nx
import numpy as np

# Create a simple graph with 2D positions
G = nx.Graph()
G.add_node(0, pos=(0.0, 0.0))
G.add_node(1, pos=(1.0, 0.5))
G.add_node(2, pos=(0.5, 1.0))
G.add_edge(0, 1)
G.add_edge(1, 2)
G.add_edge(2, 0)

# Create labels (node colors) as array
labels = np.array([0.1, 0.5, 0.9])

# Create figure and axis for 2D plot
fig = plt.figure()
ax = fig.add_subplot(1, 1, 1)

ARGS = {
    'G': G,
    'labels': labels,
    'edge_width': 1.5,
    'edge_alpha': 0.7,
    'node_size': 100,
    'layout': 'spectral',
    'ax': ax,
    'axes_visible': True
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
matplotlib, networkx, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Axes — a Axes output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Axes 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
move_to_gpu Ran
ran on a synthesized input
Dynamics-of-Neural-Systems-Lab/MARBLE/MARBLE/utils.py
code served (permissive licence) · get_code("f3cbfb34cd6aeb7a")
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)

# Create a minimal model
class DummyModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 4)
    
    def forward(self, x):
        return self.linear(x)

# Create a data object with all required attributes
class DummyData:
    def __init__(self):
        self.x = torch.randn(2, 4)
        self.pos = torch.randn(2, 3)
        self.mask = torch.ones(2, dtype=torch.bool)
        self.L = [torch.randn(4, 4), torch.randn(4, 4)]
        self.Lc = [torch.randn(4, 4), torch.randn(4, 4)]
        self.kernels = [torch.randn(4, 5), torch.randn(4, 5)]
        self.gauges = torch.randn(2, 6)

model = DummyModel()
data = DummyData()
adjs = [torch.randn(4, 4), torch.randn(4, 4)]

ARGS = {
    'model': model,
    'data': data,
    'adjs': adjs
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
scalar_diffusion Ran
ran on a synthesized input
Dynamics-of-Neural-Systems-Lab/MARBLE/MARBLE/smoothing.py
code served (permissive licence) · get_code("99624cc21afcc509")
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 2, dtype family float, shape (4, 1) 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)

# For matrix_exp method: x is a 1D or 2D tensor, t is a scalar or tensor, par is a sparse/dense matrix
x_1d = torch.randn(4)  # Will be unsqueezed to (4, 1)
x_2d = torch.randn(4, 2)  # Already 2D

# Create a parameter matrix (4x4 for matrix multiplication with x)
par_dense = torch.randn(4, 4)

# For sparse representation, we can use a sparse tensor
par_sparse = torch.sparse_coo_tensor(
    indices=torch.tensor([[0, 1, 2, 3], [0, 1, 2, 3]]),
    values=torch.tensor([1.0, 2.0, 3.0, 4.0]),
    size=(4, 4)
)

# Time parameter
t_scalar = 0.5

# For spectral method: par should be (eigenvalues, eigenvectors)
evals = torch.randn(4)
evecs = torch.randn(4, 4)
par_spectral = (evals, evecs)

# Build ARGS for matrix_exp method with 1D x
ARGS = {
    'x': x_1d,
    't': t_scalar,
    'method': 'matrix_exp',
    'par': par_dense
}
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 2, dtype family float, shape (4, 1) — shape (4, 1) 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
solve_ODE Ran
ran on a synthesized input
Dynamics-of-Neural-Systems-Lab/MARBLE/MARBLE/dynamics.py
code served (permissive licence) · get_code("e7820c4724a4dcce")
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 3, 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
from scipy.integrate import ode, odeint

# Define a simple ODE system: dx/dt = -x
def f(t, x):
    """Time derivative function."""
    return -x

def jac(t, x):
    """Jacobian function."""
    return -np.eye(len(x))

# Initial condition
x0 = np.array([1.0, 2.0, 3.0, 4.0])

# Time points
t = np.linspace(0, 1, 5)

ARGS = {
    'f': f,
    'jac': jac,
    't': t,
    'x0': x0,
    'solver': 'standard'
}
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 tuple, ndim 3, 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
embedding Not yet run
unverified (harvested, no recorded run)
Dynamics-of-Neural-Systems-Lab/MARBLE/MARBLE/plotting.py
code served (permissive licence) · get_code("1ff583e162ae57a6")
How it ran

no invocation recorded.

fields Not yet run
unverified (harvested, no recorded run)
Dynamics-of-Neural-Systems-Lab/MARBLE/MARBLE/plotting.py
code served (permissive licence) · get_code("9725f9cdeacc9e0e")
How it ran

no invocation recorded.

load_ODE Not yet run
unverified (harvested, no recorded run)
Dynamics-of-Neural-Systems-Lab/MARBLE/MARBLE/dynamics.py
code served (permissive licence) · get_code("378cf030e47fffde")
How it ran

no invocation recorded.

parallel_proc Not yet run
unverified (harvested, no recorded run)
Dynamics-of-Neural-Systems-Lab/MARBLE/MARBLE/utils.py
code served (permissive licence) · get_code("bfc0e1a045078e27")
How it ran

no invocation recorded.

vector_diffusion Not yet run
unverified (harvested, no recorded run)
Dynamics-of-Neural-Systems-Lab/MARBLE/MARBLE/smoothing.py
code served (permissive licence) · get_code("a2caaa7f7f6ceeb8")
How it ran

no invocation recorded.

Also present in other papers' code

Of 13 samples lifted from this paper's linked repositories, 13 carry a function identity and 0 do not. Of the 13 distinct functions with an identity, 1 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 1 later) and 12 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
graph 1 other paper
1 later
2309.16746 (2023-09)
Implicit Gaussian process representation of vector fields over arbitrary latent manifolds

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 2026-11-21.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

The dynamics of neuron populations commonly evolve on low-dimensional manifolds. Thus, we need methods that learn the dynamical processes over neural manifolds to infer interpretable and consistent latent representations. We introduce a representation learning method, MARBLE, that decomposes on-manifold dynamics into local flow fields and maps them into a common latent space using unsupervised geometric deep learning. In simulated non-linear dynamical systems, recurrent neural networks, and experimental single-neuron recordings from primates and rodents, we discover emergent low-dimensional latent representations that parametrise high-dimensional neural dynamics during gain modulation, decision-making, and changes in the internal state. These representations are consistent across neural networks and animals, enabling the robust comparison of cognitive computations. Extensive benchmarking demonstrates state-of-the-art within- and across-animal decoding accuracy of MARBLE compared with current representation learning approaches, with minimal user input. Our results suggest that manifold structure provides a powerful inductive bias to develop powerful decoding algorithms and assimilate data across experiments.

For agents

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

get_harvested_code_for_paper("2304.03376")
get_code_for_paper("2304.03376")
have("2304.03376")

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

curl https://syntology.ai/api/ran/2304.03376.json

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

[![Syntology run record](https://syntology.ai/api/ran/2304.03376.svg)](https://syntology.ai/paper/2304.03376)

Connect an agent — have() is free.