SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2403.14843 · 2024

Local Causal Discovery with Linear non-Gaussian Cyclic Models

arXiv · PDF · Open in the Atlas

Code that ran

15 of 18 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; 4 honoured the contract we drafted, 2 violated it and 0 have no contract verdict; 9 failures of our instrument; 3 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2403.14843.json · badge: /api/ran/2403.14843.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
markdana/local-ling-discovery canonical (classifier label)
classifier basis: census
15 ran · 3 no recorded run
FunctionStatusWhere it lives
acyclic_sink_processing Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
markdana/local-ling-discovery/local_isa_ling/acyclic_sink_processing.py
code served (permissive licence) · get_code("04038ee5ba1c7542")
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, 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 to understand parameter requirements:
# - W: a 2D numpy array (matrix) that gets copied and manipulated
# - target: an integer index (row index to target)
# - threshold: a float used for thresholding absolute values

# Create a small test case
# W should be a square matrix where acyclic_sink_processing can process it
# The function processes a DAG (directed acyclic graph) represented as a matrix

np.random.seed(42)

# Create a small 4x4 matrix with some structure
# This should represent a system where we can find coefficients
W = np.array([
    [0.0, 0.5, 0.2, 0.0],
    [0.3, 0.0, 0.0, 0.4],
    [0.0, 0.0, 0.0, 0.1],
    [0.2, 0.0, 0.3, 0.0]
], dtype=np.float64)

# target should be a valid row index in the matrix
target = 2

# threshold should be small enough to not filter out all values
threshold = 0.05

ARGS = {
    "W": W,
    "target": target,
    "threshold": threshold
}
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, 4) — shape (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
compute_cycle_strength Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
markdana/local-ling-discovery/local_isa_ling/block_diagonal_processing.py
code served (permissive licence) · get_code("dc52901e1f5769c3")
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 float64, ndim 0, 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 a small square matrix W
# The function uses W[perm] for indexing, diagonal access, and matrix operations
# W_prime gets divided by its diagonal, so diagonal should be non-zero
np.random.seed(0)
W = np.random.rand(4, 4) + 1.0  # Add 1.0 to ensure non-zero diagonal

# Create a permutation of indices
perm = [2, 0, 3, 1]

ARGS = {
    "W": W,
    "perm": perm,
}
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 float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, which disagrees with the fixture run
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
count_accuracy_of_mb Ran
ran; our contract draft was wrong, not the code
markdana/local-ling-discovery/utils/metrics.py
code served (permissive licence) · get_code("eb5aab91f877d5bb")
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 numpy as np

# Analyze the function:
# count_accuracy_of_mb takes two parameters:
# - mb_true: a set (has .intersection() method, supports - operator)
# - mb_est: a set (has - operator, supports len())
# The function computes set operations to get true positives, false positives, false negatives

# Create example sets for Markov blanket (typically variable indices)
mb_true = {0, 1, 2, 3}
mb_est = {1, 2, 4}

ARGS = {
    'mb_true': mb_true,
    'mb_est': mb_est
}
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 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
estimate_markov_blanket Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
markdana/local-ling-discovery/utils/markov_blanket.py
code served (permissive licence) · get_code("e1c6fa33cb193a70")
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 set. 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.linear_model import Lasso

# Create sample data for the function
# X should be a 2D array where columns are variables and rows are samples
# target should be an integer index

np.random.seed(0)

# Create a small dataset: 10 samples, 5 variables
X = np.random.randn(10, 5)

# target should be a valid column index (0-4)
target = 2

ARGS = {
    "X": X,
    "target": target
}
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 set — a set output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a set 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_column_with_one_nonzero Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
markdana/local-ling-discovery/local_isa_ling/acyclic_sink_processing.py
code served (permissive licence) · get_code("08d21923803611bd")
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 numpy as np

# Construct valid arguments for get_column_with_one_nonzero
# The function expects:
# - W_iter: a 2D numpy array (matrix with some zeros and nonzeros)
# - W: a 2D numpy array (matrix to be modified)
# - rows_mapping: a numpy array or dict mapping row indices
# - columns_mapping: a numpy array or dict mapping column indices

# Create a small example W_iter matrix (5x5)
# with multiple nonzero entries in different columns
W_iter = np.array([
    [1, 0, 2, 0, 0],
    [0, 3, 0, 0, 0],
    [0, 0, 4, 5, 0],
    [0, 0, 0, 6, 0],
    [0, 0, 0, 0, 7]
], dtype=float)

# Create a corresponding W matrix (same size)
W = np.array([
    [1, 0, 2, 0, 0],
    [0, 3, 0, 0, 0],
    [0, 0, 4, 5, 0],
    [0, 0, 0, 6, 0],
    [0, 0, 0, 0, 7]
], dtype=float)

# Create mappings from original indices to potentially different indices
# These are used to map back to the original W matrix
rows_mapping = np.array([0, 1, 2, 3, 4], dtype=int)
columns_mapping = np.array([0, 1, 2, 3, 4], dtype=int)

ARGS = {
    "W_iter": W_iter,
    "W": W,
    "rows_mapping": rows_mapping,
    "columns_mapping": columns_mapping,
}
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 — 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
get_parents Ran
ran, honoured the contract we drafted
markdana/local-ling-discovery/utils/graph.py
code served (permissive licence) · get_code("44c5f0acd1db0b3c")
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 int, 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

# Create a simple matrix B where rows represent items and columns represent features/connections
# The function looks up row i and finds all column indices where B[i, :] != 0
# This is a 5x7 matrix with some non-zero entries
B = np.array([
    [0, 1, 0, 2, 0, 0, 1],
    [3, 0, 1, 0, 0, 2, 0],
    [0, 0, 0, 0, 1, 1, 0],
    [2, 2, 0, 0, 0, 0, 3],
    [0, 0, 0, 0, 0, 0, 0]
], dtype=np.float32)

# Select a valid row index to query
i = 2

ARGS = {
    "B": B,
    "i": i
}
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 int, 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_vstructs Ran
ran; our contract draft was wrong, not the code
markdana/local-ling-discovery/utils/graph.py
code served (permissive licence) · get_code("93a8f8022a34d375")
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 set. 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 itertools import combinations

# Construct a small adjacency matrix B
# B[i, j] != 0 means there's an edge from j to i (j is a parent of i)
# Let's create a 4x4 matrix representing a DAG

# Example structure:
# Node 0: parents are {}
# Node 1: parents are {0}
# Node 2: parents are {0}
# Node 3: parents are {1, 2} (v-structure: 1 -> 3 <- 2, where 1 and 2 are not connected)

B = np.array([
    [0, 0, 0, 0],  # Node 0: no parents
    [1, 0, 0, 0],  # Node 1: parent is 0
    [1, 0, 0, 0],  # Node 2: parent is 0
    [0, 1, 1, 0],  # Node 3: parents are 1 and 2 (v-structure)
], dtype=float)

ARGS = {
    "B": B,
}
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 set — a set output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a set 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
is_dag Ran
ran, violated the contract we drafted
markdana/local-ling-discovery/utils/graph.py
code served (permissive licence) · get_code("3c1bf98579e1564c")
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 bool, 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 igraph as ig

torch.manual_seed(0)

# The function expects B to be array-like with a .T (transpose) attribute
# and a .tolist() method, so a numpy array works perfectly
B = np.array([
    [0, 1, 0, 1],
    [0, 0, 1, 0],
    [0, 0, 0, 1],
    [0, 0, 0, 0]
], dtype=float)

ARGS = {
    "B": B
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
igraph, numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bool, 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
is_singleton Ran
ran, violated the contract we drafted
markdana/local-ling-discovery/local_isa_ling/acyclic_sink_processing.py
code served (permissive licence) · get_code("ba75cb86644a5849")
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 bool, 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

# Analyze the function:
# - subspaces: iterable of collections that support 'in' operator and len()
# - source_idx: value to check membership in subspaces

# Create test data
subspaces = [
    {1, 2, 3},      # set containing source_idx=1 with len > 1
    {4, 5},         # set not containing source_idx
    {7}             # singleton set not containing source_idx
]

source_idx = 1

ARGS = {
    "subspaces": subspaces,
    "source_idx": source_idx
}
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 bool, 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
lasso Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
markdana/local-ling-discovery/utils/markov_blanket.py
code served (permissive licence) · get_code("d1ff8680471e0c76")
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 sklearn.linear_model import Lasso

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

# Create a small sample dataset
# X should be a 2D array with shape (n_samples, n_features)
# Using 5 samples and 4 features
n_samples = 5
n_features = 4
X = np.random.randn(n_samples, n_features)

# target should be an integer index less than n_features
target = 1

# l1_lambda is the regularization parameter
l1_lambda = 0.1

ARGS = {
    "X": X,
    "target": target,
    "l1_lambda": l1_lambda,
}
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 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
sample_dag Ran
ran, honoured the contract we drafted
markdana/local-ling-discovery/data_loader/sample_dag.py
code served (permissive licence) · get_code("22ef3504a4923055")
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 (10, 10) 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
import igraph as ig

# Set random seed for reproducibility
np.random.seed(42)

# Define ARGS dictionary with all parameters for sample_dag
ARGS = {
    'd': 10,           # number of nodes
    'degree': 4,       # expected degree (will be converted to s0 internally)
    'graph_type': 'ER' # graph type: 'ER', 'SF', or 'BP'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
igraph, 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 (10, 10) — shape (10, 10) 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
sample_data Ran
ran; our contract draft was wrong, not the code
markdana/local-ling-discovery/data_loader/sample_data.py
code served (permissive licence) · get_code("d81335fb6c059e2a")
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 (10, 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 random seed for reproducibility
np.random.seed(0)

# Define parameters for sample_data function
B = np.array([[0.0, 0.2, 0.1, 0.0],
              [0.0, 0.0, 0.3, 0.15],
              [0.1, 0.0, 0.0, 0.2],
              [0.0, 0.1, 0.0, 0.0]])

noise_scales = [0.5, 0.3, 0.4, 0.2]

num_samples = 10

ARGS = {
    "B": B,
    "noise_scales": noise_scales,
    "num_samples": num_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 ndarray, ndim 2, dtype family float, shape (10, 4) — shape (10, 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
sample_er_dcg Ran
ran, honoured the contract we drafted
markdana/local-ling-discovery/data_loader/sample_dcg.py
code served (permissive licence) · get_code("aece1776553b8d83")
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 (5, 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 numpy as np

np.random.seed(0)

ARGS = {
    "n_vars": 5,
    "max_degree": 3,
    "max_cycle": 2,
}
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 (5, 5) — shape (5, 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
sample_param_unif Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
markdana/local-ling-discovery/data_loader/sample_data.py
code served (permissive licence) · get_code("dd0c66a9906862be")
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 numpy as np

np.random.seed(42)

# B_support: a square matrix indicating which edges exist in the graph
# Using a small 4x4 matrix with some sparsity pattern
B_support = np.array([
    [0, 1, 0, 1],
    [0, 0, 1, 0],
    [1, 0, 0, 1],
    [0, 1, 0, 0]
], dtype=float)

ARGS = {
    'B_support': B_support,
    'B_low': 0.5,
    'B_high': 0.9,
    'noise_low': 0.75,
    'noise_high': 1.25,
    'flip_sign': True,
    'max_eig': 1.0,
    'max_cond_number': float('inf')
}
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 — 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
scorecond Ran
ran, honoured the contract we drafted
markdana/local-ling-discovery/local_isa_ling/isa.py
code served (permissive licence) · get_code("71ddf6f8f03056f0")
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, 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 numpy as np
import numba as nb

np.random.seed(0)

# Create sample data: (n_samples, n_vars)
# n_samples = 5 (sample size)
# n_vars = 3 (number of variables)
data = np.random.randn(5, 3).astype(np.float64)

ARGS = {
    'data': data
}
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, 5) — shape (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
block_diagonal_processing Not yet run
unverified (harvested, no recorded run)
markdana/local-ling-discovery/local_isa_ling/block_diagonal_processing.py
code served (permissive licence) · get_code("436c036bed175033")
How it ran

no invocation recorded.

get_pr Not yet run
unverified (harvested, no recorded run)
markdana/local-ling-discovery/local_isa_ling/isa.py
code served (permissive licence) · get_code("a62c2c6ad341ab35")
How it ran

no invocation recorded.

get_psi Not yet run
unverified (harvested, no recorded run)
markdana/local-ling-discovery/local_isa_ling/isa.py
code served (permissive licence) · get_code("7550c08b662ee3c0")
How it ran

no invocation recorded.

Also present in other papers' code

Of 18 samples lifted from this paper's linked repositories, 18 carry a function identity and 0 do not. Of the 18 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 18 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

Local causal discovery is of great practical significance, as there are often situations where the discovery of the global causal structure is unnecessary, and the interest lies solely on a single target variable. Most existing local methods utilize conditional independence relations, providing only a partially directed graph, and assume acyclicity for the ground-truth structure, even though real-world scenarios often involve cycles like feedback mechanisms. In this work, we present a general, unified local causal discovery method with linear non-Gaussian models, whether they are cyclic or acyclic. We extend the application of independent component analysis from the global context to independent subspace analysis, enabling the exact identification of the equivalent local directed structures and causal strengths from the Markov blanket of the target variable. We also propose an alternative regression-based method in the particular acyclic scenarios. Our identifiability results are empirically validated using both synthetic and real-world datasets.

For agents

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

get_harvested_code_for_paper("2403.14843")
get_code_for_paper("2403.14843")
have("2403.14843")

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

curl https://syntology.ai/api/ran/2403.14843.json

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

[![Syntology run record](https://syntology.ai/api/ran/2403.14843.svg)](https://syntology.ai/paper/2403.14843)

Connect an agent — have() is free.