SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2408.16862 · NeurIPS · 2024

Probabilistic Decomposed Linear Dynamical Systems for Robust Discovery of Latent Neural Dynamics

Christopher Rozell, Noga Mudrik, Adam Charles, Yenho Chen, Kyle Johnsen, Sankaraleengam Alagapan

arXiv · PDF · Open in the Atlas

Code that ran

12 of 17 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, 2 violated it and 1 has no contract verdict; 9 failures of our instrument; 5 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/2408.16862.json · badge: /api/ran/2408.16862.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
siplab-gt/probabilistic-decomposed-linear-dynamical-systems canonical (classifier label)
classifier basis: classifier_2025
12 ran · 5 no recorded run
FunctionStatusWhere it lives
LoessInference Ran
ran on a synthesized input
siplab-gt/probabilistic-decomposed-linear-dynamical-systems/pdLDS/pdLDS.py
code served (permissive licence) · get_code("86139453c8bce202")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) 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
import torch
from scipy.spatial import cKDTree

def tricube_weight(d):
    return (1 - d ** 3) ** 3

class LoessInference:

    def __init__(self, T, x=None, best=None, init=True):
        self.T = T
        self.t = torch.arange(self.T)
        self.kd = cKDTree(self.t[:, None])
        if init:
            assert x is not None, 'Need to give x if init'
            self.x = x
            self.best = best
            self.bestx = self.best

    def Loess(self, y, S=2, weight=tricube_weight):
        """
        First order local linear regression, but a bit faster

        Parameters:
        -----------
        x:
        y:
        s: float (0,1]
            smoothing parameter

        Returns: 
        --------
        values, (M,B)
        """
        T, d = y.shape
        k = S
        di, ix = self.kd.query(self.t[:, None], np.arange(1, k + 2))
        ix = torch.tensor(ix)
        di = torch.tensor(di)
        di = di
        sd = di / di.max(1).values[:, None]
        w = torch.sqrt(weight(sd))
        X_b = torch.ones([T, k + 1, 2])
        Y = torch.ones([T, k + 1, d])
        for i, (k_ix, w_) in enumerate(zip(ix, w)):
            X_b[i, :, 0] = self.t[k_ix]
            Y[i, :, :] = y[k_ix]
        lstsq = torch.linalg.lstsq(X_b * w_[None, :, None], Y * w_[None, :, None])
        M = lstsq.solution[:, 0, :]
        B = lstsq.solution[:, 1, :]
        values = M * self.t[:, None] + B
        return (values, (M, B))

    def loess_detrend(self, y, S=3):
        trend, (m, b) = self.Loess(y, S=S)
        lds = y - trend
        return (trend, lds)

    def skknee(self, s):
        """
        Objective to optimize
        
        """
        s = s[0]
        loss = self.compute_loess_score(self.x, s=s)
        score = loss - (self.Sm * s + self.Sb)
        return -score.item()

    def knee(self, s):
        """
        Objective to optimize
        
        """
        s = s[0]
        loss = self.compute_loess_score(self.x, s=s)
        score = loss - (self.Sm * s + self.Sb)
        return score

    def compute_loess_score(self, y, s=0.1):
        """
        Compute Loess. 
        """
        trend, lds = self.loess_detrend(y, s=s)
        a, _, _, _ = torch.linalg.lstsq(lds[:-1], lds[1:])
        a_loss = torch.mean((lds[1:] - lds[:-1] @ a) ** 2)
        return a_loss


T = 8
x = torch.randn(T, 4)
best = torch.randn(T, 4)

INIT_ARGS = {
    'T': T,
    'x': x,
    'best': best,
    'init': True
}

y = torch.randn(T, 4)

ARGS = {
    'y': y,
    'S': 2
}

METHOD = 'Loess'

if __name__ == '__main__':
    loess = LoessInference(**INIT_ARGS)
    result = getattr(loess, METHOD)(**ARGS)
    print("Result shape:", result[0].shape)
    print("M shape:", result[1][0].shape)
    print("B shape:", result[1][1].shape)
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
convertfastSBL Ran
ran; our contract draft was wrong, not the code
siplab-gt/probabilistic-decomposed-linear-dynamical-systems/pdLDS/pdLDS.py
code served (permissive licence) · get_code("598a3a7e7d9b2fb8")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned tuple. 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
import torch

torch.manual_seed(0)

# Create small numpy arrays to simulate fastSBL output
c_filt_np = np.random.randn(2, 4)
gammas_np = np.random.randn(2, 4)
lambdas_np = np.random.randn(4)
a_np = np.random.randn(2, 4)
b_np = np.random.randn(2, 4)
sigmas_np = np.random.randn(2, 4)

# Pack into tuple as expected by the function
fastOutput = (c_filt_np, gammas_np, lambdas_np, a_np, b_np, sigmas_np)

ARGS = {
    "fastOutput": fastOutput
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
ensure_input_torch_tensor Ran
ran, violated the contract we drafted
siplab-gt/probabilistic-decomposed-linear-dynamical-systems/pdLDS/pdLDS.py
code served (permissive licence) · get_code("94722e3dceeec582")
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 list, 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 torch
import numpy as np

torch.manual_seed(0)

# Construct test data: a list of mixed types (tensors and non-tensors)
# The function iterates over Y and converts non-tensors to torch.Tensor
Y = [
    torch.tensor([[1.0, 2.0, 3.0, 4.0]]),  # Already a tensor
    [[5.0, 6.0, 7.0, 8.0]],                # List (will be converted)
    np.array([[9.0, 10.0, 11.0, 12.0]]),  # NumPy array (will be converted)
]

ARGS = {
    "Y": Y
}
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 list, ndim 3, dtype family float — a list 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
group_consecutive_ixs Ran
ran; our contract draft was wrong, not the code
siplab-gt/probabilistic-decomposed-linear-dynamical-systems/pdLDS/pdLDS.py
code served (permissive licence) · get_code("55cff33b42930a46")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned Tensor, 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

torch.manual_seed(0)

# The function expects a tensor x where:
# - x is indexed with [1:] and [:-1], so it needs at least 2 elements
# - x.float() is called, so it should be a numeric tensor
# - x.float()[1:] - x.float()[:-1] is computed and .abs().sum(1) is called
#   This means x should be 2D (batch_size, features) so that sum(1) works
# - len(x) is called, so x needs to support len()

# Create a small 2D tensor (batch_size=4, features=3)
x = torch.tensor([
    [1, 2, 3],
    [1, 2, 3],
    [4, 5, 6],
    [7, 8, 9]
], dtype=torch.int32)

ARGS = {
    "x": x
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, 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
impute_c_dynamics Ran
ran; our contract draft was wrong, not the code
siplab-gt/probabilistic-decomposed-linear-dynamical-systems/pdLDS/pdLDS.py
code served (permissive licence) · get_code("0b43c6eccccd0b14")
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 (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

torch.manual_seed(0)

# Create small tensors with distinct dimensions
c_filt = torch.randn(2, 4)  # batch=2, features=4
basis = torch.randn(2, 4)   # same shape as c_filt
impute_val = torch.randn(2, 4)  # same shape as c_filt
thresh = 0.01

ARGS = {
    "c_filt": c_filt,
    "basis": basis,
    "impute_val": impute_val,
    "thresh": thresh,
}
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 (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
kalman_filter Ran
ran; our contract draft was wrong, not the code
siplab-gt/probabilistic-decomposed-linear-dynamical-systems/pdLDS/pdLDS.py
code served (permissive licence) · get_code("8a99d76652f912b3")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned tuple. 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

torch.manual_seed(0)

# Parameters for kalman_filter
T = 2  # time steps
N = 4  # state dimension
M = 3  # measurement dimension

# y: measurements over time (T, M)
y = torch.randn(T, M)

# F: state transition matrices (T-1, N, N)
F = torch.randn(T - 1, N, N)

# C: measurement matrix (M, N)
C = torch.randn(M, N)

# x0: initial state (N,)
x0 = torch.randn(N)

# Sy: measurement noise (M, M)
Sy = torch.eye(M)

# Sx: process noise (N, N)
Sx = torch.eye(N)

# S0: initial state covariance (N, N)
S0 = torch.eye(N)

ARGS = {
    "y": y,
    "F": F,
    "C": C,
    "x0": x0,
    "Sy": Sy,
    "Sx": Sx,
    "S0": S0,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — 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
obs_ll Ran
ran; our contract draft was wrong, not the code
siplab-gt/probabilistic-decomposed-linear-dynamical-systems/pdLDS/pdLDS.py
code served (permissive licence) · get_code("c1cbd3502f9d2ead")
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

torch.manual_seed(0)

# y: (T, M) - observations over time
# x: (T, N) - state variables over time
# C: (M, N) - observation matrix
# S: (M, M) - observation covariance matrix

T = 2  # time steps
M = 4  # observation dimension
N = 3  # state dimension

y = torch.randn(T, M)
x = torch.randn(T, N)
C = torch.randn(M, N)
S = torch.randn(M, M)
S = S @ S.T  # make symmetric positive definite

ARGS = {
    "y": y,
    "x": x,
    "C": C,
    "S": S,
}
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
rts_smoother Ran
ran; our contract draft was wrong, not the code
siplab-gt/probabilistic-decomposed-linear-dynamical-systems/pdLDS/pdLDS.py
code served (permissive licence) · get_code("92c9475a50e20ad1")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned tuple. 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

torch.manual_seed(0)

T = 5  # time steps
N = 4  # state dimension

# Create valid tensors for the function
x_filt = torch.randn(T, N)
x_pred = torch.randn(T, N)
p_filt = torch.randn(T, N, N)
p_pred = torch.randn(T, N, N)
F = torch.randn(T, N, N)
KT = torch.randn(N, N)
C = torch.randn(N, N)

# Ensure p_filt and p_pred are positive definite (symmetric)
p_filt = p_filt @ p_filt.transpose(-2, -1)
p_pred = p_pred @ p_pred.transpose(-2, -1)

# Ensure p_pred is invertible for torch.linalg.solve
p_pred = p_pred + torch.eye(N) * 0.1

ARGS = {
    "x_filt": x_filt,
    "x_pred": x_pred,
    "p_filt": p_filt,
    "p_pred": p_pred,
    "F": F,
    "KT": KT,
    "C": C,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — 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
sma Ran
ran; our contract draft was wrong, not the code
siplab-gt/probabilistic-decomposed-linear-dynamical-systems/pdLDS/pdLDS.py
code served (permissive licence) · get_code("0476ab0d65525fd9")
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 (5, 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 torch.nn as nn

torch.manual_seed(0)

# x: (T, N) tensor - time steps T=5, features N=4
x = torch.randn(5, 4)

# w: window size for moving average
w = 3

ARGS = {
    'x': x,
    'w': w,
}
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 (5, 4) — shape (5, 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
smooth_support_acrosschannels Ran
ran; our contract draft was wrong, not the code
siplab-gt/probabilistic-decomposed-linear-dynamical-systems/pdLDS/pdLDS.py
code served (permissive licence) · get_code("407c1b04ee600304")
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 int, shape (8, 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

torch.manual_seed(0)

# Construct a support tensor
# The function expects a tensor where:
# - It iterates through windows of size w+1
# - It computes rates based on changes across channels (dim 1)
# - It performs operations like indexing, cloning, and rounding
# Shape: (sequence_length, num_channels)
support = torch.tensor([
    [1, 0, 1, 0],
    [1, 0, 1, 0],
    [1, 1, 1, 0],
    [1, 1, 0, 0],
    [1, 1, 0, 1],
    [1, 1, 0, 1],
    [1, 0, 0, 1],
    [1, 0, 0, 1],
], dtype=torch.long)

# Window size for smoothing
w = 2

# Rate threshold for filtering
rate_threshold = 0.05

ARGS = {
    'support': support,
    'w': w,
    'rate_threshold': rate_threshold,
}
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 int, shape (8, 4) — shape (8, 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
smooth_support_channelwise Ran
ran; our contract draft was wrong, not the code
siplab-gt/probabilistic-decomposed-linear-dynamical-systems/pdLDS/pdLDS.py
code served (permissive licence) · get_code("ba7dfc3867f3a4d3")
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 (5, 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

torch.manual_seed(0)

# Analyze the function:
# - support: indexed with [i:i+w+1], .float(), .mean(0) called on it, torch.round() applied
#   This suggests it's a tensor with shape (sequence_length, channels)
# - w: used in range(len(support) - w) and w // 2 for indexing
#   This is an integer window size

# Create a small support tensor with sequence dimension and channel dimension
support = torch.tensor([
    [1.0, 2.0, 3.0, 4.0],
    [2.0, 3.0, 4.0, 5.0],
    [3.0, 4.0, 5.0, 6.0],
    [4.0, 5.0, 6.0, 7.0],
    [5.0, 6.0, 7.0, 8.0],
], dtype=torch.float32)

# Window size must be less than len(support)
w = 2

ARGS = {
    "support": support,
    "w": w,
}
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 (5, 4) — shape (5, 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
tricube_weight Ran
ran, violated the contract we drafted
siplab-gt/probabilistic-decomposed-linear-dynamical-systems/pdLDS/pdLDS.py
code served (permissive licence) · get_code("50c52f6414793099")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned 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)

# The function tricube_weight takes a single parameter 'd' and computes (1 - d ** 3) ** 3
# The parameter 'd' is used in arithmetic operations, so it should be a numeric value or tensor
# Let's create a small tensor to pass as 'd'

d = torch.tensor([0.1, 0.5, 0.9], dtype=torch.float32)

ARGS = {
    "d": d
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
dyn_ll Not yet run
unverified (harvested, no recorded run)
siplab-gt/probabilistic-decomposed-linear-dynamical-systems/pdLDS/pdLDS.py
code served (permissive licence) · get_code("d0c08ec91757149b")
How it ran

no invocation recorded.

fastSBLDFInference Not yet run
unverified (harvested, no recorded run)
siplab-gt/probabilistic-decomposed-linear-dynamical-systems/pdLDS/pdLDS.py
code served (permissive licence) · get_code("5952c498911b256b")
How it ran

no invocation recorded.

pdLDS Not yet run
unverified (harvested, no recorded run)
siplab-gt/probabilistic-decomposed-linear-dynamical-systems/pdLDS/pdLDS.py
code served (permissive licence) · get_code("482a9f52a775403c")
How it ran

no invocation recorded.

solve_segmented_restricted_lstsq Not yet run
unverified (harvested, no recorded run)
siplab-gt/probabilistic-decomposed-linear-dynamical-systems/pdLDS/pdLDS.py
code served (permissive licence) · get_code("30ef48202e7b5f62")
How it ran

no invocation recorded.

sparseSmoother_opt Not yet run
unverified (harvested, no recorded run)
siplab-gt/probabilistic-decomposed-linear-dynamical-systems/pdLDS/pdLDS.py
code served (permissive licence) · get_code("f9eedbd7372c4ccf")
How it ran

no invocation recorded.

Also present in other papers' code

Of 17 samples lifted from this paper's linked repositories, 17 carry a function identity and 0 do not. Of the 17 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 17 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

Time-varying linear state-space models are powerful tools for obtaining mathematically interpretable representations of neural signals. For example, switching and decomposed models describe complex systems using latent variables that evolve according to simple locally linear dynamics. However, existing methods for latent variable estimation are not robust to dynamical noise and system nonlinearity due to noise-sensitive inference procedures and limited model formulations. This can lead to inconsistent results on signals with similar dynamics, limiting the model's ability to provide scientific insight. In this work, we address these limitations and propose a probabilistic approach to latent variable estimation in decomposed models that improves robustness against dynamical noise. Additionally, we introduce an extended latent dynamics model to improve robustness against system nonlinearities. We evaluate our approach on several synthetic dynamical systems, including an empirically-derived brain-computer interface experiment, and demonstrate more accurate latent variable inference in nonlinear systems with diverse noise conditions. Furthermore, we apply our method to a real-world clinical neurophysiology dataset, illustrating the ability to identify interpretable and coherent structure where previous models cannot. 1

For agents

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

get_harvested_code_for_paper("2408.16862")
get_code_for_paper("2408.16862")
have("2408.16862")

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

curl https://syntology.ai/api/ran/2408.16862.json

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

[![Syntology run record](https://syntology.ai/api/ran/2408.16862.svg)](https://syntology.ai/paper/2408.16862)

Connect an agent — have() is free.