SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2402.07341 · ICML · 2024

Noise-Adaptive Confidence Sets for Linear Bandits and Application to Bayesian Optimization

Jungtaek Kim, Kwang-Sung Jun

arXiv · PDF · Open in the Atlas

Code that ran

15 of 15 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 6 honoured the contract we drafted, 1 violated it and 0 have no contract verdict; 8 failures of our instrument; 0 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/2402.07341.json · badge: /api/ran/2402.07341.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
jungtaekkim/LOSAN-LOFAV canonical (classifier label)
classifier basis: census
12 ran · 0 no recorded run
jungtaekkim/losan-lofav canonical (classifier label)
classifier basis: classifier_2025
3 ran · 0 no recorded run
FunctionStatusWhere it lives
calculate_D_t Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
jungtaekkim/losan-lofav/losan_lofav/losan.py
code served (permissive licence) · get_code("80dc1d6f184bf30a")
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 float64, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np
import scipy.spatial.distance as scisd

np.random.seed(0)

# Create small test data
dim = 4
bx = np.random.randn(dim)
w = np.random.randn(1)  # scalar weight
Sigma = np.random.randn(dim, dim)
Sigma = Sigma @ Sigma.T  # make it positive definite
Sigma_inverse = np.linalg.inv(Sigma)

ARGS = {
    "bx": bx,
    "w": w,
    "Sigma_inverse": Sigma_inverse
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, scipy — 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, 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
calculate_L Ran
ran, honoured the contract we drafted
jungtaekkim/LOSAN-LOFAV/losan_lofav/lofav.py
code served (permissive licence) · get_code("923a7aa11b372188")
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 int64, ndim 0, dtype family int, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

def calculate_L(T, dim):
    L = np.ceil(1 / 2 * np.log2(T / dim)).astype(np.int64)
    L = np.maximum(1, L)
    return L

# T should be a numeric value (sequence length or total size)
# dim should be a numeric value (dimension)
T = 128
dim = 4

ARGS = {
    "T": T,
    "dim": dim
}
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 int64, ndim 0, dtype family int, 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
calculate_Sigma Ran
ran, honoured the contract we drafted
jungtaekkim/LOSAN-LOFAV/losan_lofav/linear_models.py
code served (permissive licence) · get_code("991446f285a41bf3")
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

# Create a small data matrix X with shape (batch=2, features=4)
X = np.array([
    [1.0, 2.0, 3.0, 4.0],
    [5.0, 6.0, 7.0, 8.0]
], dtype=np.float64)

# Create a regularization parameter lambda_
lambda_ = 0.1

ARGS = {
    "X": X,
    "lambda_": lambda_
}
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
calculate_Sigma_t Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
jungtaekkim/losan-lofav/losan_lofav/losan.py
code served (permissive licence) · get_code("2e28195f08ce8691")
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 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

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

# Define parameters
# X: shape (n_samples, n_features) - assert X.ndim == 2
# lambda_: scalar regularization parameter
# bw: shape (n_samples,) - assert bw.ndim == 1 and X.shape[0] == bw.shape[0]

n_samples = 2
n_features = 4

X = np.random.randn(n_samples, n_features)
lambda_ = 0.1
bw = np.random.rand(n_samples)

ARGS = {
    "X": X,
    "lambda_": lambda_,
    "bw": bw
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
calculate_predictions Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
jungtaekkim/LOSAN-LOFAV/losan_lofav/linear_models.py
code served (permissive licence) · get_code("83a1582dfbc7d4e1")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 1, dtype family float, shape (2,) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# X should be a 2D array (batch_size, n_features)
X = np.array([
    [1.0, 2.0, 3.0, 4.0],
    [5.0, 6.0, 7.0, 8.0]
], dtype=np.float64)

# theta should be a 1D array (n_features,)
theta = np.array([0.1, 0.2, 0.3, 0.4], dtype=np.float64)

ARGS = {
    "X": X,
    "theta": theta
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family float, shape (2,) — shape (2,) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
calculate_rho_l Ran
ran, honoured the contract we drafted
jungtaekkim/LOSAN-LOFAV/losan_lofav/lofav.py
code served (permissive licence) · get_code("06663716441337df")
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

torch.manual_seed(0)

# The function calculate_rho_l takes a single parameter 'l' and computes 2^(-l)
# Looking at the function body, 'l' should be a numeric value (int or float)
# that can be used as an exponent

l = 3

ARGS = {
    "l": l
}
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 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
calculate_theta_hat Ran
ran, violated the contract we drafted
jungtaekkim/LOSAN-LOFAV/losan_lofav/linear_models.py
code served (permissive licence) · get_code("fba3fac32e739f9c")
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

# Create small test data for the function
np.random.seed(0)

# X: feature matrix (n_samples, n_features)
# The function does: Sigma_inverse @ X.T @ by
# X.T will be (n_features, n_samples), so X should be (n_samples, n_features)
X = np.random.randn(2, 4)  # 2 samples, 4 features

# by: response vector (n_samples,)
by = np.random.randn(2)

# Sigma_inverse: precision matrix (n_features, n_features)
# This gets multiplied with X.T which is (n_features, n_samples)
Sigma_inverse = np.random.randn(4, 4)

ARGS = {
    "X": X,
    "by": by,
    "Sigma_inverse": Sigma_inverse,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family float, shape (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
calculate_w_t Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
jungtaekkim/losan-lofav/losan_lofav/losan.py
code served (permissive licence) · get_code("df7a616fb8f85529")
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 float64, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np
import scipy.spatial.distance as scisd

np.random.seed(0)

# Create a 1D array for bx
bx = np.array([1.5, -0.3, 2.1, 0.8])

# Create a 2D positive definite matrix for Sigma_prev_inverse
# Using a simple approach: create a random matrix and make it positive definite
dim = 4
A = np.random.randn(dim, dim)
Sigma_prev_inverse = A @ A.T + np.eye(dim)

ARGS = {
    "bx": bx,
    "Sigma_prev_inverse": Sigma_prev_inverse,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, scipy — 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, 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
calculate_w_tl Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
jungtaekkim/LOSAN-LOFAV/losan_lofav/lofav.py
code served (permissive licence) · get_code("2fdadb9d07b6f319")
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
import scipy.spatial.distance as scisd

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

# Create small valid arguments for calculate_w_tl
bx = np.array([0.5, 1.2, -0.3, 0.8])  # 1D array representing a point in 4D space

# Create a positive definite covariance matrix (4x4)
# Using a simple symmetric positive definite matrix
Sigma_prev_inverse = np.array([
    [2.0, 0.1, 0.0, 0.0],
    [0.1, 2.5, 0.1, 0.0],
    [0.0, 0.1, 2.0, 0.1],
    [0.0, 0.0, 0.1, 2.2]
])

# Ensure it's symmetric and positive definite
Sigma_prev_inverse = (Sigma_prev_inverse + Sigma_prev_inverse.T) / 2
eigenvalues = np.linalg.eigvals(Sigma_prev_inverse)
if np.any(eigenvalues <= 0):
    Sigma_prev_inverse += np.eye(4) * (abs(np.min(eigenvalues)) + 0.1)

# rho_l is a scaling parameter (positive scalar)
rho_l = 0.5

ARGS = {
    "bx": bx,
    "Sigma_prev_inverse": Sigma_prev_inverse,
    "rho_l": rho_l
}
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 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
generate_rademacher_noises Ran
ran; our contract draft was wrong, not the code
jungtaekkim/LOSAN-LOFAV/losan_lofav/utils_optimization.py
code served (permissive licence) · get_code("10e169edf72bcc15")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 2, dtype family float, shape (2, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Create a random state object that the function expects
random_state = np.random.RandomState(seed=42)

# Define the arguments
ARGS = {
    "size": (2, 4),           # Small tuple for size parameter
    "sigma": 0.5,             # Scalar value for sigma
    "random_state": random_state  # numpy RandomState object
}
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 (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
get_arms Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
jungtaekkim/LOSAN-LOFAV/losan_lofav/utils_optimization.py
code served (permissive licence) · get_code("51e29dbeb8bfda28")
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, 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

# Construct bounds as a numpy array with shape (num_dimensions, 2)
# Each row represents [lower_bound, upper_bound] for one dimension
bounds = np.array([
    [0.0, 10.0],      # dimension 0: bounds from 0 to 10
    [5.0, 15.0],      # dimension 1: bounds from 5 to 15
    [-1.0, 1.0],      # dimension 2: bounds from -1 to 1
    [100.0, 200.0]    # dimension 3: bounds from 100 to 200
])

# Number of arms to generate
num_arms = 5

# Random state for reproducibility
random_state = np.random.RandomState(42)

ARGS = {
    "bounds": bounds,
    "num_arms": num_arms,
    "random_state": random_state
}
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, 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
get_arms_unitsphere Ran
ran, honoured the contract we drafted
jungtaekkim/LOSAN-LOFAV/losan_lofav/utils_optimization.py
code served (permissive licence) · get_code("83d54d372ee996ec")
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
from numpy.random import RandomState

# Construct arguments for get_arms_unitsphere
dim_problem = 5
num_arms = 3
S = 2.0
random_state = RandomState(42)

ARGS = {
    "dim_problem": dim_problem,
    "num_arms": num_arms,
    "S": S,
    "random_state": random_state,
}
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 (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
get_inverse_with_cholesky Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
jungtaekkim/LOSAN-LOFAV/losan_lofav/utils_regression.py
code served (permissive licence) · get_code("3b4fc3f3f55e08d5")
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
import scipy.linalg as scila

# Create a small positive definite matrix for Cholesky decomposition
# Using a simple approach: create a random matrix and multiply by its transpose
np.random.seed(0)
random_matrix = np.random.randn(4, 4)
X = random_matrix @ random_matrix.T  # This ensures positive definiteness

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 — 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
get_lambda Ran
ran, honoured the contract we drafted
jungtaekkim/LOSAN-LOFAV/losan_lofav/utils_regression.py
code served (permissive licence) · get_code("63057e612b635486")
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

torch.manual_seed(0)

# The function get_lambda takes sigma and S as parameters
# It computes sigma ** 2 / S ** 2
# Both sigma and S should be numeric values (scalars or tensors)

sigma = 2.5
S = 1.5

ARGS = {
    "sigma": sigma,
    "S": S
}
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 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
sample_unknown_theta Ran
ran, honoured the contract we drafted
jungtaekkim/LOSAN-LOFAV/losan_lofav/utils_regression.py
code served (permissive licence) · get_code("420d42afe9fd5262")
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 (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 scipy.linalg as scila

# Set random seed for reproducibility
np.random.seed(42)
random_state = np.random.RandomState(42)

# Define parameters
dim_theta = 5
S = 2.0

# Construct ARGS dictionary
ARGS = {
    "dim_theta": dim_theta,
    "S": S,
    "random_state": random_state
}
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 (5,) — shape (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

Also present in other papers' code

Of 15 samples lifted from this paper's linked repositories, 15 carry a function identity and 0 do not. Of the 15 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 15 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

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

Abstract

Adapting to a priori unknown noise level is a very important but challenging problem in sequential decision-making as efficient exploration typically requires knowledge of the noise level, which is often loosely specified. We report significant progress in addressing this issue for linear bandits in two respects. First, we propose a novel confidence set that is 'semi-adaptive' to the unknown sub-Gaussian parameter σ 2 * in the sense that the (normalized) confidence width scales with dσ 2 * + σ 2 0 where d is the dimension and σ 2 0 is the specified sub-Gaussian parameter (known) that can be much larger than σ 2 * . This is a significant improvement over dσ 2 0 of the standard confidence set of Abbasi-Yadkori et al. ( 2011), especially when d is large. We show that this leads to an improved regret bound in linear bandits. Second, for bounded rewards, we propose a novel variance-adaptive confidence set that has much improved numerical performance upon prior art. We then apply this confidence set to develop, as we claim, the first practical variance-adaptive linear bandit algorithm via an optimistic approach, which is enabled by our novel regret analysis technique. Both of our confidence sets rely critically on 'regret equality' from online learning. Our empirical evaluation in diverse Bayesian optimization tasks shows that our proposed algorithms demonstrate better or comparable performance compared to existing methods. The first setup is when the noise η t | F t-1 is σ 2 * -sub-Gaussian where the sub-Gaussian parameter specified to the algorithm is σ 2 0 that can be much larger than σ 2 * . We propose a novel linear bandit algorithm called LOSAN (Linear Optimism with Semi-Adaptivity to Noise). The critical ingredient for this algorithm is a novel confidence set whose (normalized) confidence width contains online variance estimators and is no larger than Õ( dσ 2 * + σ 2 0 ) with high probability where Õ hides polylogarithmic factors. This is no worse than Õ( dσ 2 0 + σ 2 0 ) of the standard confidence 1 The implementation of our proposed methods is available at https://github.com/jungtaekkim/LOSAN-LOFAV.

For agents

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

get_harvested_code_for_paper("2402.07341")
get_code_for_paper("2402.07341")
have("2402.07341")

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

curl https://syntology.ai/api/ran/2402.07341.json

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

[![Syntology run record](https://syntology.ai/api/ran/2402.07341.svg)](https://syntology.ai/paper/2402.07341)

Connect an agent — have() is free.