SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2205.00293 · NeurIPS · 2022

TTOpt: A Maximum Volume Quantized Tensor Train-based Optimization and its Application to Reinforcement Learning

Ivan Oseledets, Anh-Huy Phan, Konstantin Sozykin, Andrei Chertkov, Roman Schutski, Andrzej Cichocki

arXiv · PDF · Open in the Atlas

Code that ran

14 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; 6 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 8 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/2205.00293.json · badge: /api/ran/2205.00293.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
AndreiChertkov/ttopt canonical (classifier label)
classifier basis: classifier_2025
1 ran · 0 no recorded run
andreichertkov/ttopt marked official in Papers with Code
Papers with Code archive
13 ran · 3 no recorded run
FunctionStatusWhere it lives
_add_random Ran
ran, honoured the contract we drafted
andreichertkov/ttopt/ttopt/ttopt.py
code served (permissive licence) · get_code("aad0069afc94eb49")
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 int, shape (3, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

np.random.seed(0)

# J should be a 2D numpy array (matrix)
# Looking at the function: it uses np.vstack to add a row to J
# So J needs to be a 2D array with some number of columns
J = np.array([[1, 2, 3, 4],
              [5, 6, 7, 8]], dtype=int)

# n should be a sequence where each element k is something that np.random.choice can sample from
# np.random.choice(k) samples from k, so each k should be array-like
# n should have the same length as the number of columns in J (4 columns)
n = [np.array([10, 11, 12]), 
     np.array([20, 21, 22]), 
     np.array([30, 31, 32]), 
     np.array([40, 41, 42])]

ARGS = {
    "J": J,
    "n": n
}
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 int, shape (3, 4) — shape (3, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
_add_row Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
andreichertkov/ttopt/ttopt/ttopt.py
code served (permissive licence) · get_code("45daa0d32d507479")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 2, dtype family float, shape (3, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Construct arguments for _add_row function
# The function takes:
# - J: a 2D numpy array (Jacobian matrix)
# - i_new: a 1D numpy array (new row to add)

# Create a small 2D array (Jacobian) with shape (2, 4)
J = np.array([[1.0, 2.0, 3.0, 4.0],
              [5.0, 6.0, 7.0, 8.0]])

# Create a small 1D array (new row) with shape (4,)
i_new = np.array([9.0, 10.0, 11.0, 12.0])

ARGS = {
    "J": J,
    "i_new": i_new
}
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, 4) — shape (3, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
_maxvol Ran
ran, honoured the contract we drafted
andreichertkov/ttopt/ttopt/ttopt.py
code served (permissive licence) · get_code("54af7ad58f9e8e83")
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 (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 numpy as np

# Create a tall matrix (n > r) for testing
np.random.seed(42)
n, r = 10, 3
A = np.random.randn(n, r).astype(np.float64)

# Ensure the matrix has full rank by using QR decomposition
Q, R = np.linalg.qr(A)
A = Q[:, :r]

ARGS = {
    "A": A,
    "tol": 1.001,
    "max_iters": 1000,
    "is_rect": False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family int, 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
_merge Ran
ran, honoured the contract we drafted
andreichertkov/ttopt/ttopt/ttopt.py
code served (permissive licence) · get_code("544dde7bda277c94")
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 int, shape (12, 7) 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 _ones(k, m=1):
    return np.ones((k, m), dtype=int)

def _merge(J1, J2, Jg):
    r1 = J1.shape[0] if J1 is not None else 1
    r2 = J2.shape[0] if J2 is not None else 1
    n = Jg.shape[0]
    I = np.kron(np.kron(_ones(r2), Jg), _ones(r1))
    if J1 is not None:
        J1_ = np.kron(_ones(n * r2), J1)
        I = np.hstack((J1_, I))
    if J2 is not None:
        J2_ = np.kron(J2, _ones(r1 * n))
        I = np.hstack((I, J2_))
    return I

# Construct valid arguments
J1 = np.array([[1, 2], [3, 4]], dtype=int)  # shape (2, 2)
J2 = np.array([[5, 6], [7, 8], [9, 10]], dtype=int)  # shape (3, 2)
Jg = np.array([[11, 12, 13], [14, 15, 16]], dtype=int)  # shape (2, 3)

ARGS = {
    "J1": J1,
    "J2": J2,
    "Jg": Jg
}
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 int, shape (12, 7) — shape (12, 7) 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
_ones Ran
ran, honoured the contract we drafted
andreichertkov/ttopt/ttopt/ttopt.py
code served (permissive licence) · get_code("29694d2ff2cf642d")
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 int, shape (5, 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 numpy as np

# The function _ones(k, m=1) returns a numpy array of ones with shape (k, m) and dtype int.
# Parameters:
# - k: the number of rows (required)
# - m: the number of columns (optional, default=1)

ARGS = {
    "k": 5,
    "m": 3,
}
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 int, shape (5, 3) — shape (5, 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
_reshape Ran
ran; our contract draft was wrong, not the code
andreichertkov/ttopt/ttopt/ttopt.py
code served (permissive licence) · get_code("b0e00d33104a35ea")
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 int, shape (4, 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 numpy as np

# The function _reshape takes two parameters:
# - A: an array-like object to reshape
# - n: the target shape (as a tuple or int)
# The function uses np.reshape with order='F' (Fortran/column-major order)

# Create a simple numpy array with distinct dimensions for probing
A = np.arange(12).reshape(3, 4)  # 3x4 array

# Define target shape - reshaping to a different size
n = (4, 3)  # Reshape to 4x3

ARGS = {
    "A": A,
    "n": n
}
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 int, shape (4, 3) — shape (4, 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
_stack Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
andreichertkov/ttopt/ttopt/ttopt.py
code served (permissive licence) · get_code("05a3404311550f0a")
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 int, shape (6, 7) 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 arguments for _stack function
# Parameters: J, Jg, l2r=True

# J is either None or a numpy array with shape (r, ?) where r is number of rows
# Jg is a numpy array with shape (n, ?) where n is number of rows
# l2r is a boolean flag

J = np.array([[1, 2, 3], [4, 5, 6]], dtype=int)  # shape (2, 3)
Jg = np.array([[7, 8, 9, 10], [11, 12, 13, 14], [15, 16, 17, 18]], dtype=int)  # shape (3, 4)
l2r = True

ARGS = {
    "J": J,
    "Jg": Jg,
    "l2r": l2r
}
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 int, shape (6, 7) — shape (6, 7) 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
_update_iter Ran
ran, honoured the contract we drafted
andreichertkov/ttopt/ttopt/ttopt.py
code served (permissive licence) · get_code("0f46f52d25d095e0")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family int; 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)

# Analyze the function:
# - d: appears to be compared with i (integer comparison), likely an int representing dimension/depth
# - i: integer index that gets incremented/decremented
# - iter: integer iteration counter
# - l2r: boolean flag (left-to-right)
# The function updates these based on boundary conditions

ARGS = {
    'd': 10,      # dimension/depth limit (int)
    'i': 5,       # current index (int)
    'iter': 0,    # iteration counter (int)
    'l2r': True   # left-to-right flag (bool)
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family int — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
f Ran
ran, honoured the contract we drafted
AndreiChertkov/ttopt/demo/base.py
code served (permissive licence) · get_code("c2cbec6be13031a9")
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 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

# Create a small numpy array for X
# The function expects X to be a 2D array where it computes operations along axis=1
# Using shape (2, 4) - batch size 2, feature dimension 4
X = np.array([
    [1.0, 2.0, 3.0, 4.0],
    [0.5, 1.5, 2.5, 3.5]
], dtype=np.float32)

ARGS = {
    "X": X
}
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 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
maxvol Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
andreichertkov/ttopt/ttopt/ttopt.py
code served (permissive licence) · get_code("d0d48af7e87ae1cf")
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

# Create a tall matrix (n > r)
# Shape: [6, 3] - 6 rows, 3 columns
np.random.seed(0)
A = np.random.randn(6, 3).astype(np.float64)

# Ensure the matrix has full column rank by using QR decomposition
Q, R = np.linalg.qr(A)
A = Q  # This ensures A is tall and has full rank

ARGS = {
    'A': A,
    'e': 1.05,
    'k': 100
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — 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
maxvol_rect Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
andreichertkov/ttopt/ttopt/ttopt.py
code served (permissive licence) · get_code("e6e1ec163ad0ad39")
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
from scipy.linalg import lu
from scipy.linalg import solve_triangular

# Create a tall matrix A with shape [n, r] where n > r
np.random.seed(42)
n = 10
r = 3
A = np.random.randn(n, r)

# Ensure A has full rank by using QR decomposition
Q, _ = np.linalg.qr(A)
A = Q[:, :r]

# Define all parameters for maxvol_rect
ARGS = {
    'A': A,
    'e': 1.1,
    'dr_min': 0,
    'dr_max': 3,
    'e0': 1.05,
    'k0': 10,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — 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
ttopt_find Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
andreichertkov/ttopt/ttopt/ttopt.py
code served (permissive licence) · get_code("c162281061210a99")
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

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

# Construct arguments for ttopt_find
# I: array of indices/points, shape (n_samples, n_dims)
I = np.array([[0, 1, 2, 3],
              [4, 5, 6, 7]], dtype=np.float32)

# y: array of function values at those points, shape (n_samples,)
y = np.array([1.5, 0.8], dtype=np.float32)

# opt: array of optimizer states or parameters, one per sample
# Since opt[ind] is accessed, it should be indexable and return something per sample
opt = np.array([[10.1, 20.2, 30.3, 40.4],
                [50.5, 60.6, 70.7, 80.8]], dtype=np.float32)

# i_opt: current best index/point
i_opt = np.array([0, 1, 2, 3], dtype=np.float32)

# y_opt: current best value (or None)
y_opt = None

# opt_opt: current best optimizer state
opt_opt = np.array([10.1, 20.2, 30.3, 40.4], dtype=np.float32)

ARGS = {
    'I': I,
    'y': y,
    'opt': opt,
    'i_opt': i_opt,
    'y_opt': y_opt,
    'opt_opt': opt_opt,
    'is_max': False
}
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
ttopt_fs Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
andreichertkov/ttopt/ttopt/ttopt.py
code served (permissive licence) · get_code("b149936290bc356d")
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

# Construct arguments for ttopt_fs
y = np.array([1.5, 2.3, 0.8, 3.1])
y0 = 0.5
opt = 1.0

ARGS = {
    "y": y,
    "y0": y0,
    "opt": opt
}
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
ttopt_init Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
andreichertkov/ttopt/ttopt/ttopt.py
code served (permissive licence) · get_code("9205862ee0ada55b")
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. 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

ARGS = {
    'n': [2, 3, 4],
    'rank': 2,
    'Y0': None,
    'seed': 42,
    'with_rank': False,
}
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 list — a list 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
TTOpt Not yet run
unverified (harvested, no recorded run)
andreichertkov/ttopt/ttopt/ttopt.py
code served (permissive licence) · get_code("5770a9f5bb8c38e9")
How it ran

no invocation recorded.

_iter Not yet run
unverified (harvested, no recorded run)
andreichertkov/ttopt/ttopt/ttopt.py
code served (permissive licence) · get_code("cc0a7bd6b0e7bd68")
How it ran

no invocation recorded.

ttopt Not yet run
unverified (harvested, no recorded run)
andreichertkov/ttopt/ttopt/ttopt.py
code served (permissive licence) · get_code("f3644d1aaa2f9c70")
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

We present a novel procedure for optimization based on the combination of efficient quantized tensor train representation and a generalized maximum matrix volume principle. We demonstrate the applicability of the new Tensor Train Optimizer (TTOpt) method for various tasks, ranging from minimization of multidimensional functions to reinforcement learning. Our algorithm compares favorably to popular gradient-free methods and outperforms them by the number of function evaluations or execution time, often by a significant margin.

For agents

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

get_harvested_code_for_paper("2205.00293")
get_code_for_paper("2205.00293")
have("2205.00293")

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

curl https://syntology.ai/api/ran/2205.00293.json

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

[![Syntology run record](https://syntology.ai/api/ran/2205.00293.svg)](https://syntology.ai/paper/2205.00293)

Connect an agent — have() is free.