SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1810.02309 · 2018

Learning Compressed Transforms with Low Displacement Rank

arXiv · PDF · Open in the Atlas

Code that ran

10 of 21 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 10 have no contract verdict; 0 failures of our instrument; 11 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/1810.02309.json · badge: /api/ran/1810.02309.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
HazyResearch/structured-nets canonical (classifier label)
classifier basis: census
10 ran · 11 no recorded run
FunctionStatusWhere it lives
circ_diag_transpose_mult_fn Ran
ran on a synthesized input
HazyResearch/structured-nets/tensorflow/krylov.py
code served (permissive licence) · get_code("93114c4a7dcf4c7e")
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 EagerTensor, 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 tensorflow as tf
import numpy as np

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

# Create small test tensors
# v_f: vector for multiplication (size 4)
v_f = tf.constant([1.0, 2.0, 3.0, 4.0], dtype=tf.float32)

# d: diagonal vector (size 4)
d = tf.constant([0.5, 1.5, 2.5, 3.5], dtype=tf.float32)

# x: input vector (size 4)
x = tf.constant([10.0, 20.0, 30.0, 40.0], dtype=tf.float32)

# n: integer parameter (batch size or dimension)
n = 4

ARGS = {
    "v_f": v_f,
    "d": d,
    "x": x,
    "n": n
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, 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
circ_transpose_mult_fn Ran
ran on a synthesized input
HazyResearch/structured-nets/tensorflow/krylov.py
code served (permissive licence) · get_code("db48787ad55ef7af")
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 EagerTensor, 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 tensorflow as tf
import numpy as np

def circ_transpose_mult_fn(v_f, x, n):
    y = tf.concat([x[1:], [x[0]]], axis=0)
    return tf.multiply(y, v_f)

# Create small test tensors
v_f = tf.constant([1.0, 2.0, 3.0, 4.0], dtype=tf.float32)
x = tf.constant([5.0, 6.0, 7.0, 8.0], dtype=tf.float32)
n = 4

ARGS = {
    "v_f": v_f,
    "x": x,
    "n": n
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, 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
construct_model Ran
ran on a synthesized input
HazyResearch/structured-nets/pytorch/models/nets.py
code served (permissive licence) · get_code("f16e103b2cd798b5")
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 SimpleModel. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn
from inspect import signature

torch.manual_seed(0)

class SimpleModel(nn.Module):
    def __init__(self, in_size, out_size, hidden_dim=8, dropout=0.1):
        super().__init__()
        self.linear = nn.Linear(in_size, out_size)
    
    @staticmethod
    def args(hidden_dim=8, dropout=0.1):
        pass

class MockArgs:
    def __init__(self):
        self.hidden_dim = 8
        self.dropout = 0.1

ARGS = {
    "cls": SimpleModel,
    "in_size": 4,
    "out_size": 3,
    "args": MockArgs()
}
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 SimpleModel — a SimpleModel output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a SimpleModel output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
create_command Ran
ran on a synthesized input
HazyResearch/structured-nets/tensorflow/compare_parallel.py
code served (permissive licence) · get_code("b1be7db8436ba3d5")
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 str. 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 argparse
import torch
import torch.nn as nn

torch.manual_seed(0)

# Create a minimal argparse Namespace with all required attributes
args = argparse.Namespace(
    name="test_experiment",
    dataset="mnist",
    result_dir="/tmp/results",
    decay_freq=10,
    steps=100,
    batch_size=32,
    test=False,
    layer_size=64,
    transform="normalize",
    torch=True,
    model="resnet"
)

# Define the ARGS dictionary mapping all parameters of create_command
ARGS = {
    'args': args,
    'method': "sgd",
    'rank': 5,
    'lr': 0.001,
    'decay_rate': 0.1,
    'mom': 0.9
}
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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
cross_entropy_loss Ran
ran on a synthesized input
HazyResearch/structured-nets/pytorch/utils.py
code served (permissive licence) · get_code("8ef54cdb8384cbab")
How it ran

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

import torch
import torch.nn as nn

torch.manual_seed(0)

# The function takes pred and true tensors
# pred is used with torch.max(pred, 1) and passed to nn.CrossEntropyLoss()
# CrossEntropyLoss expects logits of shape (batch_size, num_classes)
# true is used with torch.max(true, 1) to get argmax
# true appears to be one-hot encoded or soft targets of shape (batch_size, num_classes)

batch_size = 2
num_classes = 4

# pred: logits from model, shape (batch_size, num_classes)
pred = torch.randn(batch_size, num_classes)

# true: one-hot or soft target labels, shape (batch_size, num_classes)
# Create valid targets by one-hot encoding
true_labels = torch.tensor([1, 2])  # class indices
true = torch.zeros(batch_size, num_classes)
true.scatter_(1, true_labels.unsqueeze(1), 1.0)

ARGS = {
    "pred": pred,
    "true": true
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
descendants Ran
ran on a synthesized input
HazyResearch/structured-nets/pytorch/utils.py
code served (permissive licence) · get_code("da5caff101ceaca4")
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 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

# Construct a simple class hierarchy to test the descendants function
class Animal:
    pass

class Dog(Animal):
    pass

class Cat(Animal):
    pass

class Poodle(Dog):
    pass

class Siamese(Cat):
    pass

class Tabby(Siamese):
    pass

ARGS = {
    'cls': Animal
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, 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
get_dataset Ran
ran on a synthesized input
HazyResearch/structured-nets/pytorch/dataset.py
code served (permissive licence) · get_code("caeffaa5677b8ac4")
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
import pickle as pkl
import tempfile
import os

torch = __import__('torch')

# Create temporary directory for mock data files
temp_dir = tempfile.mkdtemp()
dataset_dir = os.path.join(temp_dir, 'mnist')
os.makedirs(dataset_dir, exist_ok=True)

# Create mock dataset files
train_data = {
    'X': np.random.randn(2, 784).astype(np.float32),
    'Y': np.random.randint(0, 10, (2, 10)).astype(np.float32)
}

test_data = {
    'X': np.random.randn(2, 784).astype(np.float32),
    'Y': np.random.randint(0, 10, (2, 10)).astype(np.float32)
}

train_loc = os.path.join(dataset_dir, 'train_normalized')
test_loc = os.path.join(dataset_dir, 'test_normalized')

with open(train_loc, 'wb') as f:
    pkl.dump(train_data, f)

with open(test_loc, 'wb') as f:
    pkl.dump(test_data, f)

ARGS = {
    'dataset_name': 'mnist',
    'data_dir': temp_dir,
    'transform': ''
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
h5py, 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
identity_mult_fn Ran
ran on a synthesized input
HazyResearch/structured-nets/tensorflow/krylov.py
code served (permissive licence) · get_code("143fa9ec9a47925c")
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
import torch.nn as nn

torch.manual_seed(0)

# The function identity_mult_fn(v, n) simply returns v without using n
# So we need to provide:
# - v: any value (could be a tensor, number, or other object)
# - n: any value (not used in the function body)

ARGS = {
    "v": torch.randn(2, 4),
    "n": 5
}
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
mse_loss Ran
ran on a synthesized input
HazyResearch/structured-nets/pytorch/utils.py
code served (permissive licence) · get_code("77458eee7fb25de2")
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 torch
import torch.nn as nn

torch.manual_seed(0)

# Create small tensors for pred and true
# Both need to be float tensors that MSELoss can compute on
pred = torch.randn(2, 4)
true = torch.randn(2, 4)

ARGS = {
    "pred": pred,
    "true": true
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
split_train_val Ran
ran on a synthesized input
HazyResearch/structured-nets/pytorch/dataset.py
code served (permissive licence) · get_code("fe4c30a21b8a8bc7")
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
import torch

torch.manual_seed(0)
np.random.seed(0)

# Create small tensors for train_X and train_Y
# train_X: shape (8, 4) - 8 samples, 4 features
# train_Y: shape (8, 3) - 8 samples, 3 output dimensions
train_X = torch.randn(8, 4)
train_Y = torch.randn(8, 3)

# Parameters for the function
val_fraction = 0.25  # 25% for validation
train_fraction = None  # Will default to remainder

ARGS = {
    'train_X': train_X,
    'train_Y': train_Y,
    'val_fraction': val_fraction,
    'train_fraction': train_fraction,
}
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 — 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
StructuredLinear Not yet run
unverified (harvested, no recorded run)
HazyResearch/structured-nets/pytorch/structure/layer.py
code served (permissive licence) · get_code("03ab2af9fe40e057")
How it ran

no invocation recorded.

check_rank Not yet run
unverified (harvested, no recorded run)
HazyResearch/structured-nets/tensorflow/model.py
code served (permissive licence) · get_code("cae5d71738786dfd")
How it ran

no invocation recorded.

circulant_sparsity Not yet run
unverified (harvested, no recorded run)
HazyResearch/structured-nets/tensorflow/learned_operators.py
code served (permissive licence) · get_code("b42d372b234d35de")
How it ran

no invocation recorded.

create_data_loaders Not yet run
unverified (harvested, no recorded run)
HazyResearch/structured-nets/pytorch/dataset.py
code served (permissive licence) · get_code("57fe2dad04e97dae")
How it ran

no invocation recorded.

forward Not yet run
unverified (harvested, no recorded run)
HazyResearch/structured-nets/tensorflow/model.py
code served (permissive licence) · get_code("948d6e2b3de233d3")
How it ran

no invocation recorded.

get_structured_W Not yet run
unverified (harvested, no recorded run)
HazyResearch/structured-nets/tensorflow/model.py
code served (permissive licence) · get_code("769fab47f27b89f1")
How it ran

no invocation recorded.

hankel_like Not yet run
unverified (harvested, no recorded run)
HazyResearch/structured-nets/tensorflow/fixed_operators.py
code served (permissive licence) · get_code("fb5f1ea3206617c6")
How it ran

no invocation recorded.

polynomial_transform Not yet run
unverified (harvested, no recorded run)
HazyResearch/structured-nets/tensorflow/learned_operators.py
code served (permissive licence) · get_code("c4119d98ae620aa5")
How it ran

no invocation recorded.

toeplitz_like Not yet run
unverified (harvested, no recorded run)
HazyResearch/structured-nets/tensorflow/fixed_operators.py
code served (permissive licence) · get_code("f6411433b636f6e4")
How it ran

no invocation recorded.

tridiagonal_corner Not yet run
unverified (harvested, no recorded run)
HazyResearch/structured-nets/tensorflow/learned_operators.py
code served (permissive licence) · get_code("68475184587cb150")
How it ran

no invocation recorded.

vandermonde_like Not yet run
unverified (harvested, no recorded run)
HazyResearch/structured-nets/tensorflow/fixed_operators.py
code served (permissive licence) · get_code("63f703ee8d1bd7fb")
How it ran

no invocation recorded.

Also present in other papers' code

Of 21 samples lifted from this paper's linked repositories, 21 carry a function identity and 0 do not. Of the 21 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 21 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

The low displacement rank (LDR) framework for structured matrices represents a matrix through two displacement operators and a low-rank residual. Existing use of LDR matrices in deep learning has applied fixed displacement operators encoding forms of shift invariance akin to convolutions. We introduce a class of LDR matrices with more general displacement operators, and explicitly learn over both the operators and the low-rank component. This class generalizes several previous constructions while preserving compression and efficient computation. We prove bounds on the VC dimension of multi-layer neural networks with structured weight matrices and show empirically that our compact parameterization can reduce the sample complexity of learning. When replacing weight layers in fully-connected, convolutional, and recurrent neural networks for image classification and language modeling tasks, our new classes exceed the accuracy of existing compression approaches, and on some tasks also outperform general unstructured layers while using more than 20x fewer parameters.

For agents

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

get_harvested_code_for_paper("1810.02309")
get_code_for_paper("1810.02309")
have("1810.02309")

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

curl https://syntology.ai/api/ran/1810.02309.json

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

[![Syntology run record](https://syntology.ai/api/ran/1810.02309.svg)](https://syntology.ai/paper/1810.02309)

Connect an agent — have() is free.