SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1901.05134 · 2019

DINGO: Distributed Newton-Type Method for Gradient-Norm Optimization

arXiv · PDF · Open in the Atlas

Code that ran

16 of 19 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 16 have no contract verdict; 0 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/1901.05134.json · badge: /api/ran/1901.05134.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
RixonC/DINGO canonical (classifier label)
classifier basis: census
16 ran · 3 no recorded run
FunctionStatusWhere it lives
Ax Ran
ran on a synthesized input
RixonC/DINGO/minres_qlp.py
code served (permissive licence) · get_code("83d9c16e37611d39")
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 (4, 1) 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)

# Option 1: A as a callable function
def matrix_vector_product(x):
    """A simple matrix-vector product function."""
    A_matrix = torch.randn(4, 3)
    return torch.mm(A_matrix, x)

# Option 2: A as an explicit tensor
A_tensor = torch.randn(4, 3)

# Vector x
x = torch.randn(3, 1)

# Create ARGS dict with A as a callable
ARGS = {
    "A": matrix_vector_product,
    "x": x
}

# Alternative ARGS with A as a tensor (uncomment to use):
# ARGS = {
#     "A": A_tensor,
#     "x": x
# }
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 (4, 1) — shape (4, 1) 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
CG Ran
ran on a synthesized input
RixonC/DINGO/cg.py
code served (permissive licence) · get_code("d21487af01879377")
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

torch.manual_seed(0)

# Define a simple matrix-vector product function A
# The function should take a vector p and return A*p
def matrix_vector_product(p):
    # Create a simple 4x4 matrix for testing
    A_matrix = torch.tensor([
        [4.0, 1.0, 0.0, 0.0],
        [1.0, 3.0, 1.0, 0.0],
        [0.0, 1.0, 2.0, 1.0],
        [0.0, 0.0, 1.0, 1.0]
    ], device=p.device)
    return torch.matmul(A_matrix, p)

# Define the vector b
b = torch.tensor([[1.0], [2.0], [3.0], [4.0]])

# Define the device
device = torch.device('cpu')

# Create the ARGS dictionary
ARGS = {
    'A': matrix_vector_product,
    'b': b,
    'device': device,
    'rtol': 0.0001,
    'maxit': 100
}
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
LSMR Ran
ran on a synthesized input
RixonC/DINGO/lsmr.py
code served (permissive licence) · get_code("4cac8089e515bce5")
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
from copy import deepcopy

torch.manual_seed(0)

# Define matrix dimensions: A is m x n, so m=5 rows, n=4 columns
m, n = 5, 4

# Create a simple matrix A for testing
A_matrix = torch.randn(m, n)

# Define A as a function that performs matrix-vector product
def A_func(v):
    """Matrix-vector product: A @ v"""
    return A_matrix @ v

# Define A_transpose as a function that performs transpose matrix-vector product
def A_transpose_func(u):
    """Transpose matrix-vector product: A.T @ u"""
    return A_matrix.T @ u

# Create a right-hand side vector b
b = torch.randn(m, 1)

# Solution dimension is the number of columns in A
dimension = n

# Device specification (CPU only)
device = torch.device("cpu")

# Build ARGS dictionary with all parameters
ARGS = {
    "A": A_func,
    "A_transpose": A_transpose_func,
    "b": b,
    "dimension": dimension,
    "device": device,
    "myRtol": 0.0001,
    "maxit": 100,
    "damp": 0.0,
    "atol": 0.0,
    "btol": 0.0,
    "conlim": 0.0,
}
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
Precond Ran
ran on a synthesized input
RixonC/DINGO/minres_qlp.py
code served (permissive licence) · get_code("95b0f58fcfa41cac")
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 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
import numpy as np

torch.manual_seed(0)

# Precond takes two arguments: M and r
# Looking at the function body, it simply returns r
# M is used as a preconditioner matrix (likely applied to r)
# r is a residual vector that gets returned

# M should be a matrix-like object (could be a tensor, array, or operator)
# r should be a vector/tensor

M = torch.randn(4, 4)  # Preconditioner matrix
r = torch.randn(4)     # Residual vector

ARGS = {
    "M": M,
    "r": r,
}
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 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
get_local_line_search_matrix Ran
ran on a synthesized input
RixonC/DINGO/DINGO.py
code served (permissive licence) · get_code("a0ca34c5f4a73d8b")
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 (20, 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 torch
import torch.nn as nn
from copy import deepcopy

torch.manual_seed(0)

# Build a minimal model
class TinyModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear1 = nn.Linear(4, 3)
        self.linear2 = nn.Linear(3, 1)
    
    def forward(self, x):
        x = self.linear1(x)
        x = self.linear2(x)
        return x

model = TinyModel()

# Build a minimal Worker class
class Worker:
    def __init__(self):
        # Create a simple batch: (input, target)
        self.x = torch.randn(2, 4)
        self.y = torch.randn(2, 1)
    
    def get_local_loss(self, model):
        """Returns loss as a scalar tensor."""
        output = model(self.x)
        loss = torch.nn.functional.mse_loss(output, self.y)
        return loss
    
    def get_local_gradient(self, model):
        """Returns gradient flattened into a 1D tensor."""
        model.zero_grad()
        output = model(self.x)
        loss = torch.nn.functional.mse_loss(output, self.y)
        loss.backward()
        
        # Flatten all gradients into a single tensor
        gradients = []
        for param in model.parameters():
            if param.grad is not None:
                gradients.append(param.grad.flatten())
        
        return torch.cat(gradients, dim=0).reshape(-1, 1)

worker = Worker()

# Create update direction: flattened parameter vector
total_params = sum(p.numel() for p in model.parameters())
update_direction = torch.randn(total_params)

# Set line search iterations
line_search_max_iterations = 2

ARGS = {
    "model": model,
    "worker": worker,
    "update_direction": update_direction,
    "line_search_max_iterations": line_search_max_iterations
}
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 (20, 2) — shape (20, 2) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_local_line_search_matrix Ran
ran on a synthesized input
RixonC/DINGO/DINGO_with_only_case_1.py
code served (permissive licence) · get_code("d0d50e8ee368ffc1")
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 (20, 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 torch.nn as nn
from copy import deepcopy

torch.manual_seed(0)

# Define a minimal model
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear1 = nn.Linear(4, 3)
        self.linear2 = nn.Linear(3, 1)
    
    def forward(self, x):
        x = torch.relu(self.linear1(x))
        x = self.linear2(x)
        return x

# Define a minimal Worker class
class Worker:
    def __init__(self):
        self.data_x = torch.randn(2, 4)
        self.data_y = torch.randn(2, 1)
    
    def get_local_loss(self, model):
        """Returns a scalar loss tensor"""
        with torch.no_grad():
            output = model(self.data_x)
            loss = torch.mean((output - self.data_y) ** 2)
        return loss
    
    def get_local_gradient(self, model):
        """Returns gradient as a 1D tensor"""
        # Compute loss with gradients
        output = model(self.data_x)
        loss = torch.mean((output - self.data_y) ** 2)
        
        # Compute gradients
        grads = torch.autograd.grad(loss, model.parameters(), create_graph=False)
        
        # Flatten and concatenate all gradients into a single tensor
        flat_grads = torch.cat([g.reshape(-1) for g in grads])
        return flat_grads.reshape(-1, 1)

# Create instances
model = SimpleModel()

# Compute the total number of parameters to create appropriate update_direction
total_params = sum(p.numel() for p in model.parameters())
update_direction = torch.randn(total_params)

worker = Worker()

line_search_max_iterations = 3

ARGS = {
    "model": model,
    "worker": worker,
    "update_direction": update_direction,
    "line_search_max_iterations": line_search_max_iterations,
}
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 (20, 3) — shape (20, 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
get_local_line_search_values Ran
ran on a synthesized input
RixonC/DINGO/GIANT.py
code served (permissive licence) · get_code("d45b190cd1294398")
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 (1, 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 torch.nn as nn
from copy import deepcopy

torch.manual_seed(0)

# Define a minimal model
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear1 = nn.Linear(4, 3)
        self.linear2 = nn.Linear(3, 2)
    
    def forward(self, x):
        x = self.linear1(x)
        x = self.linear2(x)
        return x

# Define a minimal Worker class
class Worker:
    def __init__(self):
        self.data = torch.randn(2, 4)
        self.target = torch.randn(2, 2)
    
    def get_local_loss(self, model):
        output = model(self.data)
        loss = torch.nn.functional.mse_loss(output, self.target)
        return loss

# Create instances
model = SimpleModel()

# Calculate total number of parameters for update_direction
total_params = sum(p.numel() for p in model.parameters())

# Create update_direction tensor
update_direction = torch.randn(total_params)

# Create worker
worker = Worker()

# Set line search iterations
line_search_max_iterations = 3

ARGS = {
    "model": model,
    "worker": worker,
    "update_direction": update_direction,
    "line_search_max_iterations": line_search_max_iterations
}
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 (1, 3) — shape (1, 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
get_model_weights Ran
ran on a synthesized input
RixonC/DINGO/AIDE.py
code served (permissive licence) · get_code("2d58dd6bd6825205")
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 (23, 1) 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)

class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear1 = nn.Linear(4, 3)
        self.linear2 = nn.Linear(3, 2)
    
    def forward(self, x):
        x = self.linear1(x)
        x = self.linear2(x)
        return x

model = SimpleModel()

ARGS = {
    "model": model
}
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 (23, 1) — shape (23, 1) 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_new_y_and_new_zeta Ran
ran on a synthesized input
RixonC/DINGO/AIDE.py
code served (permissive licence) · get_code("c2c7d1e297cbadcb")
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 numpy as np

torch.manual_seed(0)

# Create small tensor arguments for old_weights and new_weights
old_weights = torch.randn(4)
new_weights = torch.randn(4)

# Create scalar arguments
tau = 0.5
old_zeta = 0.3
lamda = 0.1

ARGS = {
    "old_weights": old_weights,
    "new_weights": new_weights,
    "tau": tau,
    "old_zeta": old_zeta,
    "lamda": lamda
}
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
get_updated_model Ran
ran on a synthesized input
RixonC/DINGO/DINGO.py
code served (permissive licence) · get_code("ecf4c55475f0e8c4")
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 copy import deepcopy

torch.manual_seed(0)

# Create a simple model
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear1 = nn.Linear(4, 3)
        self.linear2 = nn.Linear(3, 2)
    
    def forward(self, x):
        x = self.linear1(x)
        x = self.linear2(x)
        return x

model = SimpleModel()

# Calculate total number of parameters to create matching update_direction
total_params = sum(p.numel() for p in model.parameters())

# Create update direction tensor with same total number of elements
update_direction = torch.randn(total_params)

# Create step size
step_size = 0.01

ARGS = {
    "model": model,
    "update_direction": update_direction,
    "step_size": step_size
}
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
get_updated_model Ran
ran on a synthesized input
RixonC/DINGO/DINGO_with_only_case_1.py
code served (permissive licence) · get_code("56db0b4cabde68fe")
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 copy import deepcopy

torch.manual_seed(0)

# Build a minimal model
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear1 = nn.Linear(4, 3)
        self.linear2 = nn.Linear(3, 2)
    
    def forward(self, x):
        x = self.linear1(x)
        x = self.linear2(x)
        return x

# Create model instance
model = SimpleModel()

# Calculate total number of parameters
total_params = sum(p.numel() for p in model.parameters())

# Create update_direction tensor with same total size as all model parameters
update_direction = torch.randn(total_params)

# Create step_size
step_size = 0.01

ARGS = {
    "model": model,
    "update_direction": update_direction,
    "step_size": step_size
}
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
get_updated_model Ran
ran on a synthesized input
RixonC/DINGO/DiSCO.py
code served (permissive licence) · get_code("b0d9abbfd044b38b")
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 copy import deepcopy

torch.manual_seed(0)

# Create a simple model
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear1 = nn.Linear(4, 3)
        self.linear2 = nn.Linear(3, 2)
    
    def forward(self, x):
        x = self.linear1(x)
        x = self.linear2(x)
        return x

# Instantiate the model
model = SimpleModel()

# Calculate total number of parameters
total_params = sum(p.numel() for p in model.parameters())

# Create update_direction tensor with the same total number of elements
update_direction = torch.randn(total_params)

ARGS = {
    "model": model,
    "update_direction": update_direction
}
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
line_search Ran
ran on a synthesized input
RixonC/DINGO/GIANT.py
code served (permissive licence) · get_code("9da5748a7cdf1878")
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
from copy import deepcopy

torch.manual_seed(0)

# Create a small model
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear1 = nn.Linear(4, 3)
        self.linear2 = nn.Linear(3, 2)
    
    def forward(self, x):
        x = self.linear1(x)
        x = self.linear2(x)
        return x

model = SimpleModel()

# Create update_direction tensor (flattened parameters)
# Count total parameters in model
total_params = sum(p.numel() for p in model.parameters())
update_direction = torch.randn(total_params, 1)

# Create line_search_values tensor: shape (1, num_line_search_steps)
# Each column represents objective value at different step sizes
line_search_values = torch.tensor([[10.5, 9.8, 9.2, 8.9, 8.8]], dtype=torch.float32)

# Current loss value
loss = 11.0

# Gradient tensor (same shape as update_direction)
gradient = torch.randn(total_params, 1)

# Line search rho parameter (Armijo condition parameter)
line_search_rho = 0.1

ARGS = {
    "model": model,
    "line_search_values": line_search_values,
    "loss": loss,
    "gradient": gradient,
    "update_direction": update_direction,
    "line_search_rho": line_search_rho,
}
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
minresQLP Ran
ran on a synthesized input
RixonC/DINGO/minres_qlp.py
code served (permissive licence) · get_code("c80b3003bd596b2a")
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 numpy as np

torch.manual_seed(0)

# Define a simple matrix-vector product function
def matrix_vector_product(v):
    """Simple linear operator: multiply by a fixed matrix."""
    A_matrix = torch.tensor([
        [2.0, 0.5, 0.1, 0.0],
        [0.5, 3.0, 0.2, 0.1],
        [0.1, 0.2, 1.5, 0.3],
        [0.0, 0.1, 0.3, 2.5]
    ])
    return torch.mm(A_matrix, v)

# Create a small right-hand side vector
b = torch.tensor([[1.0], [2.0], [0.5], [1.5]])

# Set device to CPU
device = torch.device('cpu')

# Define preconditioning function (identity for simplicity)
def precond_fn(M, r):
    """Simple preconditioner - identity."""
    return r

ARGS = {
    'A': matrix_vector_product,
    'b': b,
    'device': device,
    'rtol': 1e-4,
    'maxit': 50,
    'M': None,
    'shift': 0.0,
    'maxxnorm': 1e7,
    'Acondlim': 1e15,
    'TranCond': 1e7,
    'show': False,
    'rnormvec': False
}
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
replace_model_weights Ran
ran on a synthesized input
RixonC/DINGO/AIDE.py
code served (permissive licence) · get_code("e9168bd6b6f9f588")
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 Sequential. 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 a small model with a few parameters
model = nn.Sequential(
    nn.Linear(4, 3),
    nn.ReLU(),
    nn.Linear(3, 2)
)

# Calculate total number of parameters
total_params = sum(p.numel() for p in model.parameters())

# Create new weights as a column tensor
new_weights = torch.randn(total_params, 1)

ARGS = {
    "model": model,
    "new_weights": new_weights
}
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 Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential output the recorded ndim 1 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
symOrtho Ran
ran on a synthesized input
RixonC/DINGO/lsmr.py
code served (permissive licence) · get_code("dd06bcf31465b9ea")
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

torch.manual_seed(0)

ARGS = {
    "a": 3.5,
    "b": 2.1,
    "device": torch.device("cpu")
}
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
distributed_PCG_algorithm Not yet run
unverified (harvested, no recorded run)
RixonC/DINGO/DiSCO.py
code served (permissive licence) · get_code("b2046fce1a7c26a0")
How it ran

no invocation recorded.

get_update_direction Not yet run
unverified (harvested, no recorded run)
RixonC/DINGO/DINGO.py
code served (permissive licence) · get_code("3d413d6939d03a91")
How it ran

no invocation recorded.

get_update_direction Not yet run
unverified (harvested, no recorded run)
RixonC/DINGO/DINGO_with_only_case_1.py
code served (permissive licence) · get_code("c142669863cb0c2c")
How it ran

no invocation recorded.

Also present in other papers' code

Of 19 samples lifted from this paper's linked repositories, 19 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

For optimization of a sum of functions in a distributed computing environment, we present a novel communication efficient Newton-type algorithm that enjoys a variety of advantages over similar existing methods. Similar to Newton-MR, our algorithm, DINGO, is derived by optimization of the gradient's norm as a surrogate function. DINGO does not impose any specific form on the underlying functions, and its application range extends far beyond convexity. In addition, the distribution of the data across the computing environment can be arbitrary. Further, the underlying sub-problems of DINGO are simple linear least-squares, for which a plethora of efficient algorithms exist. Lastly, DINGO involves a few hyper-parameters that are easy to tune. Moreover, we theoretically show that DINGO is not sensitive to the choice of its hyper-parameters in that a strict reduction in the gradient norm is guaranteed, regardless of the selected hyper-parameters. We demonstrate empirical evidence of the effectiveness, stability and versatility of our method compared to other relevant algorithms.

For agents

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

get_harvested_code_for_paper("1901.05134")
get_code_for_paper("1901.05134")
have("1901.05134")

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

curl https://syntology.ai/api/ran/1901.05134.json

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

[![Syntology run record](https://syntology.ai/api/ran/1901.05134.svg)](https://syntology.ai/paper/1901.05134)

Connect an agent — have() is free.