SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2103.12032 · CVPR · 2021

Tangent Space Backpropagation for 3D Transformation Groups

Jia Deng, Zachary Teed

arXiv · PDF · Open in the Atlas

Code that ran

7 of 8 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 7 have no contract verdict; 0 failures of our instrument; 1 no recorded run), measured 2026-11-22; 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/2103.12032.json · badge: /api/ran/2103.12032.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
princeton-vl/lietorch canonical (classifier label)
classifier basis: census
7 ran · 1 no recorded run
FunctionStatusWhere it lives
broadcast_inputs Ran
ran on a synthesized input
princeton-vl/lietorch/lietorch/broadcasting.py
code served (permissive licence) · get_code("85c90ef2973afc32")
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 test tensors with distinct dimensions
# x: shape (2, 3, 4) - batch dims (2, 3), feature dim 4
# y: shape (2, 3, 5) - batch dims (2, 3), feature dim 5
x = torch.randn(2, 3, 4)
y = torch.randn(2, 3, 5)

ARGS = {
    'x': x,
    'y': y
}
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
cat Ran
ran on a synthesized input
princeton-vl/lietorch/lietorch/groups.py
code served (permissive licence) · get_code("5389af4c9463e592")
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 DataGroup. 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)

# Create a simple class that mimics the expected behavior
class DataGroup:
    def __init__(self, data):
        self.data = data

# Create small tensors with distinct dimensions
tensor1 = torch.randn(2, 4)
tensor2 = torch.randn(2, 4)
tensor3 = torch.randn(2, 4)

# Create DataGroup instances
group1 = DataGroup(tensor1)
group2 = DataGroup(tensor2)
group3 = DataGroup(tensor3)

# Build arguments
ARGS = {
    "group_list": [group1, group2, group3],
    "dim": 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 DataGroup — a DataGroup output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataGroup 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_numerical_jacobian Ran
ran on a synthesized input
princeton-vl/lietorch/lietorch/gradcheck.py
code served (permissive licence) · get_code("9acafca512a7baca")
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 (6, 6) 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 collections.abc as container_abcs

torch.manual_seed(0)

# Define a simple function that takes input and produces output
def simple_fn(input):
    """A simple differentiable function for testing."""
    if isinstance(input, torch.Tensor):
        return input ** 2
    elif isinstance(input, (list, tuple)):
        return torch.stack([simple_fn(x) for x in input])
    else:
        return input ** 2

# Create a small input tensor with requires_grad=True
input_tensor = torch.randn(2, 3, requires_grad=True, dtype=torch.float32)

# Create target tensor - can be different from input
target_tensor = torch.randn(2, 3, requires_grad=True, dtype=torch.float32)

ARGS = {
    "fn": simple_fn,
    "input": input_tensor,
    "target": target_tensor,
    "eps": 0.001,
    "grad_out": 1.0,
}
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 Tensor, ndim 2, dtype family float, shape (6, 6) — shape (6, 6) 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
make_homogeneous Ran
ran on a synthesized input
princeton-vl/lietorch/lietorch/run_tests.py
code served (permissive licence) · get_code("61494705bbb0f6ce")
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, 5) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# Create a small tensor with shape (2, 4)
# The function expects p to have a last dimension that will be concatenated with ones
p = torch.randn(2, 4)

ARGS = {
    "p": p
}
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, 5) — shape (2, 5) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
make_jacobian Ran
ran on a synthesized input
princeton-vl/lietorch/lietorch/gradcheck.py
code served (permissive licence) · get_code("6a9c78ecbe54f2b0")
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 (8, 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 torch.overrides import is_tensor_like
from collections import abc as container_abcs

torch.manual_seed(0)

# Create a floating-point tensor that requires gradients
input_tensor = torch.randn(2, 4, dtype=torch.float32, requires_grad=True)

# num_out is the number of outputs
num_out = 3

ARGS = {
    "input": input_tensor,
    "num_out": num_out,
}
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 (8, 3) — shape (8, 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
matv Ran
ran on a synthesized input
princeton-vl/lietorch/lietorch/run_tests.py
code served (permissive licence) · get_code("75ce2587ebcce9d6")
How it ran

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

import torch

torch.manual_seed(0)

# A is a 2D matrix, b is a 1D vector
# matv performs matrix-vector multiplication: A @ b
# A shape: (batch, matrix_rows, matrix_cols) or (matrix_rows, matrix_cols)
# b shape: (batch, matrix_cols) or (matrix_cols,)
# The function adds a dimension to b, multiplies, then removes it

A = torch.randn(2, 4, 3)  # batch=2, rows=4, cols=3
b = torch.randn(2, 3)     # batch=2, cols=3

ARGS = {
    "A": A,
    "b": b,
}
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
stack Ran
ran on a synthesized input
princeton-vl/lietorch/lietorch/groups.py
code served (permissive licence) · get_code("3305de8d30fbfbd2")
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 DataContainer. 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 simple stand-in class that mimics the structure expected by stack()
# The function accesses .data attribute and calls .__class__() constructor
class DataContainer:
    def __init__(self, data):
        self.data = data

# Create a list of DataContainer objects with small tensors
# Each tensor has shape (2, 4) - batch size 2, dimension 4
tensor1 = torch.randn(2, 4)
tensor2 = torch.randn(2, 4)
tensor3 = torch.randn(2, 4)

container1 = DataContainer(tensor1)
container2 = DataContainer(tensor2)
container3 = DataContainer(tensor3)

group_list = [container1, container2, container3]

# dim parameter: stack along dimension 0
dim = 0

ARGS = {
    "group_list": group_list,
    "dim": dim
}
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 DataContainer — a DataContainer output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataContainer 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_analytical_jacobian Not yet run
unverified (harvested, no recorded run)
princeton-vl/lietorch/lietorch/gradcheck.py
code served (permissive licence) · get_code("e94d1233772c60f6")
How it ran

no invocation recorded.

Also present in other papers' code

Of 8 samples lifted from this paper's linked repositories, 8 carry a function identity and 0 do not. Of the 8 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 8 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 address the problem of performing backpropagation for computation graphs involving 3D transformation groups SO(3), SE(3), and Sim(3). 3D transformation groups are widely used in 3D vision and robotics, but they do not form vector spaces and instead lie on smooth manifolds. The standard backpropagation approach, which embeds 3D transformations in Euclidean spaces, suffers from numerical difficulties. We introduce a new library, which exploits the group structure of 3D transformations and performs backpropagation in the tangent spaces of manifolds. We show that our approach is numerically more stable, easier to implement, and beneficial to a diverse set of tasks. Our plug-and-play PyTorch library is available at https: //github.com/princeton-vl/lietorch.

For agents

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

get_harvested_code_for_paper("2103.12032")
get_code_for_paper("2103.12032")
have("2103.12032")

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

curl https://syntology.ai/api/ran/2103.12032.json

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

[![Syntology run record](https://syntology.ai/api/ran/2103.12032.svg)](https://syntology.ai/paper/2103.12032)

Connect an agent — have() is free.