SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2405.05079 · ECCV · 2024

Power Variable Projection for Initialization-Free Large-Scale Bundle Adjustment

Daniel Cremers, Simon Weber, Je Hong

arXiv · PDF · Open in the Atlas

Code that ran

6 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; 1 honoured the contract we drafted, 1 violated it and 0 have no contract verdict; 4 failures of our instrument; 2 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/2405.05079.json · badge: /api/ran/2405.05079.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
tum-vision/povar canonical (classifier label)
classifier basis: census
6 ran · 2 no recorded run
FunctionStatusWhere it lives
compute_caching_hash Ran
ran; our contract draft was wrong, not the code
tum-vision/povar/python/rootba/experiments.py
code served (permissive licence) · get_code("90ee2642ab24e622")
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 json
import hashlib

# The function compute_caching_hash takes a dictionary and converts it to JSON,
# then computes a SHA1 hash. We need to provide a dictionary that is JSON-serializable.

ARGS = {
    "d": {
        "experiment_name": "test_experiment",
        "learning_rate": 0.001,
        "batch_size": 32,
        "num_epochs": 10,
        "model_type": "transformer",
        "seed": 42,
        "nested_config": {
            "layer_count": 3,
            "hidden_dim": 256,
            "dropout": 0.1
        },
        "flags": [True, False, True],
        "values": [1, 2, 3, 4, 5]
    }
}
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 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
copy_subdict Ran
ran; our contract draft was wrong, not the code
tum-vision/povar/python/rootba/util.py
code served (permissive licence) · get_code("88bee645d9cfe9a9")
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 dict. 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)

# Analyze the function: copy_subdict(d, keys)
# - d: a dictionary
# - keys: an iterable of keys to filter from d
# The function iterates through keys and copies matching entries from d

# Create a source dictionary with various types of values
d = {
    'a': 42,
    'b': 'hello',
    'c': [1, 2, 3],
    'd': torch.randn(2, 4),
    'e': np.array([1.0, 2.0, 3.0]),
    'f': {'nested': 'dict'},
}

# Create a list of keys, some that exist in d and some that don't
keys = ['a', 'c', 'e', 'nonexistent', 'z']

ARGS = {
    'd': d,
    'keys': keys,
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict 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
detect_log_filepath Ran
ran; our contract draft was wrong, not the code
tum-vision/povar/python/rootba/log.py
code served (permissive licence) · get_code("71fa49405353db9d")
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 os
import tempfile

# Create a temporary directory for testing
tmpdir = tempfile.mkdtemp()

# Create test files to make the function work properly
test_basename = "test_log"
ubjson_path = os.path.join(tmpdir, test_basename + ".ubjson")
with open(ubjson_path, 'w') as f:
    f.write("test")

ARGS = {
    'dir': tmpdir,
    'basename': test_basename
}
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 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
get_linear_solver_time Ran
ran, honoured the contract we drafted
tum-vision/povar/python/rootba/metric.py
code served (permissive licence) · get_code("27c1ffb436390cc9")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned float, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

class StaticSolver:
    def __init__(self):
        self.solver = self
        self.linear_solver_time_in_seconds = 1.5

class CeresLinearSolver:
    def __init__(self):
        self._static = StaticSolver()
    
    def is_ceres(self):
        return True

class NonCeresLinearSolver:
    def __init__(self):
        self.stage1_time = torch.tensor([0.1, 0.2, 0.3])
        self.stage2_time = torch.tensor([0.15, 0.25])
        self.step_solver_time = torch.tensor([0.05, 0.1, 0.15, 0.2])
        self.back_substitution_time = torch.tensor([0.02, 0.03, 0.04])
    
    def is_ceres(self):
        return False

ARGS = {
    "l": CeresLinearSolver()
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
load_json_if_exists Ran
ran; our contract draft was wrong, not the code
tum-vision/povar/python/rootba/util.py
code served (permissive licence) · get_code("5e5b2363bf21d50b")
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 Munch. 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 os
import json
import tempfile
from munch import munchify

# Create a temporary directory and a test JSON file
temp_dir = tempfile.mkdtemp()
test_filepath = os.path.join(temp_dir, "test_config.json")

# Write a simple JSON file for testing
test_data = {
    "model": "resnet50",
    "learning_rate": 0.001,
    "epochs": 10,
    "batch_size": 32
}

with open(test_filepath, 'w') as f:
    json.dump(test_data, f)

# Define ARGS with the filepath parameter
ARGS = {
    "filepath": test_filepath
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
munch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Munch — a Munch output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Munch 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
version_less Ran
ran, violated the contract we drafted
tum-vision/povar/python/rootba/experiments.py
code served (permissive licence) · get_code("f692ff215fbcca14")
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 bool, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

def version_less(vstr1, vstr2):
    """Order for sorting versions in the format a.b.c"""
    return vstr1.split('.') < vstr2.split('.')

ARGS = {
    'vstr1': '1.2.3',
    'vstr2': '1.2.4'
}
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 bool, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
load_bal Not yet run
unverified (harvested, no recorded run)
tum-vision/povar/python/rootba/fp_tree_prototype/fp_tree.py
code served (permissive licence) · get_code("94927b9692321919")
How it ran

no invocation recorded.

metrics_from_config Not yet run
unverified (harvested, no recorded run)
tum-vision/povar/python/rootba/metric.py
code served (permissive licence) · get_code("8228f65e8fbb26f2")
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

Most Bundle Adjustment (BA) solvers like the Levenberg-Marquard algorithm require a good initialization. Instead, initializationfree BA remains a largely uncharted territory. The under-explored Variable Projection algorithm (VarPro) exhibits a wide convergence basin even without initialization. Coupled with object space error formulation, recent works have shown its ability to solve small-scale initializationfree bundle adjustment problem. To make such initialization-free BA approaches scalable, we introduce Power Variable Projection (PoVar), extending a recent inverse expansion method based on power series. Importantly, we link the power series expansion to Riemannian manifold optimization. This projective framework is crucial to solve large-scale bundle adjustment problems without initialization. Using the real-world BAL dataset, we experimentally demonstrate that our solver achieves state-of-the-art results in terms of speed and accuracy. To our knowledge, this work is the first to address the scalability of BA without initialization opening new venues for initialization-free structure-from-motion.

For agents

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

get_harvested_code_for_paper("2405.05079")
get_code_for_paper("2405.05079")
have("2405.05079")

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

curl https://syntology.ai/api/ran/2405.05079.json

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

[![Syntology run record](https://syntology.ai/api/ran/2405.05079.svg)](https://syntology.ai/paper/2405.05079)

Connect an agent — have() is free.