SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2602.18482 · 2026

Boltzmann Generators for Condensed Matter via Riemannian Flow Matching

Frank Noé, Leon Klein, Jutta Rogal, Emil Hoffmann, Maximilian Schebek

arXiv · PDF · Open in the Atlas

Code that ran

5 of 6 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (4 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 4 have no contract verdict; 1 failure 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/2602.18482.json · badge: /api/ran/2602.18482.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
emil-ho/bg-cnf-fm-condensed-matter community
Papers with Code archive
5 ran · 1 no recorded run
FunctionStatusWhere it lives
ExponentialTimeEmbedding Ran
ran on a synthesized input · constructed an object; no result was computed
emil-ho/bg-cnf-fm-condensed-matter/src/vecfield.py
code served (permissive licence) · get_code("ba8cac6526b4ab6c")
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 an instance of its own class, ExponentialTimeEmbedding: the run constructed the object and did not call it, so no result was computed. 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
import math

torch.manual_seed(0)

# The function is ExponentialTimeEmbedding.__init__ and forward
# __init__ parameters: emb_dim, max_period, max_freq
# forward parameter: t (torch.Tensor)

# Create a simple test case
ARGS = {
    'emb_dim': 4,
    'max_period': 100.0,
    'max_freq': 32.0,
}
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 ExponentialTimeEmbedding — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a ExponentialTimeEmbedding 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
GaussianSmearing Ran
ran on a synthesized input · constructed an object; no result was computed
emil-ho/bg-cnf-fm-condensed-matter/src/vecfield.py
code served (permissive licence) · get_code("1a4c16fe71e8300a")
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 an instance of its own class, GaussianSmearing: the run constructed the object and did not call it, so no result was computed. 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)

# The GaussianSmearing class takes the following parameters:
# - rmin: float (minimum radius)
# - rmax: float (maximum radius)
# - num_rbf: int (number of radial basis functions)
# - trainable: bool (whether parameters are trainable)

# Create a small distance tensor to use as input
dist_input = torch.linspace(0.5, 4.5, 2)

ARGS = {
    'rmin': 0.0,
    'rmax': 5.0,
    'num_rbf': 50,
    'trainable': 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 GaussianSmearing — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a GaussianSmearing 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
PEGNN Ran
ran on a synthesized input · constructed an object; no result was computed
emil-ho/bg-cnf-fm-condensed-matter/src/vecfield.py
code served (permissive licence) · get_code("a268d29162914dcd")
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 an instance of its own class, PEGNN: the run constructed the object and did not call it, so no result was computed. 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 minimal valid arguments for PEGNN
ARGS = {
    'in_node_nf': 4,
    'in_edge_nf': 2,
    'hidden_nf': 8,
    'box_edges': torch.tensor([1.0, 1.0, 1.0]),
    'act_fn': nn.SiLU(),
    'n_layers': 2,
    'norm_diff': True,
    'out_node_nf': 3,
    'coords_range': 15.0,
    'agg': 'sum',
    'gain': 0.001,
    'distance_expansion': nn.Identity(),
}
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 PEGNN — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a PEGNN 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
PE_GCL Ran
ran on a synthesized input · constructed an object; no result was computed
emil-ho/bg-cnf-fm-condensed-matter/src/vecfield.py
code served (permissive licence) · get_code("c5fa8707e25e2b9c")
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 an instance of its own class, PE_GCL: the run constructed the object and did not call it, so no result was computed. 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)

# Small stand-in for distance_expansion with num_rbf attribute
class SimpleRBF(nn.Module):
    def __init__(self, num_rbf=8):
        super().__init__()
        self.num_rbf = num_rbf
    
    def forward(self, x):
        # Expand input to num_rbf dimensions
        return x.repeat(1, self.num_rbf)

# Setup parameters
num_nodes = 2
num_edges = 3
input_nf = 4
output_nf = 5
hidden_nf = 6
spatial_dim = 3
edges_in_d = 2

# Create box_edges buffer
box_edges = torch.ones(spatial_dim) * 10.0

# Create node features
h = torch.randn(num_nodes, input_nf)

# Create node coordinates
x = torch.randn(num_nodes, spatial_dim)

# Create edge indices (2, num_edges)
edge_index = torch.tensor([[0, 1, 0], [1, 0, 1]], dtype=torch.long)

# Create edge attributes
edge_attr = torch.randn(num_edges, edges_in_d)

# Create node attributes
node_attr = torch.randn(num_nodes, 3)

# Create distance expansion
distance_expansion = SimpleRBF(num_rbf=8)

# Create activation function
act_fn = nn.SiLU()

ARGS = {
    'box_edges': box_edges,
    'input_nf': input_nf,
    'output_nf': output_nf,
    'hidden_nf': hidden_nf,
    'edges_in_d': edges_in_d,
    'act_fn': act_fn,
    'agg': 'sum',
    'gain': 0.001,
    'norm_diff': True,
    'distance_expansion': distance_expansion,
}
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 PE_GCL — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a PE_GCL 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
scatter_sum Ran
ran; our contract draft was wrong, not the code
emil-ho/bg-cnf-fm-condensed-matter/src/vecfield.py
code served (permissive licence) · get_code("b192acda87d4c8d7")
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 (3, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# Construct arguments for scatter_sum
# src: source tensor with shape (batch, features)
src = torch.randn(2, 4)

# index: indices for scatter operation, must have same length as src
# Values should be in range [0, dim_size)
index = torch.tensor([0, 1], dtype=torch.long)

# dim_size: size of the output dimension
dim_size = 3

ARGS = {
    "src": src,
    "index": index,
    "dim_size": dim_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 Tensor, ndim 2, dtype family float, shape (3, 4) — shape (3, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
PEGNN_dynamics Not yet run
unverified (harvested, no recorded run)
emil-ho/bg-cnf-fm-condensed-matter/src/vecfield.py
code served (permissive licence) · get_code("df87f0e8d958b50f")
How it ran

no invocation recorded.

Also present in other papers' code

Of 6 samples lifted from this paper's linked repositories, 6 carry a function identity and 0 do not. Of the 6 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 6 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

Sampling equilibrium distributions is fundamental to statistical mechanics. While flow matching has emerged as scalable state-of-the-art paradigm for generative modeling, its potential for equilibrium sampling in condensed-phase systems remains largely unexplored. We address this by incorporating the periodicity inherent to these systems into continuous normalizing flows using Riemannian flow matching. The high computational cost of exact density estimation intrinsic to continuous normalizing flows is mitigated by using Hutchinson's trace estimator, utilizing a crucial bias-correction step based on cumulant expansion to render the stochastic estimates suitable for rigorous thermodynamic reweighting. Our approach is validated on monatomic ice, demonstrating the ability to train on systems of unprecedented size and obtain highly accurate free energy estimates without the need for traditional multistage estimators.

For agents

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

get_harvested_code_for_paper("2602.18482")
get_code_for_paper("2602.18482")
have("2602.18482")

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

curl https://syntology.ai/api/ran/2602.18482.json

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

[![Syntology run record](https://syntology.ai/api/ran/2602.18482.svg)](https://syntology.ai/paper/2602.18482)

Connect an agent — have() is free.