SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2002.09405 · ICML · 2020

Learning to Simulate Complex Physics with Graph Networks

Jure Leskovec, Alvaro Sanchez-Gonzalez, Jonathan Godwin, Tobias Pfaff, Rex Ying, Peter Battaglia

arXiv · PDF · Open in the Atlas

Code that ran

17 of 24 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (5 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 1 violated it and 15 have no contract verdict; 1 failure of our instrument; 7 no recorded run), measured 2026-11-21; 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/2002.09405.json · badge: /api/ran/2002.09405.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
diogodcarvalho/gns-sheet-model community (classifier label)
classifier basis not recorded on the link
8 ran · 2 no recorded run
geoelements/gns reimplementation (classifier label)
classifier basis: classifier_2025
4 ran · 4 no recorded run
htung0101/Physion-particles community
Papers with Code archive
4 ran · 1 no recorded run
m-ulmestrand/gcn-collisions community
Papers with Code archive
1 ran · 0 no recorded run
FunctionStatusWhere it lives
Decoder Ran
ran on a synthesized input
geoelements/gns/gns/learned_simulator.py
pointer only (licence: NOASSERTION) · get_code("cf515f1865cc938d")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, geoelements/gns (NOASSERTION). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

Encoder Ran
ran on a synthesized input
geoelements/gns/gns/learned_simulator.py
pointer only (licence: NOASSERTION) · get_code("31391488600b96f6")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, geoelements/gns (NOASSERTION). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

GCN Ran
ran on a synthesized input · constructed an object; no result was computed
m-ulmestrand/gcn-collisions/physics_gnn.py
code served (permissive licence) · get_code("6ddd7d872c5465bd")
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, GCN: 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
from torch.nn import Linear, Dropout
from torch_geometric.nn import GraphConv

torch.manual_seed(0)

# Create a minimal GCN instance to inspect its forward method
# The forward method expects:
# - x: node feature tensor of shape (num_nodes, n_features)
# - edge_index: edge indices of shape (2, num_edges)

# Define parameters
n_features = 4
n_outputs = 3
hidden_channels = 5

# Create sample data
num_nodes = 2
num_edges = 2

# Node features: (num_nodes, n_features)
x = torch.randn(num_nodes, n_features)

# Edge indices: (2, num_edges) - each column represents an edge (source, target)
edge_index = torch.tensor([[0, 1], [1, 0]], dtype=torch.long).t().contiguous()

ARGS = {
    'n_features': n_features,
    'n_outputs': n_outputs,
    'hidden_channels': hidden_channels
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, torch_geometric — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type GCN — 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 GCN 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
ParticleEncoder Ran
ran on a synthesized input · constructed an object; no result was computed
htung0101/Physion-particles/models/GNS.py
pointer only (licence: NONE) · get_code("455f17e20452375a")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, htung0101/Physion-particles when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

ParticlePredictor Ran
ran on a synthesized input · constructed an object; no result was computed
htung0101/Physion-particles/models/GNS.py
pointer only (licence: NONE) · get_code("52aa4c6dfc0568cf")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, htung0101/Physion-particles when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

Propagator Ran
ran on a synthesized input · constructed an object; no result was computed
htung0101/Physion-particles/models/GNS.py
pointer only (licence: NONE) · get_code("2ec08f9afc4481a9")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, htung0101/Physion-particles when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

RelationEncoder Ran
ran on a synthesized input · constructed an object; no result was computed
htung0101/Physion-particles/models/GNS.py
pointer only (licence: NONE) · get_code("0b5db1c272bb2421")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, htung0101/Physion-particles when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

build_mlp Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
geoelements/gns/gns/learned_simulator.py
pointer only (licence: NOASSERTION) · get_code("b7cb81770ba30c93")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, geoelements/gns (NOASSERTION). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_E Ran
ran on a synthesized input
diogodcarvalho/gns-sheet-model/sheet_model/utils.py
code served (permissive licence) · get_code("7bf66991b886466a")
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 2, 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 numpy as np

# Create small test arrays for the function
# x: position array with shape (batch, n_sheets)
x = np.random.randn(2, 4).astype(np.float32)

# v: velocity array with shape (batch, n_sheets)
v = np.random.randn(2, 4).astype(np.float32)

# x_eq: equilibrium position array with shape (batch, n_sheets)
x_eq = np.random.randn(2, 4).astype(np.float32)

ARGS = {
    "x": x,
    "v": v,
    "x_eq": x_eq,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 2, 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
get_E Ran
ran on a synthesized input
diogodcarvalho/gns-sheet-model/gns/utils.py
code served (permissive licence) · get_code("462dc344aeb7e330")
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 2, 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 jax.numpy as jnp
import numpy as np

# Create small test data
batch_size = 2
dim = 4

# X: position array, shape (batch_size, dim)
X = jnp.array(np.random.randn(batch_size, dim), dtype=jnp.float32)

# V: velocity array, shape (batch_size, dim)
V = jnp.array(np.random.randn(batch_size, dim), dtype=jnp.float32)

# X_eq: equilibrium positions, shape (batch_size, dim) or (dim,)
X_eq = jnp.array(np.random.randn(batch_size, dim), dtype=jnp.float32)

ARGS = {
    "X": X,
    "V": V,
    "X_eq": X_eq,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
jax, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 2, 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
get_dx_eq Ran
ran on a synthesized input
diogodcarvalho/gns-sheet-model/sheet_model/utils.py
code served (permissive licence) · get_code("4eb7506dd92f169f")
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)

# Based on the function signature and body, we need:
# n_sheets: int - number of sheets
# L: float - box size (optional, defaults to 1)

ARGS = {
    "n_sheets": 5,
    "L": 10.0
}
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
get_dx_eq Ran
ran on a synthesized input
diogodcarvalho/gns-sheet-model/gns/utils.py
code served (permissive licence) · get_code("074d15f05b987c6d")
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

ARGS = {
    "n_sheets": 5,
    "L": 1.0,
}
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
get_nodes Ran
ran on a synthesized input
diogodcarvalho/gns-sheet-model/gns/graph.py
code served (permissive licence) · get_code("d1ea23cc7d33967b")
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 jax.numpy as jnp
import numpy as np

torch_seed = np.random.seed(0)

x = jnp.array([1.0, 2.0, 3.0, 4.0])
v = jnp.array([0.1, 0.2, 0.3, 0.4])
x_eq = jnp.array([0.5, 1.0, 1.5, 2.0])
dx_eq = 0.5

ARGS = {
    'x': x,
    'v': v,
    'x_eq': x_eq,
    'dx_eq': dx_eq
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
jax, numpy — 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
get_x_eq Ran
ran on a synthesized input
diogodcarvalho/gns-sheet-model/sheet_model/utils.py
code served (permissive licence) · get_code("72df1c0ddb18d5ee")
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 ndarray, ndim 1, dtype family float, shape (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 numpy as np

# Set up arguments for get_x_eq
ARGS = {
    "n_sheets": 5,
    "L": 1.0
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family float, shape (5,) — shape (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
get_x_eq Ran
ran on a synthesized input
diogodcarvalho/gns-sheet-model/gns/utils.py
code served (permissive licence) · get_code("95d152d025eca082")
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 ArrayImpl, ndim 1, dtype family float, shape (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 jax.numpy as jnp

ARGS = {
    "n_sheets": 5,
    "L": 1.0,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
jax — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ArrayImpl, ndim 1, dtype family float, shape (5,) — shape (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
sliding_windows Ran
ran on a synthesized input
diogodcarvalho/gns-sheet-model/gns/preprocess.py
code served (permissive licence) · get_code("04256ab5e36f1f1b")
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 ArrayImpl, ndim 3, dtype family float, shape (8, 3, 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 jax.numpy as jnp

# Create a small time series array
# Shape: (time_steps, features)
X = jnp.arange(20).reshape(10, 2).astype(jnp.float32)

# Window size
w_size = 3

ARGS = {
    "X": X,
    "w_size": w_size,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
jax — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ArrayImpl, ndim 3, dtype family float, shape (8, 3, 2) — shape (8, 3, 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
time_diff Ran
ran, violated the contract we drafted
geoelements/gns/gns/learned_simulator.py
pointer only (licence: NOASSERTION) · get_code("b6a7b4c1838c5223")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, geoelements/gns (NOASSERTION). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

EncodeProcessDecode Not yet run
unverified (harvested, no recorded run)
geoelements/gns/gns/learned_simulator.py
pointer only (licence: NOASSERTION) · get_code("70132227e453c957")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, geoelements/gns (NOASSERTION). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

GNS Not yet run
unverified (harvested, no recorded run)
htung0101/Physion-particles/models/GNS.py
pointer only (licence: NONE) · get_code("7404c27d0f9f05b1")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, htung0101/Physion-particles when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

InteractionNetwork Not yet run
unverified (harvested, no recorded run)
geoelements/gns/gns/learned_simulator.py
pointer only (licence: NOASSERTION) · get_code("eb8cb787e8ea787c")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, geoelements/gns (NOASSERTION). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

LearnedSimulator Not yet run
unverified (harvested, no recorded run)
geoelements/gns/gns/learned_simulator.py
pointer only (licence: NOASSERTION) · get_code("ceb019d0bc5e6951")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, geoelements/gns (NOASSERTION). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

Processor Not yet run
unverified (harvested, no recorded run)
geoelements/gns/gns/learned_simulator.py
pointer only (licence: NOASSERTION) · get_code("4e185c59b88aa7c1")
How it ran

The body of this sample is withheld: GitHub could not classify the licence of this paper's repository, geoelements/gns (NOASSERTION). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

add_guards Not yet run
unverified (harvested, no recorded run)
diogodcarvalho/gns-sheet-model/gns/graph.py
code served (permissive licence) · get_code("1c5a15d5ff809dc8")
How it ran

no invocation recorded.

get_guards Not yet run
unverified (harvested, no recorded run)
diogodcarvalho/gns-sheet-model/gns/graph.py
code served (permissive licence) · get_code("ddb2cf246db8631b")
How it ran

no invocation recorded.

Also present in other papers' code

Of 24 samples lifted from this paper's linked repositories, 24 carry a function identity and 0 do not. Of the 24 distinct functions with an identity, 1 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 1 later) and 23 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.

FunctionAlso present inEarliest-datedOthers named
build_mlp
brick of this paper's assembly: helper
1 other paper
1 later
2401.09323 (2024-01)
BENO: Boundary-embedded Neural Operators for Elliptic PDEs

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 2026-11-21.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Here we present a machine learning framework and model implementation that can learn to simulate a wide variety of challenging physical domains, involving fluids, rigid solids, and deformable materials interacting with one another. Our framework-which we term "Graph Network-based Simulators" (GNS)-represents the state of a physical system with particles, expressed as nodes in a graph, and computes dynamics via learned message-passing. Our results show that our model can generalize from single-timestep predictions with thousands of particles during training, to different initial conditions, thousands of timesteps, and at least an order of magnitude more particles at test time. Our model was robust to hyperparameter choices across various evaluation metrics: the main determinants of long-term performance were the number of message-passing steps, and mitigating the accumulation of error by corrupting the training data with noise. Our GNS framework advances the state-of-the-art in learned physical simulation, and holds promise for solving a wide range of complex forward and inverse problems.

For agents

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

get_harvested_code_for_paper("2002.09405")
get_code_for_paper("2002.09405")
have("2002.09405")

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

curl https://syntology.ai/api/ran/2002.09405.json

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

[![Syntology run record](https://syntology.ai/api/ran/2002.09405.svg)](https://syntology.ai/paper/2002.09405)

Connect an agent — have() is free.