SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2003.04630 · ICLR · 2020

Lagrangian Neural Networks

Peter Battaglia, Miles Cranmer, Sam Greydanus, Stephan Hoyer, David Spergel, Shirley Ho

arXiv · PDF · Open in the Atlas

Code that ran

3 of 3 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 0 have no contract verdict; 3 failures of our instrument; 0 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/2003.04630.json · badge: /api/ran/2003.04630.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
milescranmer/lagrangian_nns canonical (classifier label)
classifier basis: classifier_2025
3 ran · 0 no recorded run
FunctionStatusWhere it lives
lagrangian_eom Ran
ran; our contract draft was wrong, not the code
milescranmer/lagrangian_nns/lnn/core.py
code served (permissive licence) · get_code("41d946e341b9742a")
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) and returned ArrayImpl, ndim 1, dtype family float, shape (8,) 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
import jax.numpy as jnp
from functools import partial

# Define a simple Lagrangian function
# The function signature is lagrangian(q, q_t) -> scalar
# It needs to support:
# - hessian(lagrangian, 1)(q, q_t) - hessian w.r.t. second arg (q_t)
# - grad(lagrangian, 0)(q, q_t) - gradient w.r.t. first arg (q)
# - jacobian(jacobian(lagrangian, 1), 0)(q, q_t) - second derivative
def simple_lagrangian(q, q_t):
    """
    Simple quadratic Lagrangian: L = (1/2) * q_t^T @ q_t - (1/2) * q^T @ q
    This represents a free particle or harmonic oscillator-like system.
    """
    kinetic = 0.5 * jnp.sum(q_t ** 2)
    potential = 0.5 * jnp.sum(q ** 2)
    return kinetic - potential

# Create state vector: [q, q_t] where q has 4 components and q_t has 4 components
# Total state dimension is 8
q = jnp.array([0.1, 0.2, 0.3, 0.4])  # 4-dimensional position
q_t = jnp.array([0.05, 0.1, 0.15, 0.2])  # 4-dimensional velocity
state = jnp.concatenate([q, q_t])  # shape (8,)

ARGS = {
    "lagrangian": simple_lagrangian,
    "state": state,
    "t": None
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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 (8,) — shape (8,) 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
raw_lagrangian_eom Ran
ran; our contract draft was wrong, not the code
milescranmer/lagrangian_nns/lnn/core.py
code served (permissive licence) · get_code("a223f0e9b39d7f27")
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) and returned ArrayImpl, ndim 1, dtype family float, shape (8,) 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
import jax.numpy as jnp
from functools import partial

# Define a simple Lagrangian function
# The function expects lagrangian(q, q_t) to return a scalar
# It will be differentiated w.r.t. q (arg 0) and q_t (arg 1)
def lagrangian(q, q_t):
    """Simple harmonic oscillator Lagrangian: L = 0.5*q_t^2 - 0.5*q^2"""
    return 0.5 * jnp.sum(q_t**2) - 0.5 * jnp.sum(q**2)

# Create initial state: [q, q_t] concatenated
# q has 4 dimensions, q_t has 4 dimensions
q_init = jnp.array([0.1, 0.2, 0.3, 0.4])
q_t_init = jnp.array([0.5, 0.6, 0.7, 0.8])
state = jnp.concatenate([q_init, q_t_init])

# Optional time parameter (can be None)
t = None

ARGS = {
    "lagrangian": lagrangian,
    "state": state,
    "t": t,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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 (8,) — shape (8,) 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
unconstrained_eom Ran
ran; our contract draft was wrong, not the code
milescranmer/lagrangian_nns/lnn/core.py
code served (permissive licence) · get_code("858032c958b83e3d")
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) and returned ArrayImpl, ndim 1, dtype family float, shape (8,) 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
import jax.numpy as jnp
import numpy as np

class SimpleModel:
    """Stand-in model that takes q and q_t and returns their concatenation."""
    def __call__(self, q, q_t):
        # Returns time derivative of state (dq/dt, dq_t/dt)
        # Should return same shape as input state
        return jnp.concatenate([q_t, -q])

# Create a simple model instance
model = SimpleModel()

# Create state: [q, q_t] where q has shape (4,) and q_t has shape (4,)
# Total state shape is (8,)
q = jnp.array([1.0, 2.0, 3.0, 4.0])
q_t = jnp.array([0.1, 0.2, 0.3, 0.4])
state = jnp.concatenate([q, q_t])

# Time parameter (optional, not used in function but part of signature)
t = 0.0

ARGS = {
    "model": model,
    "state": state,
    "t": t,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
jax, numpy — 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 (8,) — shape (8,) 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

Also present in other papers' code

Of 3 samples lifted from this paper's linked repositories, 3 carry a function identity and 0 do not. Of the 3 distinct functions with an identity, 3 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 0 the same month, 0 later) and 0 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
lagrangian_eom 2 other papers
2 dated earlier
1906.01563 (2019-06)
Hamiltonian Neural Networks
1907.04490 (2019-07)
Deep Lagrangian Networks: Using Physics as Model Prior for Deep Learning
raw_lagrangian_eom 2 other papers
2 dated earlier
1906.01563 (2019-06)
Hamiltonian Neural Networks
1907.04490 (2019-07)
Deep Lagrangian Networks: Using Physics as Model Prior for Deep Learning
unconstrained_eom 2 other papers
2 dated earlier
1906.01563 (2019-06)
Hamiltonian Neural Networks
1907.04490 (2019-07)
Deep Lagrangian Networks: Using Physics as Model Prior for Deep Learning

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

Accurate models of the world are built upon notions of its underlying symmetries. In physics, these symmetries correspond to conservation laws, such as for energy and momentum. Yet even though neural network models see increasing use in the physical sciences, they struggle to learn these symmetries. In this paper, we propose Lagrangian Neural Networks (LNNs), which can parameterize arbitrary Lagrangians using neural networks. In contrast to models that learn Hamiltonians, LNNs do not require canonical coordinates, and thus perform well in situations where canonical momenta are unknown or difficult to compute. Unlike previous approaches, our method does not restrict the functional form of learned energies and will produce energy-conserving models for a variety of tasks. We test our approach on a double pendulum and a relativistic particle, demonstrating energy conservation where a baseline approach incurs dissipation and modeling relativity without canonical coordinates where a Hamiltonian approach fails. Finally, we show how this model can be applied to graphs and continuous systems using a Lagrangian Graph Network, and demonstrate it on the 1D wave equation.

For agents

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

get_harvested_code_for_paper("2003.04630")
get_code_for_paper("2003.04630")
have("2003.04630")

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

curl https://syntology.ai/api/ran/2003.04630.json

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

[![Syntology run record](https://syntology.ai/api/ran/2003.04630.svg)](https://syntology.ai/paper/2003.04630)

Connect an agent — have() is free.