SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2111.02434 · NeurIPS · 2021

Hamiltonian Dynamics with Non-Newtonian Momentum for Rapid Sampling

Greg Steeg, Aram Galstyan

arXiv · PDF · Open in the Atlas

Code that ran

12 of 14 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; 2 honoured the contract we drafted, 0 violated it and 5 have no contract verdict; 5 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/2111.02434.json · badge: /api/ran/2111.02434.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
eigenfoo/littlemcmc canonical (classifier label)
classifier basis not recorded on the link
5 ran · 2 no recorded run
gregversteeg/esh_dynamics canonical (classifier label)
classifier basis: classifier_2025
5 ran · 0 no recorded run
wgrathwohl/JEM community
Papers with Code archive
1 ran · 0 no recorded run
copy not recorded community
not in the paper's declared repositories
1 ran · 0 no recorded run
FunctionStatusWhere it lives
billiards Ran
ran, honoured the contract we drafted
gregversteeg/esh_dynamics/esh/core.py
code served (permissive licence) · get_code("246b3036cf5c3d38")
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 3, 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 torch as t

t.manual_seed(0)

# Create tensors for position and velocity
# Both need to be float tensors that support abs(), max(), sign(), where() operations
# x: position tensor - should have some values outside [-1, 1] to trigger the while loop
# v: velocity tensor - same shape as x

x = t.tensor([
    [1.5, -0.3, 0.8, -1.2],
    [0.2, -1.8, 0.5, 0.9]
], dtype=t.float32)

v = t.tensor([
    [0.1, -0.2, 0.3, -0.4],
    [0.5, -0.1, 0.2, 0.3]
], dtype=t.float32)

ARGS = {
    "x": x,
    "v": v,
}
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, ndim 3, 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
esh_leap_step Ran
ran; our contract draft was wrong, not the code
gregversteeg/esh_dynamics/esh/core.py
code served (permissive licence) · get_code("ad0a2d24d12e2af3")
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 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 as t
import torch.nn as nn

t.manual_seed(0)

# Define a simple energy function (scalar output per batch item)
class SimpleEnergy(nn.Module):
    def forward(self, xs):
        # Return a scalar per batch item
        return (xs ** 2).sum(dim=tuple(range(1, len(xs.shape))))

# Create small tensors
batch_size = 2
spatial_dims = (4,)

# xs: position, shape (batch_size, *spatial_dims)
xs = t.randn(batch_size, *spatial_dims)

# u: unit vector velocity, must be normalized
u = t.randn(batch_size, *spatial_dims)
u = u / u.flatten(start_dim=1).norm(dim=1, keepdim=True).view(-1, *([1] * len(spatial_dims)))

# rs: tracks dynamics for velocity magnitude
rs = t.zeros(batch_size)

# energy: a callable that takes xs and returns scalar per batch item
energy = SimpleEnergy()

# epsilon: step size
epsilon = 0.1

# Optional parameters
grad = None  # Will be computed if None
energy_scale = 1.0
debug = False
constraint = False

ARGS = {
    "xs": xs,
    "u": u,
    "rs": rs,
    "energy": energy,
    "epsilon": epsilon,
    "grad": grad,
    "energy_scale": energy_scale,
    "debug": debug,
    "constraint": constraint,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
IPython, 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
esh_leap_step Ran
ran; our contract draft was wrong, not the code
gregversteeg/esh_dynamics/esh/core.py
code served (permissive licence) · get_code("90eed7b2ef9c7cc9")
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 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 as t
import torch.nn as nn

t.manual_seed(0)

# Define a simple energy function as an nn.Module
class SimpleEnergy(nn.Module):
    def forward(self, x):
        # Return a scalar energy per batch item
        return (x ** 2).flatten(start_dim=1).sum(dim=1)

# Create batch_size=2, spatial_dim=4
batch_size = 2
spatial_dim = 4

# Initialize position tensor
xs = t.randn(batch_size, spatial_dim) * 0.5  # Keep in reasonable range

# Initialize unit vector velocity (must be normalized)
u_raw = t.randn(batch_size, spatial_dim)
u = u_raw / u_raw.flatten(start_dim=1).norm(dim=1, keepdim=True).view(batch_size, 1)

# Initialize rs (velocity magnitude tracking)
rs = t.zeros(batch_size)

# Create energy function
energy = SimpleEnergy()

# Set epsilon (step size)
epsilon = 0.1

ARGS = {
    'xs': xs,
    'u': u,
    'rs': rs,
    'energy': energy,
    'epsilon': epsilon,
    'grad': None,
    'energy_scale': 1.0,
    'debug': False,
    'constraint': False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
IPython, 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
get_sample_q Ran
ran; our contract draft was wrong, not the code
wgrathwohl/JEM/train_wrn_ebm.py
code served (permissive licence) · get_code("29ae6da71890a0da")
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 function. 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 as t
import torch.nn as nn

t.manual_seed(0)

# Build a minimal model that the function uses
class DummyEBM(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(3 * 32 * 32, 1)
    
    def forward(self, x, y=None):
        # x shape: (bs, 3, 32, 32)
        bs = x.shape[0]
        x_flat = x.view(bs, -1)
        return self.linear(x_flat)

# Build args object with all required attributes
class Args:
    def __init__(self):
        self.n_steps = 2
        self.batch_size = 2
        self.sgld_lr = 0.1
        self.sgld_std = 0.01
        self.reinit_freq = 0.05
        self.uncond = True
        self.n_classes = 4

args = Args()

# Create a small replay buffer (list of tensors)
replay_buffer = []

# Create the model
model = DummyEBM()

# Set device to CPU
device = t.device('cpu')

ARGS = {
    'args': args,
    'device': device
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
init_random Ran
ran, honoured the contract we drafted
this paper's copy was not recorded; identical code first harvested from wgrathwohl/JEM
pointer only · get_code("77d3d7b7c1ab83ce")
How it ran

The body of this sample is withheld: this paper's copy of the code was not recorded (its link names no repository or source path), so no licence was read for it. 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.

isquadpotential Ran
ran on a synthesized input
eigenfoo/littlemcmc/littlemcmc/quadpotential.py
code served (permissive licence) · get_code("916d7e0ce470f1bb")
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
import numpy as np


class ConcreteQuadPotential:
    """Concrete implementation of QuadPotential for testing."""
    
    def __init__(self, ndim=4):
        self.ndim = ndim
        self._mass_matrix = np.eye(ndim)
    
    def velocity(self, x, out=None):
        """Compute velocity."""
        return np.dot(self._mass_matrix, x)
    
    def energy(self, x, velocity=None):
        """Compute kinetic energy."""
        if velocity is None:
            velocity = self.velocity(x)
        return 0.5 * np.dot(velocity, velocity)
    
    def random(self):
        """Draw random value."""
        return np.random.randn(self.ndim)
    
    def velocity_energy(self, x, v_out):
        """Compute velocity and energy."""
        vel = self.velocity(x)
        energy = self.energy(x, vel)
        return vel, energy
    
    def update(self, sample, grad, tune):
        """Update potential during tuning."""
        pass
    
    def raise_ok(self, vmap=None):
        """Check if mass matrix is valid."""
        return None
    
    def reset(self):
        """Reset adaptation."""
        pass


class QuadPotential:
    """Base class for quadpotentials."""

    def velocity(self, x, out=None):
        """Compute the current velocity at a position in parameter space."""
        raise NotImplementedError('Abstract method')

    def energy(self, x, velocity=None):
        """Compute kinetic energy at a position in parameter space."""
        raise NotImplementedError('Abstract method')

    def random(self):
        """Draw random value from QuadPotential."""
        raise NotImplementedError('Abstract method')

    def velocity_energy(self, x, v_out):
        """Compute velocity and return kinetic energy at a position in parameter space."""
        raise NotImplementedError('Abstract method')

    def update(self, sample, grad, tune):
        """Inform the potential about a new sample during tuning."""
        pass

    def raise_ok(self, vmap=None):
        """Check if the mass matrix is ok, and raise ValueError if not."""
        return None

    def reset(self):
        """Reset quadpotential adaptation routine."""
        pass


def isquadpotential(value):
    """Check whether an object might be a QuadPotential object."""
    return isinstance(value, QuadPotential)


# Create a concrete instance that inherits from QuadPotential
class TestQuadPotential(QuadPotential):
    """Test implementation of QuadPotential."""
    
    def __init__(self, ndim=4):
        self.ndim = ndim
        self._mass_matrix = np.eye(ndim)
    
    def velocity(self, x, out=None):
        return np.dot(self._mass_matrix, x)
    
    def energy(self, x, velocity=None):
        if velocity is None:
            velocity = self.velocity(x)
        return 0.5 * np.dot(velocity, velocity)
    
    def random(self):
        return np.random.randn(self.ndim)
    
    def velocity_energy(self, x, v_out):
        vel = self.velocity(x)
        energy = self.energy(x, vel)
        return vel, energy


torch.manual_seed(0)

# Create test instances
quad_potential_instance = TestQuadPotential(ndim=4)
non_quad_potential = "not a quad potential"
none_value = None

ARGS = {
    "value": quad_potential_instance,
}
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 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
jarzynski_sample Ran
ran; our contract draft was wrong, not the code
gregversteeg/esh_dynamics/esh/core.py
code served (permissive licence) · get_code("5420df8af89cc7cd")
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 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 as t
import torch.nn as nn

t.manual_seed(0)

# Define a simple energy function as an nn.Module
class SimpleEnergy(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 1)
    
    def forward(self, x):
        # x shape: (batch_size, 4)
        batch_size = x.shape[0]
        x_flat = x.view(batch_size, -1)
        return self.linear(x_flat).squeeze(-1)

# Create energy model
energy_fn = SimpleEnergy()

# Initial batch of states: batch_size=2, feature_dim=4
x_init = t.randn(2, 4)

# Number of steps
n_steps = 3

# Step size
epsilon = 0.1

# Energy scale (temperature)
energy_scale = 1.0

# Initial energy
E0 = 0.0

# Initial velocity (optional, can be None to let function generate)
u0 = None

ARGS = {
    "energy": energy_fn,
    "x": x_init,
    "n_steps": n_steps,
    "epsilon": epsilon,
    "energy_scale": energy_scale,
    "E0": E0,
    "u0": u0,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
IPython, 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
log1mexp_numpy Ran
ran on a synthesized input
eigenfoo/littlemcmc/littlemcmc/math.py
code served (permissive licence) · get_code("b0fc8c54e4f85a50")
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 (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 numpy as np

# The function takes a single parameter 'x' which is a float
# However, looking at the implementation, it uses np.where and numpy functions
# that work with arrays, so x can be a scalar or array-like

# Create test values - mix of scalars and arrays
# Test with values both below and above the threshold of 0.683
x_value = np.array([0.5, 0.683, 1.0, 2.0])

ARGS = {
    "x": x_value
}
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 (4,) — shape (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
logbern Ran
ran on a synthesized input
eigenfoo/littlemcmc/littlemcmc/math.py
code served (permissive licence) · get_code("ce692f1e057145d0")
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 numpy as np
import numpy.random as nr

# Set seed for reproducibility
np.random.seed(0)

# The function logbern takes a log_p parameter (float) and performs a Bernoulli trial
# It checks that log_p is not NaN, then compares log(uniform()) with log_p
# We need to provide a valid log probability value

# A valid log probability should be a negative float (since probabilities are between 0 and 1)
# For example, log(0.5) ≈ -0.693
log_p_value = np.log(0.5)  # This gives approximately -0.693, a valid log probability

ARGS = {
    "log_p": log_p_value
}
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 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
logdiffexp_numpy Ran
ran on a synthesized input
eigenfoo/littlemcmc/littlemcmc/math.py
code served (permissive licence) · get_code("5b291226145b9ec1")
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 float64, ndim 0, 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

# Test with scalar float values
a = 5.0
b = 3.0

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
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, which disagrees with the fixture run
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
rebuild_exc Ran
ran on a synthesized input
eigenfoo/littlemcmc/littlemcmc/parallel_sampling.py
code served (permissive licence) · get_code("5b85f7845f5fe619")
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 ValueError. 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 traceback

class RemoteTraceback(Exception):
    def __init__(self, tb):
        self.tb = tb

    def __str__(self):
        return self.tb

def rebuild_exc(exc, tb):
    exc.__cause__ = RemoteTraceback(tb)
    return exc

# Create a valid exception object
exc = ValueError("Test error message")

# Create a valid traceback string
tb = "Traceback (most recent call last):\n  File \"test.py\", line 1, in <module>\n    raise ValueError('Test error')\nValueError: Test error"

ARGS = {
    "exc": exc,
    "tb": tb
}
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 ValueError — a ValueError output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a ValueError 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
u_r_step Ran
ran; our contract draft was wrong, not the code
gregversteeg/esh_dynamics/esh/core.py
code served (permissive licence) · get_code("ce2f28911e5c38cf")
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 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 as t

t.manual_seed(0)

# Construct valid arguments for u_r_step
batch_size = 2
spatial_dims = (4,)  # 1D spatial dimension for simplicity
d = 4

# u: tensor of shape (batch_size, *spatial_dims) representing unit vectors
# Initialize as random normalized vectors
u = t.randn(batch_size, *spatial_dims)
u = u / u.flatten(start_dim=1).norm(dim=1, keepdim=True).view(-1, *([1] * len(spatial_dims)))

# rs: tensor of shape (batch_size,) representing log time scales
rs = t.randn(batch_size)

# grad: tensor of same shape as u, representing gradients
grad = t.randn(batch_size, *spatial_dims)

# epsilon: scalar learning rate parameter
epsilon = 0.1

ARGS = {
    "u": u,
    "rs": rs,
    "grad": grad,
    "epsilon": epsilon,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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
init_nuts Not yet run
unverified (harvested, no recorded run)
eigenfoo/littlemcmc/littlemcmc/sampling.py
code served (permissive licence) · get_code("bf4617984094c1f5")
How it ran

no invocation recorded.

quad_potential Not yet run
unverified (harvested, no recorded run)
eigenfoo/littlemcmc/littlemcmc/quadpotential.py
code served (permissive licence) · get_code("91493b1dc16affd9")
How it ran

no invocation recorded.

Also present in other papers' code

Of 14 samples lifted from this paper's linked repositories, 14 carry a function identity and 0 do not. Of the 13 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 0 later) and 12 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
init_random
brick of this paper's assembly: helper
1 other paper
1 dated earlier
1912.03263 (2019-12)
Your Classifier is Secretly an Energy Based Model and You Should Treat it Like One

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 from an unnormalized probability distribution is a fundamental problem in machine learning with applications including Bayesian modeling, latent factor inference, and energy-based model training. After decades of research, variations of MCMC remain the default approach to sampling despite slow convergence. Auxiliary neural models can learn to speed up MCMC, but the overhead for training the extra model can be prohibitive. We propose a fundamentally different approach to this problem via a new Hamiltonian dynamics with a non-Newtonian momentum. In contrast to MCMC approaches like Hamiltonian Monte Carlo, no stochastic step is required. Instead, the proposed deterministic dynamics in an extended state space exactly sample the target distribution, specified by an energy function, under an assumption of ergodicity. Alternatively, the dynamics can be interpreted as a normalizing flow that samples a specified energy model without training. The proposed Energy Sampling Hamiltonian (ESH) dynamics have a simple form that can be solved with existing ODE solvers, but we derive a specialized solver that exhibits much better performance. ESH dynamics converge faster than their MCMC competitors enabling faster, more stable training of neural network energy models.

For agents

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

get_harvested_code_for_paper("2111.02434")
get_code_for_paper("2111.02434")
have("2111.02434")

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

curl https://syntology.ai/api/ran/2111.02434.json

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

[![Syntology run record](https://syntology.ai/api/ran/2111.02434.svg)](https://syntology.ai/paper/2111.02434)

Connect an agent — have() is free.