SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2009.14148 · NeurIPS · 2020

Unbalanced Sobolev Descent

Youssef Mroueh, Mattia Rigotti

arXiv · PDF · Open in the Atlas

Code that ran

11 of 15 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (2 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 8 have no contract verdict; 3 failures of our instrument; 4 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/2009.14148.json · badge: /api/ran/2009.14148.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
ibm/usd canonical (classifier label)
classifier basis: census
6 ran · 2 no recorded run
IBM/USD community
Papers with Code archive
5 ran · 2 no recorded run
FunctionStatusWhere it lives
D_forward_weights Ran
ran; our contract draft was wrong, not the code
IBM/USD/unbalanced_descent.py
code served (permissive licence) · get_code("666b93c844016302")
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. 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)

# Define a simple discriminator model
class SimpleDiscriminator(nn.Module):
    def __init__(self, input_dim, output_dim=1):
        super().__init__()
        self.linear = nn.Linear(input_dim, output_dim)
    
    def forward(self, x):
        return self.linear(x)

# Create the discriminator
D = SimpleDiscriminator(input_dim=4, output_dim=1)

# Create input tensors
# x_p: batch of positive samples, shape (2, 4)
x_p = torch.randn(2, 4, requires_grad=False)

# x_q: batch of query samples, shape (2, 4)
x_q = torch.randn(2, 4, requires_grad=False)

# Weights for positive samples
w_p = torch.ones(2, 1)

# Weights for query samples
w_q = torch.ones(2, 1)

# Hyperparameters
lambda_aug = 1.0
alpha = 0.5
rho = 2.0

ARGS = {
    'D': D,
    'x_p': x_p,
    'w_p': w_p,
    'x_q': x_q,
    'w_q': w_q,
    'lambda_aug': lambda_aug,
    'alpha': alpha,
    'rho': rho,
}
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 — 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
MMD_RFF Ran
ran on a synthesized input · constructed an object; no result was computed
IBM/USD/unbalanced_descent.py
code served (permissive licence) · get_code("a74a618e9386ec3d")
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, MMD_RFF: 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 numpy as np
from torch import nn

torch.manual_seed(0)

# Define the MMD_RFF module
class RFFEmbedding(nn.Module):
    def __init__(self, num_features, num_outputs=100, sigma=1.0):
        super(RFFEmbedding, self).__init__()
        self.num_features = num_features
        self.num_outputs = num_outputs
        self.sigma = sigma
        self.weight = nn.Parameter(torch.Tensor(num_features, num_outputs).normal_().mul_(np.sqrt(2) / sigma))
        self.bias = nn.Parameter(torch.Tensor(num_outputs).uniform_(-np.pi, np.pi))

    def forward(self, inputs):
        h = inputs @ self.weight + self.bias
        return torch.cos(h).mul(np.sqrt(2 / self.num_outputs))


class MMD_RFF(nn.Module):
    def __init__(self, num_features, num_outputs=100, sigma=1.0):
        super(MMD_RFF, self).__init__()
        self.num_features = num_features
        self.num_outputs = num_outputs
        self.rff_emb = RFFEmbedding(num_features, num_outputs, sigma=sigma)

    def forward(self, X, Y, weights_X=None, weights_Y=None):
        fX, fY = (self.rff_emb(X), self.rff_emb(Y))
        if weights_X is None:
            mu_X = fX.mean(0)
        else:
            mu_X = (weights_X.view(-1, 1) / weights_X.sum() * fX).sum(0)
        if weights_Y is None:
            mu_Y = fY.mean(0)
        else:
            mu_Y = (weights_Y.view(-1, 1) / weights_Y.sum() * fY).sum(0)
        d_XY = mu_X - mu_Y
        return d_XY.norm()


# Create sample data
X = torch.randn(2, 4)  # batch1=2, num_features=4
Y = torch.randn(3, 4)  # batch2=3, num_features=4
weights_X = torch.ones(2)
weights_Y = torch.ones(3)

ARGS = {
    "num_features": 4,
    "num_outputs": 50,
    "sigma": 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 MMD_RFF — 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 MMD_RFF 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
RFFEmbedding Ran
ran on a synthesized input · constructed an object; no result was computed
IBM/USD/unbalanced_descent.py
code served (permissive licence) · get_code("7b1e4a57e169fde5")
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, RFFEmbedding: 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 numpy as np
from torch import nn

torch.manual_seed(0)

# Create a small batch of input data
# batch size: 2, num_features: 4
batch_size = 2
num_features = 4
inputs = torch.randn(batch_size, num_features)

# Define parameters for RFFEmbedding
num_outputs = 8
sigma = 1.0

ARGS = {
    'num_features': num_features,
    'num_outputs': num_outputs,
    'sigma': sigma,
}
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 RFFEmbedding — 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 RFFEmbedding 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
describe_data Ran
ran on a synthesized input
ibm/usd/datasets.py
code served (permissive licence) · get_code("535e52d7bda9b841")
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 torch
import numpy as np

torch.manual_seed(0)

# Create a small torch.Tensor
tensor_data = torch.randn(2, 4)

# Create a small numpy.ndarray
numpy_data = np.random.randn(3, 5)

# Test with tensor
ARGS = {
    'D': tensor_data
}
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 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_devices Ran
ran on a synthesized input
ibm/usd/utils.py
code served (permissive licence) · get_code("290087b87df2ac2e")
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 1, dtype family float; no output shape is stated. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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

torch.manual_seed(0)

ARGS = {
    'cuda_device': 'cpu',
    'seed': 1,
}
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 1, 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_loader Ran
ran; our contract draft was wrong, not the code
IBM/USD/unbalanced_descent.py
code served (permissive licence) · get_code("1beaa0f96fbe0db6")
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 DataLoader. 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

torch.manual_seed(0)

# Create small sample tensors
# Input features: batch_size=2, features=4
x = torch.randn(2, 4)

# Target labels: batch_size=2
y = torch.randn(2, 3)

# Define arguments for get_loader
ARGS = {
    'data': (x, y),
    'batch_size': 2,
    'shuffle': False
}
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 DataLoader — a DataLoader output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a DataLoader 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
load_img Ran
ran on a synthesized input
ibm/usd/datasets.py
code served (permissive licence) · get_code("5912b3a04cff8c04")
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 2, dtype family float, shape (50, 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 numpy as np
from PIL import Image
import io

# Create a small test image in memory
def create_test_image(size=200):
    """Create a simple test image with some black pixels"""
    img_array = np.ones((size, size), dtype=np.uint8) * 255  # white background
    # Add some black pixels to make nonzero work
    img_array[50:100, 50:100] = 0  # black square
    img = Image.fromarray(img_array, mode='L')
    return img

# Create image and save to bytes buffer
test_img = create_test_image(size=200)
img_buffer = io.BytesIO()
test_img.save(img_buffer, format='JPEG')
img_buffer.seek(0)

# Create a temporary file path that we'll mock
# Since we can't do file I/O, we need to patch Image.open
# Instead, let's create a minimal mock setup

import tempfile
import os

# Create temporary directory and save test image
temp_dir = tempfile.mkdtemp()
test_img_path = os.path.join(temp_dir, 'test_img.jpg')
test_img.save(test_img_path)

ARGS = {
    'fn': test_img_path,
    'size': 200,
    'max_samples': 50
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family float, shape (50, 2) — shape (50, 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
load_weights Ran
ran on a synthesized input
ibm/usd/datasets.py
code served (permissive licence) · get_code("82adb6728cac0e04")
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 (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 torch
import numpy as np
from PIL import Image
import io

torch.manual_seed(0)

# Create a small dummy image and save it to a bytes buffer
# We'll create a simple grayscale image
img_array = np.random.randint(0, 256, (200, 200), dtype=np.uint8)
img = Image.fromarray(img_array, mode='L')
img_buffer = io.BytesIO()
img.save(img_buffer, format='JPEG')
img_buffer.seek(0)

# Create a temporary in-memory image file path workaround
# Since we can't do file I/O, we'll mock the image loading
import tempfile
import os

# Create a temporary directory and file
temp_dir = tempfile.gettempdir()
temp_img_path = os.path.join(temp_dir, 'test_spiral.jpg')
img.save(temp_img_path)

# Create position tensor: batch size 2, 2D coordinates (x, y)
# Values should be in range [-1, 1] based on how the function normalizes them
pos_tensor = torch.tensor([
    [-0.5, 0.3],
    [0.2, -0.8]
], dtype=torch.float32)

ARGS = {
    'pos': pos_tensor,
    'img': temp_img_path,
    'size': 200
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
PIL, numpy, torch — 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 (2,) — shape (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
minibatch Ran
ran; our contract draft was wrong, not the code
IBM/USD/unbalanced_descent.py
code served (permissive licence) · get_code("194fa2394685944b")
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 list. 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

torch.manual_seed(0)

# Create small test tensors
data_tensor_1 = torch.randn(2, 4)  # batch size 2, features 4
data_tensor_2 = torch.randn(2, 3)  # batch size 2, features 3

ARGS = {
    "data": (data_tensor_1, data_tensor_2),
    "batch_size": None,
}
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 list — a list 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
sparsemax Ran
ran on a synthesized input
ibm/usd/modules.py
code served (permissive licence) · get_code("8ae5e6dfcb32482c")
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 (2, 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)

# Create a 2D tensor with batch size 2 and 4 features
logits = torch.randn(2, 4)

ARGS = {
    'logits': logits
}
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 (2, 4) — shape (2, 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
swish Ran
ran on a synthesized input
ibm/usd/modules.py
code served (permissive licence) · get_code("fc3718baf8f65341")
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 (2, 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)

# The function swish takes inputs and applies the swish activation function
# It expects a tensor as input
inputs = torch.randn(2, 4)

ARGS = {
    "inputs": inputs
}
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 (2, 4) — shape (2, 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
log_to_dict Not yet run
unverified (harvested, no recorded run)
ibm/usd/utils.py
code served (permissive licence) · get_code("88a867cdff5a2e0b")
How it ran

no invocation recorded.

make_data_parallel Not yet run
unverified (harvested, no recorded run)
ibm/usd/utils.py
code served (permissive licence) · get_code("0b8190dcac4fd585")
How it ran

no invocation recorded.

manual_sgd_ Not yet run
unverified (harvested, no recorded run)
IBM/USD/unbalanced_descent.py
code served (permissive licence) · get_code("13c848a877c3681a")
How it ran

no invocation recorded.

train_unbalanced_descent Not yet run
unverified (harvested, no recorded run)
IBM/USD/unbalanced_descent.py
code served (permissive licence) · get_code("debaae28c8c3b426")
How it ran

no invocation recorded.

Also present in other papers' code

Of 15 samples lifted from this paper's linked repositories, 15 carry a function identity and 0 do not. Of the 15 distinct functions with an identity, 2 are also present in other papers whose code we hold (2 with at least one paper dated earlier, 0 the same month, 0 later) and 13 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
get_devices 1 other paper
1 dated earlier
1606.07892 (2016-06)
Large-Scale Kernel Methods for Independence Testing
make_data_parallel 1 other paper
1 dated earlier
1606.07892 (2016-06)
Large-Scale Kernel Methods for Independence Testing

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

We introduce Unbalanced Sobolev Descent (USD), a particle descent algorithm for transporting a high dimensional source distribution to a target distribution that does not necessarily have the same mass. We define the Sobolev-Fisher discrepancy between distributions and show that it relates to advection-reaction transport equations and the Wasserstein-Fisher-Rao metric between distributions. USD transports particles along gradient flows of the witness function of the Sobolev-Fisher discrepancy (advection step) and reweighs the mass of particles with respect to this witness function (reaction step). The reaction step can be thought of as a birth-death process of the particles with rate of growth proportional to the witness function. When the Sobolev-Fisher witness function is estimated in a Reproducing Kernel Hilbert Space (RKHS), under mild assumptions we show that USD converges asymptotically (in the limit of infinite particles) to the target distribution in the Maximum Mean Discrepancy (MMD) sense. We then give two methods to estimate the Sobolev-Fisher witness with neural networks, resulting in two Neural USD algorithms. The first one implements the reaction step with mirror descent on the weights, while the second implements it through a birthdeath process of particles. We show on synthetic examples that USD transports distributions with or without conservation of mass faster than previous particle descent algorithms, and finally demonstrate its use for molecular biology analyses where our method is naturally suited to match developmental stages of populations of differentiating cells based on their single-cell RNA sequencing profile. Code is available at http://github.com/ibm/usd.

For agents

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

get_harvested_code_for_paper("2009.14148")
get_code_for_paper("2009.14148")
have("2009.14148")

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

curl https://syntology.ai/api/ran/2009.14148.json

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

[![Syntology run record](https://syntology.ai/api/ran/2009.14148.svg)](https://syntology.ai/paper/2009.14148)

Connect an agent — have() is free.