SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2303.04842 · 2023

Distributed Potential iLQR: Scalable Game-Theoretic Trajectory Planning for Multi-Agent Interactions

arXiv · PDF · Open in the Atlas

Code that ran

9 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; 0 honoured the contract we drafted, 0 violated it and 9 have no contract verdict; 0 failures of our instrument; 5 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/2303.04842.json · badge: /api/ran/2303.04842.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
labicon/dp-ilqr canonical (classifier label)
classifier basis not recorded on the link
9 ran · 5 no recorded run
FunctionStatusWhere it lives
compute_pairwise_distance Ran
ran on a synthesized input
labicon/dp-ilqr/dpilqr/util.py
code served (permissive licence) · get_code("306214f70893433d")
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 (5, 3) 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 computes pairwise distances between agents
# X is reshaped as (-1, n_agents, n_states) then swapped to (n_states, -1, n_agents)
# x_dims is a list where each element represents the state dimension of an agent
# n_d is the number of dimensions to use (default 2)

# Let's create valid arguments:
# - We need at least 2 agents (function raises error for 1 agent)
# - All agents must have the same state dimension (assertion checks len(set(x_dims)) == 1)
# - X must be reshapeable as (-1, n_agents, n_states)

n_agents = 3
n_states = 4
batch_size = 5  # arbitrary batch dimension

# x_dims: list of state dimensions for each agent (all must be equal)
x_dims = [n_states, n_states, n_states]

# X: flat array that can be reshaped to (batch_size, n_agents, n_states)
# Total elements: batch_size * n_agents * n_states
X = np.random.randn(batch_size * n_agents * n_states)

# n_d: number of dimensions to use for distance (must be <= n_states)
n_d = 2

ARGS = {
    "X": X,
    "x_dims": x_dims,
    "n_d": n_d
}
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 2, dtype family float, shape (5, 3) — shape (5, 3) 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
compute_pairwise_distance_nd Ran
ran on a synthesized input
labicon/dp-ilqr/dpilqr/util.py
code served (permissive licence) · get_code("8eda000c67df9307")
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 (2, 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
import itertools

# Set up small test data
# The function expects:
# - X: array of shape (batch_size, n_agents * n_states)
# - x_dims: list where x_dims[0] is n_states
# - n_dims: list of dimensionality for each agent
# - dec_ind: optional agent index for filtering pairs

# Create test data
batch_size = 2
n_agents = 3
n_states = 4  # states per agent
n_dim_agent_0 = 2  # agent 0 uses 2D distance
n_dim_agent_1 = 2  # agent 1 uses 2D distance
n_dim_agent_2 = 3  # agent 2 uses 3D distance

# X: shape (batch_size, n_agents * n_states)
X = np.random.randn(batch_size, n_agents * n_states)

# x_dims: [n_states, n_states, n_states, ...]
x_dims = [n_states, n_states, n_states]

# n_dims: dimensionality for each agent
n_dims = [n_dim_agent_0, n_dim_agent_1, n_dim_agent_2]

# dec_ind: optional, filter pairs to those containing agent 1
dec_ind = 1

ARGS = {
    "X": X,
    "x_dims": x_dims,
    "n_dims": n_dims,
    "dec_ind": dec_ind,
}
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 2, dtype family float, shape (2, 2) — shape (2, 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
forward_euler_integration Ran
ran on a synthesized input
labicon/dp-ilqr/dpilqr/dynamics.py
code served (permissive licence) · get_code("d380f82f423fb693")
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
import torch.nn as nn

torch.manual_seed(0)

# f is a function that takes (x, u) and returns a tensor
# x is a state tensor
# u is a control/input tensor
# h is a step size (scalar)

def f(x, u):
    """Simple dynamics function: linear combination of state and control"""
    # x shape: (2, 4), u shape: (2, 3)
    # returns shape: (2, 4)
    return 0.1 * x + 0.05 * u @ torch.randn(3, 4)

x = torch.randn(2, 4)
u = torch.randn(2, 3)
h = 0.01

ARGS = {
    'f': f,
    'x': x,
    'u': u,
    'h': h,
}
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
nchoosek Ran
ran on a synthesized input
labicon/dp-ilqr/dpilqr/graphics.py
code served (permissive licence) · get_code("565f4ae7c2af75cc")
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 int, ndim 0, dtype family int, 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
from functools import reduce
from operator import mul

torch.manual_seed(0)

# The nchoosek function takes two integer parameters: n and k
# It computes the binomial coefficient n choose k
# We need to provide valid integers where k <= n

ARGS = {
    "n": 5,
    "k": 2,
}
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 int, ndim 0, dtype family int, 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
quadraticize_distance Ran
ran on a synthesized input
labicon/dp-ilqr/dpilqr/cost.py
code served (permissive licence) · get_code("d85d2ab328afec78")
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 numpy as np

class Point:
    def __init__(self, x, y, z=0):
        self.x = x
        self.y = y
        self.z = z
        self.ndim = 1
    
    def hypot2(self):
        return self.x ** 2 + self.y ** 2 + self.z ** 2

# Create two point instances
point_a = Point(x=3.0, y=4.0, z=0.0)
point_b = Point(x=0.0, y=0.0, z=0.0)

# Parameters for the function
radius = 3.0
n_d = 2

ARGS = {
    "point_a": point_a,
    "point_b": point_b,
    "radius": radius,
    "n_d": n_d
}
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 — 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
quadraticize_finite_difference Ran
ran on a synthesized input
labicon/dp-ilqr/dpilqr/cost.py
code served (permissive licence) · get_code("ab22271cb30f1f84")
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 numpy as np

def dummy_cost(x, u, terminal):
    """Simple quadratic cost function for testing"""
    return float(np.sum(x**2) + np.sum(u**2))

# Create small test inputs
x = np.array([1.0, 2.0, 3.0, 4.0])  # state dimension 4
u = np.array([0.5, 1.5, 2.5])  # control dimension 3

ARGS = {
    "cost": dummy_cost,
    "x": x,
    "u": u,
    "terminal": False,
    "jac_eps": None,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — 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
rk4_integration Ran
ran on a synthesized input
labicon/dp-ilqr/dpilqr/dynamics.py
code served (permissive licence) · get_code("b8607f54318df6e0")
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 torch
import numpy as np

torch.manual_seed(0)

# Define a simple dynamics function f(x, u) that returns dx/dt
# The function is called with state x and control u
# It should return a derivative vector of the same shape as x
def dynamics_fn(x, u):
    """Simple linear dynamics: dx/dt = -x + u"""
    return -x + u

# Initial state x0 - needs .copy() method, so use numpy array
x0 = np.array([1.0, 2.0, 3.0, 4.0])

# Control input u - passed to dynamics function but not modified
u = np.array([0.1, 0.2, 0.3, 0.4])

# Integration horizon h
h = 1.0

# Sub-integration step size dh (optional)
dh = 0.1

ARGS = {
    "f": dynamics_fn,
    "x0": x0,
    "u": u,
    "h": h,
    "dh": dh
}
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 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
scipy_integration Ran
ran on a synthesized input
labicon/dp-ilqr/dpilqr/dynamics.py
code served (permissive licence) · get_code("37b64fe3fa4467f7")
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 torch
import numpy as np
from scipy.integrate import solve_ivp

torch.manual_seed(0)

# Define a simple dynamics function f(x, u) that returns dx/dt
# This function is called as f(x, u) where x is state and u is control input
def f(x, u):
    """
    Simple linear dynamics: dx/dt = A*x + B*u
    x: state vector (shape: (4,))
    u: control input (shape: (3,))
    returns: dx/dt (shape: (4,))
    """
    A = np.array([
        [-0.1, 0.2, 0.0, 0.1],
        [0.1, -0.15, 0.05, 0.0],
        [0.0, 0.1, -0.2, 0.1],
        [0.05, 0.0, 0.1, -0.1]
    ])
    B = np.array([
        [0.1, 0.0, 0.05],
        [0.0, 0.1, 0.0],
        [0.05, 0.0, 0.1],
        [0.0, 0.05, 0.0]
    ])
    x_flat = np.asarray(x).flatten()
    u_flat = np.asarray(u).flatten()
    dx_dt = A @ x_flat + B @ u_flat
    return dx_dt

# Initial state: 4-dimensional
x = np.array([1.0, 0.5, -0.3, 0.2])

# Control input: 3-dimensional (distinct from state dimension)
u = np.array([0.1, -0.05, 0.15])

# Time horizon for integration
h = 0.1

# Additional kwargs for solve_ivp (optional)
kwargs = {
    'method': 'RK45',
    'dense_output': False
}

ARGS = {
    'f': f,
    'x': x,
    'u': u,
    'h': h,
    **kwargs
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy, 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 (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
split_agents Ran
ran on a synthesized input
labicon/dp-ilqr/dpilqr/util.py
code served (permissive licence) · get_code("0c86a3c8208ecab0")
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 numpy as np

# Z is a 2D array-like state/control vector
# For a batch of 2 samples with 7 total dimensions
Z = np.array([[1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0],
              [8.0, 9.0, 10.0, 11.0, 12.0, 13.0, 14.0]])

# z_dims specifies how to partition the columns
# Example: 3 agents with dimensions [2, 3, 2] summing to 7
z_dims = [2, 3, 2]

ARGS = {
    "Z": Z,
    "z_dims": z_dims
}
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 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
define_inter_graph_threshold Not yet run
unverified (harvested, no recorded run)
labicon/dp-ilqr/dpilqr/distributed.py
code served (permissive licence) · get_code("8a4081f81bb6722e")
How it ran

no invocation recorded.

solve_distributed Not yet run
unverified (harvested, no recorded run)
labicon/dp-ilqr/dpilqr/distributed.py
code served (permissive licence) · get_code("aee96831e4daa63d")
How it ran

no invocation recorded.

solve_rhc Not yet run
unverified (harvested, no recorded run)
labicon/dp-ilqr/dpilqr/distributed.py
code served (permissive licence) · get_code("071f43ca77a0ebc6")
How it ran

no invocation recorded.

solve_subproblem Not yet run
unverified (harvested, no recorded run)
labicon/dp-ilqr/dpilqr/problem.py
code served (permissive licence) · get_code("946a58b2ed451d55")
How it ran

no invocation recorded.

solve_subproblem_starmap Not yet run
unverified (harvested, no recorded run)
labicon/dp-ilqr/dpilqr/problem.py
code served (permissive licence) · get_code("8f4c5f787e70f798")
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 14 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 14 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

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

Abstract

In this work, we develop a scalable, local trajectory optimization algorithm that enables robots to interact with other robots. It has been shown that agents' interactions can be successfully captured in game-theoretic formulations, where the interaction outcome can be best modeled via the equilibria of the underlying dynamic game. However, it is typically challenging to compute equilibria of dynamic games as it involves simultaneously solving a set of coupled optimal control problems. Existing solvers operate in a centralized fashion and do not scale up tractably to multiple interacting agents. We enable scalable distributed game-theoretic planning by leveraging the structure inherent in multi-agent interactions, namely, interactions belonging to the class of dynamic potential games. Since equilibria of dynamic potential games can be found by minimizing a single potential function, we can apply distributed and decentralized control techniques to seek equilibria of multi-agent interactions in a scalable and distributed manner. We compare the performance of our algorithm with a centralized interactive planner in a number of simulation studies and demonstrate that our algorithm results in better efficiency and scalability. We further evaluate our method in hardware experiments involving multiple quadcopters.

For agents

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

get_harvested_code_for_paper("2303.04842")
get_code_for_paper("2303.04842")
have("2303.04842")

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

curl https://syntology.ai/api/ran/2303.04842.json

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

[![Syntology run record](https://syntology.ai/api/ran/2303.04842.svg)](https://syntology.ai/paper/2303.04842)

Connect an agent — have() is free.