SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2007.03155 · 2020

Decentralized policy learning with partial observation and mechanical constraints for multiperson modeling

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; 1 honoured the contract we drafted, 0 violated it and 11 have no contract verdict; 0 failures of our instrument; 2 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/2007.03155.json · badge: /api/ran/2007.03155.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
keisuke198619/PO-MC-DHVRNN canonical (classifier label)
classifier basis: census
12 ran · 2 no recorded run
FunctionStatusWhere it lives
check_game_roles_duplicates Ran
ran on a synthesized input
keisuke198619/PO-MC-DHVRNN/utilities.py
code served (permissive licence) · get_code("5a1f9577a7a8d923")
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

ARGS = {
    "id_role_mapping": {
        "player_1": ["role_a"],
        "player_2": ["role_b", "role_c"],
        "player_3": ["role_d", "role_e", "role_f"],
        "player_4": ["role_g"]
    }
}
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 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
create_static_features Ran
ran on a synthesized input
keisuke198619/PO-MC-DHVRNN/features.py
code served (permissive licence) · get_code("7b1232890f213204")
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 numpy as np
import pandas as pd

# Create a small stand-in DataFrame with moments
# The function expects events_df to have a 'moments' column
# Each moment is a flattened array of shape (n_pl * 4 + 2 + 2,) for a single moment
# or a list/array of such moments

def create_sample_moments(n_pl):
    """Create sample moments for testing"""
    # Each moment has:
    # - n_pl * 4 values for player positions (n_pl players * 2 coordinates)
    # - 2 values for ball position
    # - 2 values for hoop position (but these are computed, not in input)
    # Total input: n_pl * 4 + 2
    moment_size = n_pl * 4 + 2
    
    # Create 3 sample moments
    moments_list = [
        np.random.randn(moment_size).astype(np.float32),
        np.random.randn(moment_size).astype(np.float32),
        np.random.randn(moment_size).astype(np.float32),
    ]
    return moments_list

# Create events_df with n_pl=5 (basketball)
n_pl_val = 5
events_data = {
    'moments': [
        create_sample_moments(n_pl_val),
        create_sample_moments(n_pl_val),
    ]
}
events_df = pd.DataFrame(events_data)

ARGS = {
    'events_df': events_df,
    'n_pl': n_pl_val,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — 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
flatten_moments Ran
ran on a synthesized input
keisuke198619/PO-MC-DHVRNN/features.py
code served (permissive licence) · get_code("5614066a824dbc29")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 2, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np
import pandas as pd

# Create a minimal events_df with the structure expected by flatten_moments
# The function expects:
# - A 'moments' column containing lists of moments
# - Each moment is a tuple/list where:
#   - moment[0] is a numeric value (game_clock or similar)
#   - moment[2] is a numeric value
#   - moment[3] is a numeric value
#   - moment[5] is a numpy array with shape (11, 5) representing 10 players + 1 ball
#     - rows 1:6 are team 1 players (5 players)
#     - rows 6:11 are team 2 players (5 players)
#     - rows 0 is the ball
#     - columns are [team_id, player_id, x, y, z] where x, y are in feet

def create_moment(game_clock, val2, val3, team1_id=1, team2_id=2):
    """Create a valid moment structure"""
    # Create the player/ball array: 11 rows x 5 columns
    m = np.zeros((11, 5))
    
    # Ball (row 0): [team_id, player_id, x, y, z]
    m[0] = np.array([0, -1, 47.0, 25.0, 0.0])
    
    # Team 1 players (rows 1-5)
    for i in range(1, 6):
        m[i] = np.array([team1_id, i, 20.0 + i, 10.0 + i, 0.0])
    
    # Team 2 players (rows 6-11)
    for i in range(6, 11):
        m[i] = np.array([team2_id, i, 60.0 + i, 30.0 + i, 0.0])
    
    # moment is [game_clock, ?, val2, val3, ?, moments_array]
    moment = [game_clock, None, val2, val3, None, m]
    return moment

# Create sample data
moments_list_1 = [
    create_moment(600.0, 100.0, 200.0, team1_id=1, team2_id=2),
    create_moment(599.5, 101.0, 201.0, team1_id=1, team2_id=2),
]

moments_list_2 = [
    create_moment(598.0, 102.0, 202.0, team1_id=1, team2_id=2),
    create_moment(597.5, 103.0, 203.0, team1_id=1, team2_id=2),
]

events_df = pd.DataFrame({
    'moments': [moments_list_1, moments_list_2]
})

ARGS = {
    'events_df': events_df,
    'normalize': True,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 2, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_params_str Ran
ran on a synthesized input
keisuke198619/PO-MC-DHVRNN/vrnn/models/utils.py
code served (permissive licence) · get_code("a2044933b17736c4")
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

torch.manual_seed(0)

def get_params_str(model_args, params):
    ret = ''
    for arg in model_args:
        ret += ' {} {} |'.format(arg, params[arg])
    return ret[1:-2]

model_args = ['learning_rate', 'batch_size', 'hidden_dim']
params = {
    'learning_rate': 0.001,
    'batch_size': 32,
    'hidden_dim': 64
}

ARGS = {
    'model_args': model_args,
    'params': params
}
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 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_sequences Ran
ran on a synthesized input
keisuke198619/PO-MC-DHVRNN/sequencing.py
code served (permissive licence) · get_code("355994bf463714bb")
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

# Create a minimal single_game structure
# Based on the code, it accesses indices up to around 551 for basketball
# We need shape[1] to be at least 552 for n_pl=5 case
single_game = [np.random.randn(10, 600)]  # 10 timesteps, 600 features

# policy is iterated over, appears to be a list/array of indices
policy = [0, 1]

# sequence_length: used for creating sequences
sequence_length = 5

# overlap: used for sequence generation
overlap = 2

# n_pl: number of players (5 for basketball, 11 for soccer)
n_pl = 5

# k_nearest: number of nearest players
k_nearest = 3

# n_feat: number of features per nearest player
n_feat = 4

# velocity: flag for velocity (0, 1, or 2)
velocity = 1

# in_sma: boolean flag
in_sma = False

ARGS = {
    'single_game': single_game,
    'policy': policy,
    'sequence_length': sequence_length,
    'overlap': overlap,
    'n_pl': n_pl,
    'k_nearest': k_nearest,
    'n_feat': n_feat,
    'velocity': velocity,
    'in_sma': in_sma
}
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
id_player Ran
ran on a synthesized input
keisuke198619/PO-MC-DHVRNN/utilities.py
code served (permissive licence) · get_code("df8527ed08f0823e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import pandas as pd

# Create a small dataframe with the structure expected by id_player
# The function expects each row to have 'home' and 'visitor' columns
# containing dicts with 'players' lists

home_team_1 = {
    'players': [
        {'playerid': 1, 'firstname': 'John', 'lastname': 'Doe'},
        {'playerid': 2, 'firstname': 'Jane', 'lastname': 'Smith'},
    ]
}

away_team_1 = {
    'players': [
        {'playerid': 3, 'firstname': 'Bob', 'lastname': 'Johnson'},
        {'playerid': 4, 'firstname': 'Alice', 'lastname': 'Williams'},
    ]
}

home_team_2 = {
    'players': [
        {'playerid': 1, 'firstname': 'John', 'lastname': 'Doe'},
        {'playerid': 5, 'firstname': 'Charlie', 'lastname': 'Brown'},
    ]
}

away_team_2 = {
    'players': [
        {'playerid': 3, 'firstname': 'Bob', 'lastname': 'Johnson'},
        {'playerid': 6, 'firstname': 'Diana', 'lastname': 'Miller'},
    ]
}

event_df = pd.DataFrame({
    'home': [home_team_1, home_team_2],
    'visitor': [away_team_1, away_team_2]
})

ARGS = {
    'event_df': event_df
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
id_position Ran
ran on a synthesized input
keisuke198619/PO-MC-DHVRNN/utilities.py
code served (permissive licence) · get_code("eb7beff2f5f867dc")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import pandas as pd
import numpy as np

# Create a minimal DataFrame with the structure expected by id_position
# The function expects event_df to have rows where each row contains 'home' and 'visitor' data
# Each of those contains a 'players' list with dicts having 'playerid' and 'position' keys

event_data = []

# First event with home and away players
event_1 = {
    'home': {
        'players': [
            {'playerid': 1, 'position': 'goalkeeper'},
            {'playerid': 2, 'position': 'defender'},
            {'playerid': 3, 'position': 'midfielder'},
        ]
    },
    'visitor': {
        'players': [
            {'playerid': 10, 'position': 'goalkeeper'},
            {'playerid': 11, 'position': 'defender'},
            {'playerid': 12, 'position': 'forward'},
        ]
    }
}

# Second event with some overlapping and some new players
event_2 = {
    'home': {
        'players': [
            {'playerid': 1, 'position': 'goalkeeper'},
            {'playerid': 2, 'position': 'defender'},
            {'playerid': 4, 'position': 'forward'},
        ]
    },
    'visitor': {
        'players': [
            {'playerid': 10, 'position': 'goalkeeper'},
            {'playerid': 11, 'position': 'defender'},
            {'playerid': 13, 'position': 'midfielder'},
        ]
    }
}

event_data.append(event_1)
event_data.append(event_2)

event_df = pd.DataFrame({
    'home': [event_data[0]['home'], event_data[1]['home']],
    'visitor': [event_data[0]['visitor'], event_data[1]['visitor']]
})

ARGS = {
    'event_df': event_df
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
num_trainable_params Ran
ran, honoured the contract we drafted
keisuke198619/PO-MC-DHVRNN/vrnn/models/utils.py
pointer only (licence: MIT); the stored copy get_code serves, from felixykliu/NAOMI, is pointer only (licence: NONE) · get_code("dcc2aedb4ca9a4e8")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from felixykliu/NAOMI, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

parse_model_params Ran
ran on a synthesized input
keisuke198619/PO-MC-DHVRNN/vrnn/models/utils.py
code served (permissive licence) · get_code("1b07b23dc05534d2")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import argparse
import torch
import torch.nn as nn

torch.manual_seed(0)

# Create a minimal argument parser
parser = argparse.ArgumentParser()

# Define model arguments - these will be added as command-line arguments
model_args = ['hidden_size', 'num_layers', 'dropout']

# Create a params dict to be updated
params = {
    'learning_rate': 0.001,
    'batch_size': 32
}

# Simulate command-line arguments by setting sys.argv
import sys
sys.argv = ['script.py', '--hidden_size', '128', '--num_layers', '3', '--dropout', '1']

ARGS = {
    'model_args': model_args,
    'params': params,
    'parser': parser
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
remove_non_eleven Ran
ran on a synthesized input
keisuke198619/PO-MC-DHVRNN/preprocessing.py
code served (permissive licence) · get_code("6c4fda6e44757193")
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 pandas as pd
import numpy as np

# Construct a minimal events_df with the required structure
# The function expects a DataFrame with 'moments' column and home/visitor info

# Create sample moments data
# For n_pl=5 case: moments[i][5] should contain player data (11 elements when valid)
# For n_pl=10 case: moments[i][0] should contain player data (46 elements when valid)

def create_sample_moments(n_moments, n_pl, valid=True):
    """Create sample moments with the correct structure."""
    moments = []
    for i in range(n_moments):
        if n_pl == 5:
            # Structure: [time, quarter, ..., players_list, ...]
            # moments[i][5] should be the players list
            if valid:
                players = [{'x': j, 'y': j+1} for j in range(11)]
            else:
                players = [{'x': j, 'y': j+1} for j in range(5)]  # Invalid: fewer than 11
            moment = [0, 1, 2, 3, 4, players]
        else:
            # For n_pl=10: moments[i][0] should be the players list
            if valid:
                players = [{'x': j, 'y': j+1} for j in range(46)]
            else:
                players = [{'x': j, 'y': j+1} for j in range(20)]  # Invalid: fewer than 46
            moment = [players]
        moments.append(moment)
    return moments

# Create events dataframe with valid and invalid moments
valid_moments = create_sample_moments(12, n_pl=5, valid=True)
invalid_moments = create_sample_moments(2, n_pl=5, valid=False)
mixed_moments = valid_moments[:5] + invalid_moments + valid_moments[5:]

events_df = pd.DataFrame({
    'moments': [mixed_moments, valid_moments],
    'home': [
        {'teamid': 'home_team_1'},
        {'teamid': 'home_team_2'}
    ],
    'visitor': [
        {'teamid': 'away_team_1'},
        {'teamid': 'away_team_2'}
    ]
})

ARGS = {
    'events_df': events_df,
    'event_length_th': 3,
    'n_pl': 5,
    'verbose': False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — 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
remove_outlier Ran
ran on a synthesized input
keisuke198619/PO-MC-DHVRNN/preprocessing.py
code served (permissive licence) · get_code("fbdba52436f13206")
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
import pandas as pd

# Create sample data for the function
np.random.seed(0)

# Create a DataFrame with 'moments' column
# Each moment is an array with shape (num_frames, 528)
# Indices 506:526 are player velocities (10 players * 2 coords = 20 values)
# Indices 526:528 are ball velocities (2 coords)

n_players = 10
n_frames = 5

# Create moments array with shape (n_frames, 528)
# First 506 columns are dummy data
# Columns 506:526 are player velocities (10 players)
# Columns 526:528 are ball velocities
moments_1 = np.random.randn(n_frames, 528) * 0.1
# Set velocities to valid range (below thresholds: ball < 20, players < 13)
moments_1[:, 506:526] = np.random.randn(n_frames, 20) * 5  # player velocities < 13
moments_1[:, 526:528] = np.random.randn(n_frames, 2) * 8   # ball velocities < 20

moments_2 = np.random.randn(n_frames, 528) * 0.1
moments_2[:, 506:526] = np.random.randn(n_frames, 20) * 5
moments_2[:, 526:528] = np.random.randn(n_frames, 2) * 8

# Create DataFrame with moments column
events_df = pd.DataFrame({
    'moments': [moments_1, moments_2],
    'other_col': [1, 2]
})

ARGS = {
    'events_df': events_df,
    'n_pl': n_players,
    'verbose': False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — 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
subsample_sequence Ran
ran on a synthesized input
keisuke198619/PO-MC-DHVRNN/sequencing.py
code served (permissive licence) · get_code("621c945f84c310f5")
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
from scipy import signal

# Create sample event sequences
# The function expects a list of arrays (events), where each array has shape (n_moments, n_features)
events = [
    np.random.randn(10, 4),  # First sequence: 10 time steps, 4 features
    np.random.randn(12, 4),  # Second sequence: 12 time steps, 4 features
]

# Define arguments
ARGS = {
    "events": events,
    "subsample_factor": 2,
    "random_sample": False,
}
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 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
filters Not yet run
unverified (harvested, no recorded run)
keisuke198619/PO-MC-DHVRNN/preprocessing.py
code served (permissive licence) · get_code("3387d8452713966d")
How it ran

no invocation recorded.

flatten_moments_soccer Not yet run
unverified (harvested, no recorded run)
keisuke198619/PO-MC-DHVRNN/features.py
code served (permissive licence) · get_code("cd67ed55902cba6d")
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, 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 11 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
num_trainable_params 2 other papers
2 dated earlier
1803.07612 (2018-03)
Generating Multi-Agent Trajectories using Programmatic Weak Supervision
1901.10946 (2019-01)
NAOMI: Non-Autoregressive Multiresolution Sequence Imputation
get_params_str 1 other paper
1 dated earlier
1803.07612 (2018-03)
Generating Multi-Agent Trajectories using Programmatic Weak Supervision
parse_model_params 1 other paper
1 dated earlier
1803.07612 (2018-03)
Generating Multi-Agent Trajectories using Programmatic Weak Supervision

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

Extracting the rules of real-world multi-agent behaviors is a current challenge in various scientific and engineering fields. Biological agents independently have limited observation and mechanical constraints; however, most of the conventional data-driven models ignore such assumptions, resulting in lack of biological plausibility and model interpretability for behavioral analyses. Here we propose sequential generative models with partial observation and mechanical constraints in a decentralized manner, which can model agents' cognition and body dynamics, and predict biologically plausible behaviors. We formulate this as a decentralized multi-agent imitation-learning problem, leveraging binary partial observation and decentralized policy models based on hierarchical variational recurrent neural networks with physical and biomechanical penalties. Using real-world basketball and soccer datasets, we show the effectiveness of our method in terms of the constraint violations, long-term trajectory prediction, and partial observation. Our approach can be used as a multi-agent simulator to generate realistic trajectories using real-world data.

For agents

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

get_harvested_code_for_paper("2007.03155")
get_code_for_paper("2007.03155")
have("2007.03155")

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

curl https://syntology.ai/api/ran/2007.03155.json

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

[![Syntology run record](https://syntology.ai/api/ran/2007.03155.svg)](https://syntology.ai/paper/2007.03155)

Connect an agent — have() is free.