Skip to main content

Implementation of World Particle

Project description

WorldParticle (wip)

Implementation of WorldParticle, Unified Simulation of Lagrangian Particle Dynamics via Transformer

Install

$ pip install worldparticle

Usage

import torch
from worldparticle import WorldParticle, ParticleTokenizer

tokenizer = ParticleTokenizer(
    dim = 64,
    dim_attr = 1,
    grid_res = 5,
    spatial_radius = 2.0,
    boundary_radius = 2.0
)

model = WorldParticle(
    predictor = dict(
        delta_time = 0.01
    ),
    corrector = dict(
        dim = 64,
        enc_depth = 4,
        dec_depth = 4,
        enc_dim_head = 96,
        enc_heads = 8,
        dec_dim_head = 96,
        dec_heads = 8
    ),
    tokenizer = tokenizer
)

pos = torch.randn(2, 512, 3)
vel = torch.randn(2, 512, 3)
forces = torch.randn(2, 512, 3)
mass = torch.ones(2, 512)

boundary_pos = torch.randn(2, 128, 3)

# forward 1 step

out = model(
    pos = pos,
    vel = vel,
    mass = mass,
    forces = forces,
    tokenizer_kwargs = dict(
        attrs = mass,
        boundary_pos = boundary_pos
    )
)

# out.pos (2, 512, 3)
# out.vel (2, 512, 3)

# or multi-step rollout

out_trajectory = model(
    num_steps = 10,
    pos = pos,
    vel = vel,
    mass = mass,
    forces = forces,
    return_initial_state = True,
    tokenizer_kwargs = dict(
        attrs = mass,
        boundary_pos = boundary_pos
    )
)

# out_trajectory.pos (2, 11, 512, 3)
# out_trajectory.vel (2, 11, 512, 3)

Citations

@misc{wang2026unifiedsimulationlagrangianparticle,
    title   = {Unified Simulation of Lagrangian Particle Dynamics via Transformer},
    author  = {Caoliwen Wang and Minghao Guo and Siyuan Chen and Heng Zhang and Mengdi Wang and Xingyu Ni and Hanson Sun and Kunyi Wang and Zherong Pan and Kui Wu and Lingjie Liu and Yin Yang and Chenfanfu Jiang and Taku Komura and Wojciech Matusik and Peter Yichen Chen},
    year    = {2026},
    eprint  = {2605.15305},
    archivePrefix = {arXiv},
    primaryClass = {cs.GR},
    url     = {https://arxiv.org/abs/2605.15305},
}

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

worldparticle-0.0.10.tar.gz (11.1 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

worldparticle-0.0.10-py3-none-any.whl (10.4 kB view details)

Uploaded Python 3

File details

Details for the file worldparticle-0.0.10.tar.gz.

File metadata

  • Download URL: worldparticle-0.0.10.tar.gz
  • Upload date:
  • Size: 11.1 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: uv/0.8.17

File hashes

Hashes for worldparticle-0.0.10.tar.gz
Algorithm Hash digest
SHA256 64ab841f7dcdafc1f987d7d2be6c2a8fec10e580366a71648d80ba04dd337bbe
MD5 3016e3eff47a07e2d67c22f2acf6d653
BLAKE2b-256 b0b8f30aef5f3c11685eb4de42dff5ed1156420fa9b22095a0b1a29d431e072c

See more details on using hashes here.

File details

Details for the file worldparticle-0.0.10-py3-none-any.whl.

File metadata

File hashes

Hashes for worldparticle-0.0.10-py3-none-any.whl
Algorithm Hash digest
SHA256 016a46238b43fa5b558bb06caf29baf51527be8fff262a8628b4d82873e0b5b5
MD5 6615264149e8e5ded6f69a2007bb66d3
BLAKE2b-256 3cde50f7f0419417e08997567c6e4eee1b651c9521d6561b19eeae0634c68552

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page