Skip to main content

Crate physics_in_parallel

Crate physics_in_parallel 

Source
Expand description

Physics in Parallel provides backend-agnostic numerical containers, spatial tools, and reusable physical models.

§Breaking release: 4.1.0-alpha

PiP 4.1 alpha replaces the 3.x API and tightens validation compared with the 4.0 alphas. Review the migration notes before upgrading code or persisted data. Pin the exact alpha version; public APIs and serialized representations may change between alpha releases.

Start with the documentation index for installation, examples, performance, correctness, and advanced API guides.

§Thread-pool responsibility

PiP uses the active Rayon pool and never creates one. Applications must configure one shared pool before concurrent work begins or let their workflow runner do so. Independent pools can oversubscribe the machine. Normally call threading::set_max_threads once during startup to set the maximum worker participation requested by any single PiP method.

§API tiers

Domain roots are authoritative. Independent convenience preludes expose basic numerics, ready physical models, and opt-in advanced facilities:

use physics_in_parallel::prelude::basic::*;
use physics_in_parallel::prelude::models::*;
use physics_in_parallel::prelude::advanced::*;

Start with the basic or model API. Raw storage and generic engines are advanced; implementation modules are private. PiP owns runtime scientific values, while applications own user configuration, persistence policy, and workflow integration.

Modules§

advanced
Opt-in backend, interchange, and generic-engine facilities.
math
Core math foundations.
models
Domain-level model modules.
prelude
Explicit API-tier preludes.
rng
PiP-owned, fully resolved randomness.
space
Core space abstractions and utilities.
threading
Process-wide parallel execution policy.