cubie.batchsolving.BatchSolverConfig

Configuration utilities for the batch solver.

This module provides BatchSolverConfig, a small container that holds settings used when compiling and running the CUDA integration kernel.

Classes

BatchSolverConfig(precision, algorithm, ...)

Settings for configuring a batch solver kernel.

class cubie.batchsolving.BatchSolverConfig.BatchSolverConfig(precision: type = <class 'numpy.float32'>, algorithm: str = 'euler', duration: float = 1.0, warmup: float = 0.0, stream: int = 0, profileCUDA: bool = False)[source]

Bases: object

Settings for configuring a batch solver kernel.

precision

Data type used for computation. Defaults to float32.

Type:: type, optional

algorithm

Name of the integration algorithm. Defaults to 'euler'.

Type:: str, optional

duration

Total integration duration in seconds. Defaults to 1.0.

Type:: float, optional

warmup

Length of the warm-up period before outputs are stored. Defaults to 0.0.

Type:: float, optional

stream

Identifier for the CUDA stream to execute on. None defaults to the solver’s stream. Defaults to 0.

Type:: int, optional

profileCUDA

If True CUDA profiling is enabled. Defaults to False.

Type:: bool, optional

algorithm: str

duration: float

precision: type

profileCUDA: bool

stream: int

warmup: float