solprop.backends.mlx_gpu.batch#

Fixed-schedule lockstep batches of mlx members (SEP 33 section 6.3, Q10).

LockstepRunner asks each member for its trial and its host norm in turn, so a member whose trial synchronises could not queue the others first. MlxBatchRunner keeps the public select_batch interface and every member’s own state, ledgers and statistics, and splits each trial: it builds every member’s step graph, evaluates them together, then reads each member’s norm.

Classes

MlxBatchRunner(members)

Advance loaded mlx members through one fixed schedule, evaluated together.

class solprop.backends.mlx_gpu.batch.MlxBatchRunner(members)[source]#

Bases: LockstepRunner

Advance loaded mlx members through one fixed schedule, evaluated together.

Parameters:

memberssequence of MlxBackend: Loaded members with matching shapes, fixed schedules and Runge-Kutta tableaus. Ownership transfers on successful construction.

metadata()[source]#

Return the batch’s settings.

Returns:

dict: The runner’s, with the trials evaluated together.

name = 'mlx'#