`culebra.trainer.topology` module¶

Topologies for distributed trainer algorithms.

Functions¶

ring_destinations(origin: int, num_subtrainers: int, offset: int = 1) → List[int]¶

Return the destinations reachable from origin.

Parameters:

origin (int) – The index of the origin subpopulation trainer
num_subtrainers (int) – The number of subpopulations
offset (int, optional) – Offset applied to origin, defaults to DEFAULT_RING_OFFSET

Returns:

The direct destinations from origin

Return type:

list of subpopulation trainer indexes

full_connected_destinations(origin: int, num_subtrainers: int) → List[int]¶

Return the destinations reachable from origin.

Parameters:

origin (int) – The index of the origin subpopulation trainer
num_subtrainers (int) – The number of subpopulations

Returns:

The direct destinations from origin

Return type:

list of subpopulation trainer indexes

Attributes¶

DEFAULT_RING_OFFSET = 1¶: Default offset for the ring topology.

This work is supported by projects New Computing Paradigms and Heterogeneous Parallel Architectures for High-Performance and Energy Efficiency of Classification and Optimization Tasks on Biomedical Engineering Applications (HPEE-COBE), with reference PGC2018-098813-B-C31, and Advanced Methods of Biomedical Data Analysis and Brain Modeling Optimized for High-Performance and Energy-Efficient Computing (HPEEC-BIOBRAIN), with reference PID2022-137461NB-C31, both funded by the Spanish Ministerio de Ciencia, Innovación y Universidades, and by the European Regional Development Fund (ERDF).