Parallelisation at C or R level

ppapasaikas / griph

Graph Inference of Population Heterogeneity.

https://ppapasaikas.github.io/griph/

GNU General Public License v3.0

10 stars 3 forks source link

Parallelisation at C or R level #8

Open ppapasaikas opened 7 years ago

ppapasaikas commented 7 years ago

Parallelisation currently is done both at the c level using openMP (code ported from largeVis) and at the R level. It would make sense to switch everything to C level implementation (with the possible exception of QUIC for glasso).

mbstadler commented 7 years ago

this has been achieved for hellinger, canberra and correlation distance calculations (now all using OpenMP if it is available). The former R-based parallelization remains as a fall-back for environments without OpenMP support.

Currently, only dense matrices are supported.