I need to compute the Laplacian spectrum of a ton of graphs and was wondering if it's possible to use all CPU cores instead of only one.
SInce you have to run the computation on various graphs, you can do some basic parallelism provided by the @paralleldecorator, see https://doc.sagemath.org/html/en/refe...
EDIT if tou use the @parallel decorator, you have to specify the number of cpus, e.g. @parallel(ncpus=8)
Thanks! I was trying to use Python's multiprocessing library before, I'll try using the decorator then.
I'm experimenting with the @parallel decorator but the spectrum() method doesn't seem to accept more than one input at a time.
vertex = 7
parameters = str(vertex) + " -c " + str(vertex) + ":" + str(vertex)
graphs_nauty = list(graphs.nauty_geng(parameters))@parallel
def laplacian_spectrum(graph):
return graph.spectrum(laplacian = True)
spectrum_all = laplacian_spectrum(graphs_nauty)
for i in spectrum_all:
print(i)
@parallel with the spectrum() method instead of using it on a function I defined myself?The way you did works for me, if i replace @parallel with @parallel(ncpus=8)
It isn't working for me at all. Which Sage version are you using?
Asked: 3 years ago
Seen: 314 times
Last updated: Jun 30 '21
Copyright Sage, 2010. Some rights reserved under creative commons license. Content on this site is licensed under a Creative Commons Attribution Share Alike 3.0 license.
