Parallele composition of FSTs?

[Chenxi-Nie]

Dear K2 developer,

I need to do FST composition 1,000 times, right now I do this using a for loop on one single thread, like this

fsa_l = [] # create a list of 1000 FSAs
fst = # a fst to compose 
result = []
for a in range(1000):
     fsa_a = fsa_l[a]
     result.append(k2.compose(fsa_a, fst)

I tried to use create_fsa_vec to create a fsa vector and do compose using the fsa vector and the fst, but I do not think k2.compose works directly on a fsa vector.

Are there any (straight forward) way to do this? Or should I just use other python parallelization packages to do this task?

Best regards,
Chenxi