fixed fast coefficient computation #515

Job	Run time
build-mpi (3.9)	5m 34s
build-docs (3.8)	11s
build-mpi (3.10)	1m 36s
build-numpy (3.8, short)	1m 49s
build-jax (3.8)	1m 52s
build-numpy (3.8, optimize)	1m 50s
build-numpy (3.8, multivariate)	1m 27s
build-jax (3.9)	51s
build-torch (3.8)	2m 13s
build-jax (3.10)	58s
build-numpy (3.8, optimize-experimental)	2m 6s
build-torch (3.9)	3m 15s
build-numpy (3.9, short)	52s
build-torch (3.10)	2m 3s
build-numpy (3.9, optimize)	51s
build-numpy (3.9, multivariate)	49s
build-numpy (3.9, optimize-experimental)	50s
build-numpy (3.10, short)	55s
build-numpy (3.10, optimize)	50s
build-numpy (3.10, multivariate)	53s
build-numpy (3.10, optimize-experimental)	50s
	32m 35s

Provide feedback