Adding @inline compiler hints to the Boltzmann integration functions, which in turn are called by the cost functions sent to NLOpt.jl, may give us a little speedup for little work. Will investigate over the next week or two.
Tested using non-singular, constant sample rate data, @inline gave no improvement. Julia compiler probably smart enough to optimise where appropriate anyway. Closing for now.
Adding
@inline
compiler hints to the Boltzmann integration functions, which in turn are called by the cost functions sent to NLOpt.jl, may give us a little speedup for little work. Will investigate over the next week or two.