Could this simple function go faster?

[lrnv]

@stevengj

The basic issue here is that, for BigFloat and BigInt (“bignum”) calculations, the arithmetic cost will typically dominate any algorithm. Almost all that matters is probably the number of bignum operations. Type stability, allocation of temporary arrays, and all of the other usual Julia considerations are likely to have relatively small impact. So, you need to focus on algorithmic improvements that reduce the number of arithmetic operations.

This is clearly an issue. I could do more precomputations to replace prod(P.BINS[CartesianIndex.(k_minus_one_in_deg,j_arr)]), which demands n product of bigfloats by only one indexing, but it gave me an OutOfMemory error if i want to cover all cases upfront

What about GitHub - JuliaArbTypes/ArbFloats.jl: Much faster than BigFloat at precisions up to 3,500 bits (1050 digits) ?