I am new to this language, tried a few speed up tips and this is the best i got. Is there anything else that can be done to get a significant performance increase? Can you show me how? using Plots, Printf, OffsetArrays function loops(nn::Integer, kk::Integer) if nn < 0 || kk <= 0 retur…

[image] gdalle: fac = factorial.(0:BigInt(n)) It would be more modular to separate the factorial precomputation from the other parts of the code, like this: module FactorialTables struct FactorialTable{T <: Number} table::Vector{T} FactorialTable{T}(n::Integer) where {T <: Number} = …

It’s a cool idea indeed but perhaps mildly scary for a beginner :scream: I tried to stick to the original code as much as possible, but I agree that we can do much cleaner

cnt += recur * (fac[n+1] ÷ fac[n-j*k+1]) ÷ (fac[j+1] * (k^j)) This for instance still has a lot of duplicate computations, the division of factorials is only there for notational simplicity

[image] gdalle: n - j * k This expression also repeats multiple times. I don’t think Julia can do common subexpression elimination on account of BigInt operations being done by external C code.

Poor OP, by the time they come back we’ll have completely disfigured their code :rofl:

Thanks, yes I read about BigInts but really didn’t need the accuracy and things were working with floating points so I settled on the table i created or using the gamma() function which works with floating points for real value “factorials”. The latter took twice as long as the table but I am new at…

Thanks for the tips about Julia speed ups and I do appreciate the memoization. That was the next step I planned to look at so you saved me a lot of time. Good job!

If there’s a need to make gdalle’s dynamic programming solution even more performant, while still using BigInts, look into the MutableArithmetics package: https://juliahub.com/ui/Packages/MutableArithmetics/EoEec/1.3.0

If not all performance needs to be squeezed out, just memoizing the straight-forward implementation might already be enough: using Memoize @memoize Dict function facm(n::T) where {T <: Integer} if iszero(n) one(T) else n * facm(n - 1) end end @memoize Dict function loo…

[image] Mark_Szlazak: if k == 1 || n == 0 return 1.0 end My generic advice here is that if you want recursion to be fast (in any language), you should ideally enlarge the base case (“coarsen” the recursion) — see this comment . (That being said, maybe with memo-ization here you won’t cal…

Help with speeding up this code

General Usage Performance

Topic		Replies	Views
Repeatedly multiplying a Float64 is fastest in chunks of <51 terms Performance	11	1392	March 28, 2020
Need help with performance in a very large loop Performance performance	28	1697	May 28, 2020
Reduce allocations and speed up Performance	15	1767	July 27, 2021
For loop performance vs "functional" performance Performance advent-of-code	17	703	January 4, 2025
Can this be written even faster? (CPU) Performance	34	1310	February 9, 2024

Help with speeding up this code

Related topics