Numerical integration over given integral. How to do it in Julia?

[KZiemian]

In my current work I integrate numericaly some function over [0, \infty) using NumPy calling of Fortran libraries. I want to try do my problem using Julia, but I can’t find out-of-the-box library computing integrals.

Can someone tell my how numerical integration look now in Julia? Do I need call Fortran code directly? I try google something, but find almost nothing.

[HCrepaz]

Hi, There are several packages for numerical integration in Julia.

• For one-dimensional numerical integration QuadGK.jl https://github.com/JuliaMath/QuadGK.jl
• For higher dimensional quadrature HCubature.jl https://github.com/stevengj/HCubature.jl or NIntegration.jl https://github.com/pabloferz/NIntegration.jl

[stevengj]

More explicitly, do using Pkg; Pkg.add("QuadGK") to install the QuadGK package, and then do

using QuadGK
quadgk(f, 0, Inf, rtol=1e-3)

to compute \int_0^\infty f(x)dx (along with an error estimate) for a function f, to about 3–4 digits.

[CarloLucibello]

Does anyone know how to perfom numerical integration on a gpu?

[ChrisRackauckas]

@CarloLucibello

Does anyone know how to perfom numerical integration on a gpu?

Use GitHub - JuliaApproximation/FastGaussQuadrature.jl: Julia package for Gaussian quadrature to get the quadrature rates, use a CUArray and broadcast your function across the array, and then accumulate according to the quadrature weights.

[CarloLucibello]

this would work, thanks!