I’m doing a VQE simulation that is very similar to
I’m doing much larger simulations, where I cap the MPS bond dimension. I’m a little lost on how to optimize this script by taking advantage of the many threads I have.
My current execution of my script looks like
export OMP_NUM_THREADS=16
julia -t $OMP_NUM_THREADS --sysimage ~/.julia/sysimages/sys_itensors.so -i hubbard.jl
Also, I have the following libraries imported
using ITensors
using ITensorParallel
using OptimKit
using Random
using Zygote
using JLD2
using Printf
using LinearAlgebra
using MKL
- Am doing the right thing in how I’m addressing the number of threads I should use? Is there an optimal number of threads to use?
- Do I have right libraries included?
- Can I easily port something like this onto GPUs?
Thanks in advance!