Baiyi Yu

Is there a way to calculate differentiation of `f(eigvals(::Hermitian))` where `f(x)` is pure julia? @dlfivefifty

I just got this error message: ``` ERROR: MethodError: no method matching eigvals!(::Hermitian{ForwardDiff.Dual{ForwardDiff.Tag{var"#102#103"{Float64, Float64, PositionBasis{-54.0, 9.0, Int64, Float64}, Vector{Float64}, PreallocationTools.DiffCache{Matrix{ComplexF64}, Vector{ComplexF64}}}, Float64}, Float64, 1}, Matrix{ForwardDiff.Dual{ForwardDiff.Tag{var"#102#103"{Float64, Float64, PositionBasis{-54.0, 9.0, Int64, Float64},...

Speculate that adding fields of `Ket`, `Bra`, and `combinedoperater` may resolve the use of tmp, but it may make `LazyProduct` a little bit fat.

As a test add `KTL` and `BTL` ``` mutable struct LazyProduct{BL,BR,F,T,KTL,BTR}

OK, I’ll open a PR. But I’m new in programming, may take some time to know how to use GitHub PR.

``` > git push -u origin EffientLazyProduct_mul remote: Permission to qojulia/QuantumOptics.jl.git denied to Lightup1. fatal: unable to access 'https://github.com/qojulia/QuantumOptics.jl.git/': The requested URL returned error: 403 ``` After setup for a...

Oh, thank you. Got it!😅

Thanks for your suggestions. >A third way is to put a cache in the operator struct. That way only needed arrays get created and the memory is freed when the...

You may just write a disrete distribution for you mixed state by using StatsBase with function `Weights()`. And for every trajetory, you choose an random initial state from this distribution.

OK thanks very much.