Stackoverflow Error with QobjEvo depending on multiplication order

[Gavin-Rockwood]

Bug Description

There is a stack overflow error depending on the order of scalar and QobjEvo multiplication. I discovered this while trying to write a positive integer power function for QobjEvo objects. I'm not sure if this is an issue with purely SciMLOperators or ifs a bug in QobjEvo.

Code to Reproduce the Bug

a = QobjEvo((qeye(2), (p,t) -> 1))
println("This works, ", 1*a*a*1)
println("This works, ", 1*(a*a))
println("This works, ", (a*a)*1)
println("This does not work, ", 1*(a*a)*1)

Code Output

This works, 
Quantum Object Evo.:   type=Operator()   dims=[2]   size=(2, 2)   ishermitian=true   isconstant=false
ScalarOperator(1) * (ScalarOperator(0.0 + 0.0im) * DiagonalOperator(2 × 2) * ScalarOperator(0.0 + 0.0im) * DiagonalOperator(2 × 2))
This works, 
Quantum Object Evo.:   type=Operator()   dims=[2]   size=(2, 2)   ishermitian=true   isconstant=false
ScalarOperator(1) * (ScalarOperator(0.0 + 0.0im) * DiagonalOperator(2 × 2) * ScalarOperator(0.0 + 0.0im) * DiagonalOperator(2 × 2))
This works, 
Quantum Object Evo.:   type=Operator()   dims=[2]   size=(2, 2)   ishermitian=true   isconstant=false
ScalarOperator(1) * (ScalarOperator(0.0 + 0.0im) * DiagonalOperator(2 × 2) * ScalarOperator(0.0 + 0.0im) * DiagonalOperator(2 × 2))
StackOverflowError:

Stacktrace:
 [1] *(α::ScalarOperator{Int64, SciMLOperators.FilterKwargs{typeof(SciMLOperators.DEFAULT_UPDATE_FUNC), Tuple{}}}, x::UniformScaling{Int64})
   @ SciMLOperators ~/.julia/packages/SciMLOperators/DKrpP/src/scalar.jl:366
 [2] *(α::ScalarOperator{Int64, SciMLOperators.FilterKwargs{typeof(SciMLOperators.DEFAULT_UPDATE_FUNC), Tuple{}}}, x::UniformScaling{Int64})
   @ SciMLOperators ~/.julia/packages/SciMLOperators/DKrpP/src/scalar.jl:369

Expected Behaviour

I would have expected all 4 lines to return the same object.

Your Environment

QuantumToolbox.jl: Quantum Toolbox in Julia
≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡≡
Copyright © QuTiP team 2022 and later.
Current admin team:
    Alberto Mercurio and Yi-Te Huang

Package information:
====================================
Julia              Ver. 1.11.1
QuantumToolbox     Ver. 0.34.0
SciMLOperators     Ver. 1.4.0
LinearSolve        Ver. 3.24.0
OrdinaryDiffEqCore Ver. 1.26.2

System information:
====================================
OS       : macOS (arm64-apple-darwin22.4.0)
CPU      : 14 × Apple M4 Pro
Memory   : 48.0 GB
WORD_SIZE: 64
LIBM     : libopenlibm
LLVM     : libLLVM-16.0.6 (ORCJIT, apple-m1)
BLAS     : libopenblas64_.dylib (ilp64)
Threads  : 1 (on 10 virtual cores)

Additional Context

No response

[Gavin-Rockwood]

I can also add that the same issue arises here:

a = QobjEvo((qeye(2), (p,t) -> 1))
println("This works, ", 1*(a+a*a), "\n")
println("This works, ", (a+a*a*1), "\n")
println("This works, ", (a+a*(a*1))*1, "\n")
println("This does not work, ", (a+a*a*1)*1)

The last line returns the exact same error message as above.

[ytdHuang]

The issue originates from the redundant way we apply (*) multiplication between an AbstractQuantumObject and a Number.

I have made PR #525 to fix it

[albertomercurio]

The problem can be fixed directly from SciMLOperators.jl, so that we keep the QuantumToolbox.jl code clean. I will make a PR there soon.

[albertomercurio]

I made a PR here

• Fix StackOverFlowError on multiplication between ScalarOperator and UniformScalar SciML/SciMLOperators.jl#300

Let's wait it to be merged and a new version of SciMLOperators.jl to be released. I have already tested it locally with QuantumToolbox.jl

[ytdHuang]

Hmm, I didn't notice that we need that I for GPU compatibility.