Add extension to convert ITensorMPS.MPS to Tenet.MPS and viceversa

[jofrevalles]

Summary

This PR resolves #247 by adding TenetITensorMPSExt extension, which adds two convert functions to convert ITensorMPS.MPS to Tenet.MPS and viceversa. Additionally, we also added some tests to cover for this function. @starsfordummies

TODO

• Clean some code.

Example

julia> using Tenet

julia> using ITensors; using ITensorMPS

julia> tenet_mps = rand(MPS; n=5, maxdim=30)
MPS (inputs=0, outputs=5)

julia> itensor_mps = convert(ITensorMPS.MPS, tenet_mps) # Tenet.MPS to ITensorMPS conversion
ITensors.ITensorMPS.MPS
[1] ((dim=2|id=565|"Site,n=1"), (dim=2|id=954|"Link,l=1"))
[2] ((dim=2|id=557|"Site,n=2"), (dim=2|id=954|"Link,l=1"), (dim=4|id=965|"Link,l=2"))
[3] ((dim=2|id=556|"Site,n=3"), (dim=4|id=965|"Link,l=2"), (dim=4|id=651|"Link,l=3"))
[4] ((dim=2|id=195|"Site,n=4"), (dim=4|id=651|"Link,l=3"), (dim=2|id=558|"Link,l=4"))
[5] ((dim=2|id=263|"Site,n=5"), (dim=2|id=558|"Link,l=4"))

julia> @test isapprox(parent(Tenet.contract(tenet_mps)), Array(ITensorMPS.contract(itensor_mps).tensor))
Test Passed

julia> itensor_mps = ITensorMPS.random_mps(siteinds(4, 5); linkdims=7)
ITensors.ITensorMPS.MPS
[1] ((dim=4|id=51|"Site,n=1"), (dim=7|id=149|"Link,l=1"))
[2] ((dim=7|id=149|"Link,l=1"), (dim=4|id=113|"Site,n=2"), (dim=7|id=746|"Link,l=2"))
[3] ((dim=7|id=746|"Link,l=2"), (dim=4|id=227|"Site,n=3"), (dim=7|id=555|"Link,l=3"))
[4] ((dim=7|id=555|"Link,l=3"), (dim=4|id=400|"Site,n=4"), (dim=4|id=984|"Link,l=4"))
[5] ((dim=4|id=984|"Link,l=4"), (dim=4|id=459|"Site,n=5"))

julia> tenet_mps = convert(MPS, itensor_mps) # ITensorMPS to Tenet.MPS conversion
MPS (inputs=0, outputs=5)

julia> form(tenet_mps)
MixedCanonical(1)

julia> contracted = permutedims(Tenet.contract(tenet_mps), [inds(tenet_mps; at=Site(i)) for i in 1:length(tensors(tenet_mps))])
4×4×4×4×4 Tensor{Float64, 5, Array{Float64, 5}}: ...

julia> @test isapprox(parent(contracted), Array(ITensorMPS.contract(itensor_mps).tensor))
Test Passed

[mofeing]

super cool!

[mofeing]

mmm if we set the patch version, will we have problems when v0.2.7 lands?

like, will it force it to use v0.2.6 or will it be compatible (but not included) up to v0.3.0?

in principle we should only mark major and minor versions, not patch version (unless a patch is completely required for it to work)

[jofrevalles]

Mmmm I don't know, I just put the version since we also had that for ITensors and ITensorNetworks. Do you really think this is a problem?

[mofeing]

I think we should try to just set the major.minor version. If you need a higher patch version, you should just call Pkg.update() then.

[starsfordummies]

nicee! It should be easy to make it work for MPO as well, yes?
so we can finally have MPO to MPS application in Tenet:

function evolve!(tenet_mpo, tenet_mps)
   itensor_mps = convert(ITensorMPS.MPS, tenet_mps)
   itensor_mpo = convert(ITensorMPS.MPS, tenet_mpo)
   o_psi = apply(itensor_mpo, itensor_mps)
   o_psi_tenet = convert(MPS, o_psi) 
   println("🙃 ")
end 

[jofrevalles]

I want to merge this, do you approve the PR? @mofeing @starsfordummies

[starsfordummies]

great, thanks!

[mofeing]

The tests failing are my fault. If they pass locally, you can merge.

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 52 lines in your changes missing coverage. Please review.

Project coverage is 0.04%. Comparing base (8a576c5) to head (a510320).
Report is 10 commits behind head on master.

Files with missing lines	Patch %	Lines
ext/TenetITensorMPSExt.jl	0.00%	52 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff            @@
##           master    #251      +/-   ##
=========================================
- Coverage    0.05%   0.04%   -0.01%     
=========================================
  Files          30      32       +2     
  Lines        1708    2036     +328     
=========================================
  Hits            1       1              
- Misses       1707    2035     +328     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.