[NDTensors] [BUG] Constructing a Tensor from a view of another tensor can fail

[kmp5VT]

Description of bug

There is an issue when one tries to construct a new tensor from the view of another tensor. It comes from the parameters of the new tensor being defined before the storage is copied, i.e. a type mismatch.

Minimal runnable code

using ITensors: Index
using NDTensors: randomTensor, Tensor
i,j = Index.((4,4))
t = randomTensor((i,j))
Tensor(storage(t[1:2, 1:2]), (2, 2))

Output of minimal runnable code

ERROR: MethodError: Cannot `convert` an object of type Vector{Float64} to an object of type Base.ReshapedArray{Float64, 1, SubArray{Float64, 2, Matrix{Float64}, Tuple{UnitRange{Int64}, UnitRange{Int64}}, false}, Tuple{Base.MultiplicativeInverses.SignedMultiplicativeInverse{Int64}}}

Closest candidates are:
  convert(::Type{T}, ::T) where T
   @ Base Base.jl:84
  convert(::Type{T}, ::LinearAlgebra.Factorization) where T<:AbstractArray
   @ LinearAlgebra ~/.julia/juliaup/julia-1.10.2+0.aarch64.apple.darwin14/share/julia/stdlib/v1.10/LinearAlgebra/src/factorization.jl:108
  convert(::Type{T}, ::T) where T<:AbstractArray
   @ Base abstractarray.jl:16
  ...

Stacktrace:
 [1] convert(::Type{NDTensors.Dense{Float64, Base.ReshapedArray{…}}}, D::NDTensors.Dense{Float64, Vector{Float64}})
   @ NDTensors ~/.julia/dev/ITensors/NDTensors/src/dense/dense.jl:142
 [2] NDTensors.DenseTensor{…}(::NDTensors.AllowAlias, storage::NDTensors.Dense{…}, inds::Tuple{…})
   @ NDTensors ~/.julia/dev/ITensors/NDTensors/src/tensor/tensor.jl:27
 [3] NDTensors.DenseTensor{…}(::NDTensors.NeverAlias, storage::NDTensors.Dense{…}, inds::Tuple{…})
   @ NDTensors ~/.julia/dev/ITensors/NDTensors/src/tensor/tensor.jl:36
 [4] Tensor(as::NDTensors.NeverAlias, storage::NDTensors.Dense{Float64, Base.ReshapedArray{…}}, inds::Tuple{Int64, Int64})
   @ NDTensors ~/.julia/dev/ITensors/NDTensors/src/tensor/tensor.jl:67
 [5] Tensor(storage::NDTensors.Dense{Float64, Base.ReshapedArray{…}}, inds::Tuple{Int64, Int64})
   @ NDTensors ~/.julia/dev/ITensors/NDTensors/src/tensor/tensor.jl:81
 [6] top-level scope
   @ Untitled-1:16
Some type information was truncated. Use `show(err)` to see complete types.

Version information

• Output from versioninfo():

julia> versioninfo()
Julia Version 1.10.2
Commit bd47eca2c8a (2024-03-01 10:14 UTC)
Build Info:
  Official https://julialang.org/ release
Platform Info:
  OS: macOS (arm64-apple-darwin22.4.0)
  CPU: 10 × Apple M1 Max
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-15.0.7 (ORCJIT, apple-m1)
Threads: 1 default, 0 interactive, 1 GC (on 8 virtual cores)
Environment:
  LD_LIBRARY_PATH = /Users/kpierce/Software/triqs/triqs_install/lib::/opt/intel/oneapi/mkl/latest/lib
  JULIA_EDITOR = code
  JULIA_NUM_THREADS = 

Proposed solution

The code stack calls

function Tensor(as::AliasStyle, storage::TensorStorage, inds::Tuple)
  return Tensor{eltype(storage),length(inds),typeof(storage),typeof(inds)}(
    as, storage, inds
  )
end

which calls

function Tensor{ElT,N,StoreT,IndsT}(
  ::NeverAlias, storage::TensorStorage, inds
) where {ElT,N,StoreT<:TensorStorage,IndsT}
  return Tensor{ElT,N,StoreT,IndsT}(AllowAlias(), copy(storage), inds)
end

In the first call

typeof(storage(t[1:2, 1:2])) = NDTensors.Dense{Float64, Base.ReshapedArray{Float64, 1, SubArray{Float64, 2, Matrix{Float64}, Tuple{UnitRange{Int64}, UnitRange{Int64}}, false}, Tuple{Base.MultiplicativeInverses.SignedMultiplicativeInverse{Int64}}}}

and when the type of the copied storage is

typeof(copy(storage(t[1:2, 1:2])))
NDTensors.Dense{Float64, Vector{Float64}}

So my proposal is to modify the second tensor constructor to be

function Tensor{ElT,N,StoreT,IndsT}(
  ::NeverAlias, storage::TensorStorage, inds
) where {ElT,N,StoreT<:TensorStorage,IndsT}
  store = copy(storage)
  return Tensor{ElT,N,typeof(store),IndsT}(AllowAlias(), store, inds)
end

which returns the following output

julia> Tensor(storage(t[1:2, 1:2]), (2,2)) 
Dim 1: 2
Dim 2: 2
NDTensors.Dense{Float64, Vector{Float64}}
 2×2
 -0.8736954509592826   0.30952483312000373
  1.6882572502541735  -0.7830143316102207