Issue
One may not want to add TensorCrossInterpolation.jl as a dependency to use ACI. For example, I only have T4ATensorCI.jl and would get name clashes if I were to add TensorCrossInterpolation.jl . Also, the TensorTrain type exists in multiple tensor4all Julia repositories.
Proposed solution
One could just add an overload to elementwise that takessitetensors::Array{Array{ValueType,N}} for each input TT and also outputs an Array{Array{ValueType,N}} instead of a TensorTrain.
Issue
One may not want to add
TensorCrossInterpolation.jlas a dependency to use ACI. For example, I only haveT4ATensorCI.jland would get name clashes if I were to addTensorCrossInterpolation.jl. Also, theTensorTraintype exists in multiple tensor4all Julia repositories.Proposed solution
One could just add an overload to
elementwisethat takessitetensors::Array{Array{ValueType,N}}for each input TT and also outputs anArray{Array{ValueType,N}}instead of aTensorTrain.