Handle duplicate index

[subrata]

A correct way to define a ITensor is the following

using ITensors

i = Index(2)
j = Index(2)
A = random_itensor(i, j)

Now what if I do the following

A = random_itensor(i,i)

It does not raise any warning or error.

But more than this, it gives incorrect results when setting values at specific positions. See the following example

i=Index(2)
j=Index(2)
A=random_itensor(i,i)
B=random_itensor(i,j)
@show A
@show B

println("After setting value at 1,2 for both A and B")
A[i=>1,i=>2]=999
B[i=>1,j=>2]=999
@show A
@show B

And here is the output:

A = ITensor ord=2
Dim 1: (dim=2|id=168)
Dim 2: (dim=2|id=168)
NDTensors.Dense{Float64, Vector{Float64}}
 2×2
 -0.8951148439378779  0.36977219901378616
 -0.4788833082871866  0.34125666612909394
B = ITensor ord=2
Dim 1: (dim=2|id=168)
Dim 2: (dim=2|id=84)
NDTensors.Dense{Float64, Vector{Float64}}
 2×2
 -1.5456158351721794  -0.26621937104864807
 -0.217078897761148    0.29058283067273155

After setting value at 1,2 for both A and B
A = ITensor ord=2
Dim 1: (dim=2|id=168)
Dim 2: (dim=2|id=168)
NDTensors.Dense{Float64, Vector{Float64}}
 2×2
 999.0                 0.36977219901378616
  -0.4788833082871866  0.34125666612909394
B = ITensor ord=2
Dim 1: (dim=2|id=168)
Dim 2: (dim=2|id=84)
NDTensors.Dense{Float64, Vector{Float64}}
 2×2
 -1.5456158351721794  999.0
 -0.217078897761148     0.29058283067273155

You can see A[i=>1,i=>2]=999 replace the value incorrectly at A[1,1] position.

I know nobody should use the Index in such a way to define an ITensor, but the ITensor should properly handle this situation.

Version:
Julia Version 1.11.2
ITensors v0.7.11

[mtfishman]

Duplicate indices are only checked when debug checks are enabled:

julia> using ITensors

julia> ITensors.enable_debug_checks()

julia> i = Index(2)
(dim=2|id=970)

julia> ITensor(i, i)
ERROR: Trying to create ITensors with collection of indices ((dim=2|id=970), (dim=2|id=970)). Indices must be unique.
Stacktrace:
  [1] error(s::String)
    @ Base ./error.jl:35
  [2] macro expansion
    @ ~/.julia/packages/ITensors/KvHvB/src/itensor.jl:94 [inlined]
  [3] macro expansion
    @ ~/.julia/packages/ITensors/KvHvB/src/global_variables.jl:180 [inlined]
  [4] ITensor
    @ ~/.julia/packages/ITensors/KvHvB/src/itensor.jl:91 [inlined]
  [5] ITensor(as::NDTensors.AllowAlias, st::NDTensors.EmptyStorage{NDTensors.EmptyNumber, NDTensors.Dense{…}}, is::Tuple{Index{…}, Index{…}})
    @ ITensors ~/.julia/packages/ITensors/KvHvB/src/itensor.jl:122
  [6] itensor(::NDTensors.EmptyStorage{NDTensors.EmptyNumber, NDTensors.Dense{NDTensors.EmptyNumber, Vector{NDTensors.EmptyNumber}}}, ::Vararg{Any}; kwargs::@Kwargs{})
    @ ITensors ~/.julia/packages/ITensors/KvHvB/src/itensor.jl:138
  [7] itensor
    @ ~/.julia/packages/ITensors/KvHvB/src/itensor.jl:138 [inlined]
  [8] ITensor
    @ ~/.julia/packages/ITensors/KvHvB/src/itensor.jl:181 [inlined]
  [9] ITensor(::Type{NDTensors.EmptyNumber}, ::Index{Int64}, ::Vararg{Index{Int64}})
    @ ITensors ~/.julia/packages/ITensors/KvHvB/src/itensor.jl:184
 [10] ITensor(::Index{Int64}, ::Vararg{Index{Int64}})
    @ ITensors ~/.julia/packages/ITensors/KvHvB/src/itensor.jl:186
 [11] top-level scope
    @ REPL[4]:1
Some type information was truncated. Use `show(err)` to see complete types.

We do that for efficiency, otherwise we would be checking for duplicate indices a lot. It is good practice to enable debug checks when developing ITensor code, and once it is working disable them for efficiency.

[subrata]

Okay, got it.