solve_mixed should check arguments for consistency

[HechtiDerLachs]

Try the following code:

A = ZZ[2 1 0;-1 -1 -1]
v = ZZ[2; -1]
Id = identity_matrix(ZZ, ncols(A))
solve_mixed(A, v, Id, zero(v))

According to the documentation of solve_mixed, this should look for solutions of A*x = v under the condition that x >= 0. The obvious solution is x = [1; 0; 0], but it throws an error.

Interestingly, the code runs when we replace A by A = ZZ[2 1; -1 -1].

CC: @lkastner @benlorenz

[thofma]

What is the error?

[HechtiDerLachs]

Yes, sorry. I get the following message:

ERROR: LoadError: DimensionMismatch("tried to assign 2-element array to 3×1 destination")
Stacktrace:
  [1] throw_setindex_mismatch(X::Vector{BigInt}, I::Tuple{Int64, Int64})
    @ Base ./indices.jl:193
  [2] setindex_shape_check
    @ ./indices.jl:248 [inlined]
  [3] _unsafe_setindex!(::IndexLinear, ::Matrix{BigInt}, ::Vector{BigInt}, ::Base.Slice{Base.OneTo{Int64}}, ::Int64)
    @ Base ./multidimensional.jl:896
  [4] _setindex!
    @ ./multidimensional.jl:887 [inlined]
  [5] setindex!
    @ ./abstractarray.jl:1267 [inlined]
  [6] augment(mat::Matrix{BigInt}, vec::Vector{BigInt})
    @ Oscar ~/Kaiserslautern/neuer_Oscar_Klon/Oscar.jl/src/PolyhedralGeometry/helpers.jl:158
  [7] affine_matrix_for_polymake(x::Tuple{Matrix{BigInt}, Vector{BigInt}})
    @ Oscar ~/Kaiserslautern/neuer_Oscar_Klon/Oscar.jl/src/PolyhedralGeometry/helpers.jl:35
  [8] Polyhedron{fmpq}(I::Tuple{Matrix{BigInt}, Vector{BigInt}}, E::Tuple{Matrix{BigInt}, Vector{BigInt}})
    @ Oscar ~/Kaiserslautern/neuer_Oscar_Klon/Oscar.jl/src/PolyhedralGeometry/Polyhedron/constructors.jl:101
  [9] Polyhedron
    @ ~/Kaiserslautern/neuer_Oscar_Klon/Oscar.jl/src/PolyhedralGeometry/Polyhedron/constructors.jl:23 [inlined]
 [10] solve_mixed(A::fmpz_mat, b::fmpz_mat, C::fmpz_mat, d::fmpz_mat)
    @ Oscar ~/Kaiserslautern/neuer_Oscar_Klon/Oscar.jl/src/PolyhedralGeometry/solving_integrally.jl:32
 [11] top-level scope
    @ /tmp/test.jl:9
 [12] include(fname::String)
    @ Base.MainInclude ./client.jl:444
 [13] top-level scope
    @ REPL[25]:1
 [14] top-level scope
    @ ~/.julia/packages/Infiltrator/r3Hf5/src/Infiltrator.jl:710

[thofma]

Probably some index error. In the meantime you could do

julia> solve_mixed(A, v, Id)
[1   0   0]`

(Assuming that you only have >= 0.)

[HechtiDerLachs]

The shortened version works? This part of code was suggested to me by @lkastner once and I just kept with it. But I can switch to the short one. However, the issue should eventually be resolved.

[fingolfin]

Your final vector in the argument list has the wrong length.

[HechtiDerLachs]

Now I really wish this was not true... But it probably is. Yes, sorry, I just didn't bother checking this again. I only remember this line being suggested to me by @lkastner once upon a time. But it had already been late in the evening at that point and probably something got lost in translation.

Either way, a plausibility check and a more informative error message would probably be helpful.

Thanks for catching my mistake and my apologies for the mess!

[thofma]

We should check the input for consistency.

[fingolfin]

I agree a more helpful error message would be good!