Error for geometric product of null basis and negative origin

[karlwessel]

I get an error for the geometric product of v∅ and -v∞:

julia> using Grassmann

julia> @basis S"∞∅+"
(⟨∞∅+⟩, v, v∞, v∅, v₁, v∞∅, v∞₁, v∅₁, v∞∅₁)

julia> v∅*v∞
-1 - 1v∞∅

julia> v∅*(-v∞)
ERROR: MethodError: no method matching SValue{⟨∞∅+⟩,G,B,T} where T where B where G(::Int64, ::MultiVector{Int64,⟨∞∅+⟩,8})
Closest candidates are:
  SValue{⟨∞∅+⟩,G,B,T} where T where B where G(::T) where {V, T} at .julia/packages/Grassmann/H9Zog/src/multivectors.jl:97
  SValue{⟨∞∅+⟩,G,B,T} where T where B where G(::Any, ::SValue{V,G,B,T} where T where B) where {V, G} at .julia/packages/Grassmann/H9Zog/src/multivectors.jl:90
  SValue{⟨∞∅+⟩,G,B,T} where T where B where G(::Any, ::MValue{V,G,B,T} where T where B) where {V, G} at .julia/packages/Grassmann/H9Zog/src/multivectors.jl:91
  ...
Stacktrace:
 [1] *(::Basis{⟨∞∅+⟩,1,0x0000000000000002}, ::SValue{⟨∞∅+⟩,1,v∞,Int64}) at .julia/packages/Grassmann/H9Zog/src/algebra.jl:157
 [2] top-level scope at none:0

[Dream Scatter]

Ah yes, that's because the product is supposed to be a MultiVector but an SValue is expected

julia> v∅*basis(-v∞)
-1 - 1v∞∅

julia> SValue{V}(-1,ans)
ERROR: MethodError: no method matching SValue{⟨∞∅+⟩,G,B,T} where T where B where G(::Int64, ::MultiVector{Int64,⟨∞∅+⟩,8})
Closest candidates are:

and the definition of the product is expecting the SValue instead of MultiVector value

*(a::Basis{V},b::SValue{V}) where V = SValue{V}(b.v,a*basis(b))

the fix will have to require a creative way of adapting the code without introducing type instability

[karlwessel]

By now it didn't happen anywhere in any useful or productive code I use.
Only in a generic testcase I've written to check the basic properties of an algebra. So no real hurry from my side at the moment for the fix.

[Dream Scatter]

You're welcome to help increase the test coverage with runtests.jl

[Dream Scatter]

@karlwessel this issue has now been resolved

julia> (-v∞) * v∅
1 - 1v∞∅

julia> v∞ * (-v∅)
1 - 1v∞∅

all your tests are now passing

[karlwessel]

Thanks for fixing, and sorry for the slow response!

With this issue fixed the associativity and distributivity tests now also pass for Vectorspace S"∞∅+". I therefore added PR #36 which enables those tests.