Remove specialized vector-matrix multiplication methods (#55538)

The specialized methods for `TransposeAbsMat` and `AdjointAbsMat` seem
unnecessary, as these are also `AbstractMatrix`es, and are treated
identically. I've also added a `require_one_based_indexing` check on the
vector to avoid accepting `OffsetArray`s.

Author: jishnub

stdlib/LinearAlgebra/src/matmul.jl

@@ -61,9 +61,10 @@ function (*)(A::AbstractMatrix{T}, x::AbstractVector{S}) where {T,S}
 end
 
 # these will throw a DimensionMismatch unless B has 1 row (or 1 col for transposed case):
-(*)(a::AbstractVector, tB::TransposeAbsMat) = reshape(a, length(a), 1) * tB
-(*)(a::AbstractVector, adjB::AdjointAbsMat) = reshape(a, length(a), 1) * adjB
-(*)(a::AbstractVector, B::AbstractMatrix) = reshape(a, length(a), 1) * B
+function (*)(a::AbstractVector, B::AbstractMatrix)
+    require_one_based_indexing(a)
+    reshape(a, length(a), 1) * B
+end
 
 # Add a level of indirection and specialize _mul! to avoid ambiguities in mul!
 @inline mul!(y::AbstractVector, A::AbstractVecOrMat, x::AbstractVector,

stdlib/LinearAlgebra/test/matmul.jl

@@ -1120,4 +1120,14 @@ end
     end
 end
 
+@testset "vector-matrix multiplication" begin
+    a = [1,2]
+    A = reshape([1,2], 2, 1)
+    B = [1 2]
+    @test a * B ≈ A * B
+    B = reshape([1,2], 2, 1)
+    @test a * B' ≈ A * B'
+    @test a * transpose(B) ≈ A * transpose(B)
+end
+
 end # module TestMatmul