Is the innovation here not so much resize, but allowing the creation of greater than 1×1 matrices in algebra (we've only allow automatic up-shifting of number into a 1×1 matrix till now?
This would allow creation of arbitrary matrices (which is useful), but I think was decided against in the past?
perhaps supporting matrix in an algebra would be enough...
mxMatrix(name="a", values =1:12, nrow=3,ncol=4)
mxAlgebra(matrix(a,1,12), name ="vector")
mxAlgebras can currently evaluate to any arbitrary size matrix, e.g. solve(I-A) %% S %% t(solve(I-A)) often is an NxN matrix. The innovation is mostly that you could take a vector, restructure it as a matrix, and then use matrix operations on the new matrix.
Comments
#1
Is the innovation here not so much resize, but allowing the creation of greater than 1×1 matrices in algebra (we've only allow automatic up-shifting of number into a 1×1 matrix till now?
This would allow creation of arbitrary matrices (which is useful), but I think was decided against in the past?
perhaps supporting matrix in an algebra would be enough...
#2
mxAlgebras can currently evaluate to any arbitrary size matrix, e.g. solve(I-A) %% S %% t(solve(I-A)) often is an NxN matrix. The innovation is mostly that you could take a vector, restructure it as a matrix, and then use matrix operations on the new matrix.
#3
Yes, I was meaning not that algebras can't output arbitrary matrices, but that these are all illegal, no?
#4
Not a good idea to have unnamed objects, and auto-named objects