Without compiling or testing:
Transpose:
Inverse (A is square):
var C = A.LU().Solve(DenseMatrix.Identity(A.ColumnCount));
Pseudo Inverse (A rows > columns):
var D = A.QR().Solve(DenseMatrix.Identity(A.RowCount));
Pseudo Inverse (A columns > rows) (we currently don't support LQ):
var E = A.Transponse().QR().Solve(DenseMatrix.Identity(A.ColumnCount)).Transpose();
(looks like we may want to implement a simple Inverse() or PseudoInverse() method that covers all these cases)
