See . I don't think Maple would be particularly useful for this sort of project. Code-breaking generally involves testing a huge number of possibilities, where a lower-level compiled language would have a great advantage in terms of speed.

Under Windows XP, Pentium 4, 3 GHz Maple V Release 4, grobner[gsolve](caprasse,{x,y,z,t}): .516 seconds discriminant4: .047 seconds gerdt: .235 seconds gonnet: .187 seconds Maple 11 Classic. caprasse: 1.078 seconds discriminant4: 0.203 seconds gerdt: 1.344 seconds gonnet: 1.047 seconds Cheers, Robert

It should be easy enough, with LinearAlgebra[MatrixFunction](A,exp(z),z); ... unless A is, say, a 3 x 3 Matrix with integer entries and an irreducible characteristic polynomial.