I'm trying to take the matrix exponential of a skew symmetric rotation matrix, S
. I expect that the result is Rodrigues' rotation formula I + sin(theta)*S + (1-cos(theta))*S*S
. However, Mathematica returns something that doesn't look like that formula and it's result is trying to take the square root of a negative number.
Here's my code:
S = { { 0, -omegaz, omegay }, {omegaz, 0, -omegax}, {-omegay, omegax, 0} };
FullSimplify[MatrixExp[S]]
That results in Mathematica:
Am I doing something wrong?
The result are the same.
In Rodrigues' rotation formula, the skew matrix is made from the unit vector, therefore you have assumption:
1 == omegax^2 + omegay^2 + omegaz^2
And you need to use:
MatrixExp[theta S]
And if you run:
rod = IdentityMatrix[3] + Sin[theta] S + (1 - Cos[theta]) MatrixPower[S, 2]
rod = FullSimplify[rod, Assumptions -> {omegax^2 + omegay^2 + omegaz^2 == 1}]
expS = FullSimplify[MatrixExp[theta S], Assumptions -> {omegax^2 + omegay^2 + omegaz^2 == 1}]
rod == b
(* True *)
Thus, the Mathematica calculates the rotation matrix correctly.