Matrix exponential Suppose that our state is a vector variable: x t Ax t We can imagine that the solution will have a “matrix exponential” form: x t x 0 exp At For any square matrix A, the matrix exponential exp(A) is a square matrix function. We can compute it using Taylor series expansion. d2 f t 2 dt t 0 df f t f 0 dt t2 exp At I At A 2 tn n exp At A n 0 n ! 2 In Matlab, exp(A) is computed as expm(A). In Mathematica, use MatrixExp[A]. t2 2 t 0 tn A n! n dn f n dt tn n! t 0 Some properties of the matrix exponential Using Taylor series expansion, one can show the following properties of the matrix exponential: exp(0t ) I exp A t1 t2 exp At1 exp At2 d exp At A exp At exp At A dt Other properties of the matrix exponential: exp( A) exp( A) I exp A exp B exp A B only if AB BA 0 Solution of continuous LTI state equations (vector condition) x t Ax t Cu t exp At x t exp At Ax t exp At Cu t d exp At x t exp At Cu t dt exp At x t 0 exp At Cu t dt 0 exp A x x 0 exp At Cu t dt 0 x exp A x 0 exp A exp At Cu t dt 0 x exp A x 0 exp A t Cu t dt 0 Solution of discrete LTI state equations x k 1 Ax Cu k x Ax Cu 1 0 k 0 x 2 Ax1 Cu 1 A2 x 0 ACu 0 Cu 1 x k Ax k 0 k 1 Ak 1 mCu m 0 m Relating discrete and continuous representation of a linear system x t exp Act x 0 exp Act exp Ac Cu d t 0 t k 1 x k 1 x k 1 x k 1 exp Ac k 1 x 0 exp Ac k 1 k 1 0 exp Ac Cc u d exp Ac k 1 x 0 exp Ac k 1 exp Ac Cc u d k 0 exp Ac k 1 k 1 k exp Ac Cc u d k exp Ac exp Ac k x 0 exp Ac k exp Ac Cc u d 0 exp Ac k 1 k 1 k exp Ac x k k 1 k exp Ac x k Ac1 exp Ac Cc u d exp A k 1 exp Ac k 1 Cc u d c exp Ac x k Ac1 I exp Ac Cc u k k 1 k C u k c Assume that u(t) is constant between the two sampling intervals. Discrete and continuous representation of a linear system (noise free scenario) Continuous system x t Ac x t Cc u t Gc y t Bc x t Dc u t Discrete system x k 1 Ad x k Cd u k Gd k k k y Bd x Dd u sampling interval Ad exp Ac Cd Ac1 exp Ac I Cc Bd Bc Dd Dc