Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Solution for Sylvester Equation

  1. Mar 26, 2012 #1
    Given n by n matrices A, B, C. I know how to solve the Sylvester equation

    AX + XB + C = 0

    using the matlab command >> X=lyap(A,B,C)

    But how do we solve the extended Sylvester equation
    AX + XB + CXD + E = 0 ?

    Either numerical or analytical method I'm willing to learn.
  2. jcsd
  3. Mar 26, 2012 #2


    User Avatar
    Science Advisor

    Hey matematikawan.

    Have you tried just expanding out the system, collecting the terms and getting a form of AX = B?

    In other words you get a matrix corresponding to ZX = F and then apply the formula X = Z^-1 x F. For the Z matrix you will need to do some algebra to get this and in this particular example, F = -E.
  4. Mar 26, 2012 #3
    I don't think it is possible to express it as AX=B.
    Even to solve the Sylvester equation you have to diagonalize the matrices.
  5. Mar 26, 2012 #4


    User Avatar
    Science Advisor

    Try pen and paper first instead of using a computer.

    What will happen is that when you collect everything together you should get a linear system in terms of your X and some matrix that is premultiplied by it. Once you have separated the matrix from your X by specifying what that matrix is then you can do normal inversion techniques.

    You might have to write the algorithm yourself after doing a pen and paper derivation, but the idea doesn't change.

    Also when you expand out everything using algebra, I'm sure you'll find conditions for when this does not hold, possibly even as a function of A, B, and C.

    Again I urge you to do the pen and paper algebraic computation if you can't use any other known results.
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook