As pointed out by @pmli in #27, instead of [computing matrix inverses](https://github.com/python-control/python-control/blob/0b26e062a6cd82da8a2420ebc3957e5267e6e116/control/statesp.py#L458), better to use `np.linalg.solve` when needed.
As pointed out by @pmli in #27, instead of computing matrix inverses, better to use
np.linalg.solvewhen needed.