This PR adds

• MIMO support to the markov function and
• it uses the TimeResponseData Type.

T = np.linspace(0, 10, 100)
U = np.ones((1, 100))
response = ct.forced_response(ct.tf([1], [1, 0.5], True), T, U)
H = ct.markov(response, 3)

This breaks old user code!
Does not break old user code!

T = np.linspace(0, 10, 100)
U = np.ones((1, 100))
T, Y = ct.forced_response(ct.tf([1], [1, 0.5], True), T, U)
H = ct.markov(Y, U, 3, transpose=False)

coverage: 94.519% (-0.01%) from 94.533%
when pulling ec38f2e on KybernetikJo:improve_markove_mimo
into 4acc78b on python-control:main.

coverage: 94.519% (-0.01%) from 94.533%
when pulling 4c1f4ca on KybernetikJo:improve_markove_mimo
into 4acc78b on python-control:main.

coverage: 94.519% (-0.01%) from 94.533%
when pulling 446a161 on KybernetikJo:improve_markove_mimo
into 4acc78b on python-control:main.

Rather than breaking old code, it would be nice to do this in a way that preserved prior functionality. So use *args processing to allow both of the following to work:

H = ct.markov(response, 3)
H = ct.markov(Y, U, 3)

It also seems a bit odd (to me) to have the return type be a TimeResponseData object. It true that the Markov parameters are the impulse response, but are people going to be plotting that impulse response as the primary way they interact with the output of the markov command?

Rather than breaking old code, it would be nice to do this in a way that preserved prior functionality. So use *args processing to allow both of the following to work:

H = ct.markov(response, 3)
H = ct.markov(Y, U, 3)

I will update this one.

It also seems a bit odd (to me) to have the return type be a TimeResponseData object. It true that the Markov parameters are the impulse response, but are people going to be plotting that impulse response as the primary way they interact with the output of the markov command?

I do not have strong feelings about that. The idea was to simplify the joint use of markov and era:

impulse_response = ct.markov(response, m=10)
sysd = ct.era(impulse_response, r=4)

Direct access of Markov parameters:

_, H = ct.markov(response, m=10)

I can change the return value to an array. I think it is also better for the old API.
I guess I should also keep the old API for era.

H = ct.markov(Y, U, m=10)
sysd = ct.era(H, r=4)

coverage: 94.538% (+0.005%) from 94.533%
when pulling 317d261 on KybernetikJo:improve_markove_mimo
into 4acc78b on python-control:main.

coverage: 94.516% (-0.02%) from 94.533%
when pulling 2b1cd21 on KybernetikJo:improve_markove_mimo
into 4acc78b on python-control:main.

coverage: 94.516% (-0.02%) from 94.533%
when pulling cee59d1 on KybernetikJo:improve_markove_mimo
into 4acc78b on python-control:main.

coverage: 94.669% (+0.04%) from 94.629%
when pulling e61ac53 on KybernetikJo:improve_markove_mimo
into da64e0e on python-control:main.

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.