adrianschlatter/tanuna

Python tools to work with dynamic systems.

tanuna's Issues

CT_System.phaseportrait(*args)

Return n-dim mesh grid of phase portrait compatible with streamplot (at least for n=2). args is a tuple of intervals giving the ranges of the internal states.

=> use this in the ml laser example

tf of system with zero DC-gain

High-pass systems have a zero at s=0. CT_LTI_System.tf() normalizes the nominator polynomials based on the DC-gain => NaN!

Create LTI from zpk

Add classmethod that creates CT_LTI from zpk.

Observer design

Add method for observer design. Probably there is more than one method => this issue addresses the most common one.

=> extend ml laser example to estimate state from observation of output power.

CT_LTI_System.links => .shape

rename .links to .shape and return (nout, nin) instead of (nin, nout) to be consistent with matrix formalism.

Feedback

Add support to connect systems in feedback configuration.

Unittest for CT_LTI_System.tf

Write one and pass it, especially for MIMO.

CT_System.__connect__

Support connection of continuous-time systems.

Powers

Support raising systems to powers. This is useful, e.g., to make a 4th-order low pass from a 1st-order low pass.

Create LTI from tf

Add classmethod to initialize CT_LTI from transfer function.

Partially connecting matrices

Currently, we cannot connect a system to only one output of a matrix. Example: Connect a SISO to one leg of a splitter (np.matrix([[1], [1]])).

=> partial connections for matrices could solve that problem.

CT_System.steadystates(t=0)

Returns list of stable and unstable steady-states at time t.

Fix zpk

zpk currently assumes 'k' to be DC-gain. This is not the usual definition and leads to problems in system with zero DC-gain. What zpk should mean is:

ODE Solver

Make CT_System solvable.

=> use this in ml laser example

pyproject.toml

pyproject.tomlize

System comparison

add support of system comparison to CT_LTI_System. Doctest currently fails because this is not yet implemented.

lqr

Controller design by lqr method.

=> extend ml laser example by adding constraints on pump power excursions.

Add Response-Plots

Add plot of step- and impulse-responses, Bode- and Nyquist-plots in doc.

Ackermann

Implement Ackermann's algorithm for pole placement.

K = ackermann(A, B, poles)

=> use this in ml laser example

Wr, Wo

Add correct output for G.Wr and G.Wo in doc

Step-response plot ml laser

... Aus missing. Build problem on RTD?

CT_LTI operations

Add multiplication and addition.

Make CT_LTI_System.tf callable

Currently, CT_LTI_System.tf returns the transfer-function representation as follows:

(matrix([[poly1d([ 5.48342176e+12])]], dtype=object),
poly1d([ 1.00000000e+00, 2.94538978e+05, 8.69700924e+12]))

As the transfer function is - as the name says - a function, it would be very convenient to make this callable. CT_LTI_System.tf could, e.g., return an object of class TransferFunction which has a nice repr and is additionally callable.

CT_LTI.__radd__

Example:

G = LowPass(10)
np.matrix(1) + G

Fails on second line because - apparently - np.matrix.add claims to know how to add a CT_LTI_System. It tries to do so by adding G to each (float-) element of matrix.

Same goes for rsub(...)

Document system operations

Add Chapter on *, +,/,-,|, connect, feedback.

High-Pass

Add 1st-order high-pass to library.

CT_System.linearize(x0)

Approximate CT_System by CT_LTI_System near x0. x0 is given by user.

=> use this in ml laser example

Diving in plot

... is too wide. At least when viewed on the iPad.

add github workflow for testing