Description

vincentchoqueuse

opened on Aug 17, 2021

Issue body actions

Hi,

This is my first contribution on GitHub ;)

Problem

The method damp() seems to return a wrong result for discrete systems with negative poles

import control as ctl
H = ctl.tf([1],[1,0.2],1)
H.damp()

Result : RuntimeWarning: invalid value encountered in log

Investigation

The problem seems to come from the type of the poles variable in the dampmethod of the LTI object

if isdtime(self, strict=True):
     splane_poles = np.log(poles)/self.dt
else:
      splane_poles = poles
wn = absolute(splane_poles)
Z = -real(splane_poles)/wn
return wn, Z, poles

Solution

The poles variable must be cast to a numpy complex type as follows

if isdtime(self, strict=True)
      splane_poles = np.log(poles.astype(complex))/self.dt
...

Activity

1. bnavigator commented on Aug 17, 2021

   

   bnavigator

   on Aug 17, 2021

   Contributor

   More actions

   Hi,

   thanks for catching this!

   This is my first contribution on GitHub ;)

   Awesome! Would you try to make the next step and create a Pull Request with your proposed changes?

   Bonus points for adding your example to test_damp() (using pytest.mark.parametrize()) :

   python-control/control/tests/lti_test.py

   Lines 54 to 71 in 6b7a9c7

   	def test_damp(self):
   	# Test the continuous time case.
   	zeta = 0.1
   	wn = 42
   	p = -wn * zeta + 1j * wn * np.sqrt(1 - zeta**2)
   	sys = tf(1, [1, 2 * zeta * wn, wn**2])
   	expected = ([wn, wn], [zeta, zeta], [p, p.conjugate()])
   	np.testing.assert_allclose(sys.damp(), expected)
   	np.testing.assert_allclose(damp(sys), expected)
   	# Also test the discrete time case.
   	dt = 0.001
   	sys_dt = c2d(sys, dt, method='matched')
   	p_zplane = np.exp(p*dt)
   	expected_dt = ([wn, wn], [zeta, zeta],
   	[p_zplane, p_zplane.conjugate()])
   	np.testing.assert_almost_equal(sys_dt.damp(), expected_dt)
   	np.testing.assert_almost_equal(damp(sys_dt), expected_dt)
2. bnavigator

   linked a pull request that will close this issuefix damp method for discrete time systems with a negative real-valued pole #647on Aug 17, 2021
3. bnavigator

   added

   bug

   on Aug 17, 2021
4. bnavigator

   closed this as completedin #647on Aug 19, 2021

Issue actions