H-infinity mixed-synthesis #151

roryyorke · 2017-07-01T20:10:16Z

This adds methods "augw" and "mixsyn", with interfaces similar to the Matlab and Octave functions of the same name. They are for so-called mixed-sensitivity design, in which the sensitivity, complementary sensitivity, and "input sensitivity" (K*S, or reference to input mapping) can be weighted and stacked for H-infinity synthesis.

As part of this, I've fixed what I think is a bug, namely that empty StateSpace objects could not be created. Such empty objects are convenient in the augw implementation.

I've added two examples for mixsyn, one SISO and one MIMO.

augw is the major part of the new code. It may be that there is some better way to implement it, but I have tried to write it in a way that no uncontrollable or unobservable modes are introduced, and so that is still (somewhat) clear.

coveralls · 2017-07-01T20:16:37Z

Coverage increased (+0.5%) to 78.162% when pulling b41fe06 on roryyorke:rory/augw into 05d6bc5 on python-control:master.

murrayrm · 2018-01-02T16:27:01Z

@roryyorke There's a small conflict with master that it would be good to fix. Will try to do a more thorough review in the coming days...

roryyorke · 2018-01-02T16:46:52Z

Will have a look at the conflict.

New functions augw and mixsyn modelled after Matlab and Octave's function of the same names. python-control doesn't currently support input group naming, so that is not implemented.

Both examples taken from Skogestad and Postlethwaite's Multivariable Feedback Control.

coveralls · 2018-01-02T17:25:49Z

Coverage increased (+0.4%) to 78.954% when pulling 7b883ef on roryyorke:rory/augw into 33bebc1 on python-control:master.

murrayrm · 2018-01-04T05:36:24Z

control/robust.py

+
+
+def _size_as_needed(w,wname,n):
+    """_size_as_needed(w,wname,n) -> w2


Nonstandard docstring. I don't think we use this type of docstring (fcn(args) -> output) anyplace else. In addition, parameters, returns, etc are documented in a different format. Would prefer to use a consistent style, as done in hinfsyn in this file.

Will fix here and elsewhere (augw, _matrix, ...). I see I also used this in xferfcn.py in 310f580 in _dc_gain_cont.

murrayrm · 2018-01-04T05:37:06Z

control/robust.py

+
+
+def augw(g,w1=None,w2=None,w3=None):
+    """augw(g,w1=None,w2=None,w3=None) -> p


Nonstandard docstring.

murrayrm · 2018-01-04T05:41:53Z

control/robust.py

+    cl: closed system mapping evaluation inputs to evaluation outputs; if p is the augmented plant, with
+        [z] = [p11 p12] [w], then cl is the system from w->z with u=-k*y.  StateSpace object.
+        [y]   [p21   g] [u]
+    info: namedtuple with fields, in order,


This is the first use of a namedtuple in python-control. Is there a more standard way to return this sort of information in numpy or scipy? If so, perhaps we use that format?

namedtuple has been around since Python 2.6, so it's hardly non-standard ;). I couldn't find namedtuple use in numpy, but it is used in scipy (stats/stats.py, for example).

I can change this to a plain tuple; if we do keep the namedtuple, I'd like to remove the leading underscore from the name.

murrayrm · 2018-01-04T05:46:57Z

control/statesp.py

@@ -54,7 +54,7 @@
 import math
 import numpy as np
 from numpy import all, angle, any, array, asarray, concatenate, cos, delete, \
-    dot, empty, exp, eye, matrix, ones, poly, poly1d, roots, shape, sin, \
+    dot, empty, exp, eye, matrix, ones, pi, poly, poly1d, roots, shape, sin, \


We should avoid using numpy.pi since this can generate errors when numpy is incorporated as a mock function. See PR 171, commit b881b7. Replace with math.pi.

unqualified pi no longer used following your aa6e0df anyway. Will remove.

murrayrm · 2018-01-04T05:49:32Z

control/statesp.py

+def _matrix(a):
+    """_matrix(a) -> numpy.matrix
+    a - passed to numpy.matrix
+    Wrapper around numpy.matrix; unlike that function,  _matrix([]) will be 0x0


Not clear to my why this is useful? Can you say a bit here (or in a comment in the code)?

I'll expand the commentary; it's just a helper to make it easier to get 0x0 matrices, which are needed for empty state space systems.

control/robust.py: - doc-strings conform to python-control conventions - use plain tuple instead of namedtuple for auxiliary output in mixsyn control/statesp.py: - don't import pi from numpy - doc-string for _matrix expanded, conforms to conventions control/tests/robust_test.py - change test to accept plain tuple

roryyorke · 2018-01-05T14:24:05Z

I thought I'd check the docstrings via sphinx, but when I run make html in the doc sub-directory, I get

  File "/home/rory/miniconda3/envs/control-python27/lib/python2.7/re.py", line 155, in sub
    return _compile(pattern, flags).sub(repl, string, count)
TypeError: expected string or buffer

I get the same error on master.

I see doc-requirements.txt references Sphinx from github, but that is somewhat older (log talks about Sphinx 1.3.1, I've got 1.6 installed from conda).

This is my conda environment (2.7-based, and AFAIK up-to-date; don't know why I don't have Scipy 1.0):

alabaster                 0.7.10                   py27_0  
babel                     2.5.0                    py27_0  
cairo                     1.14.8                        0  
certifi                   2016.2.28                py27_0  
cycler                    0.10.0                   py27_0  
dbus                      1.10.20                       0  
docutils                  0.14                     py27_0  
expat                     2.1.0                         0  
fontconfig                2.12.1                        3  
freetype                  2.5.5                         2  
funcsigs                  1.0.2                    py27_0  
functools32               3.2.3.2                  py27_0  
glib                      2.50.2                        1  
gst-plugins-base          1.8.0                         0  
gstreamer                 1.8.0                         0  
icu                       54.1                          0  
imagesize                 0.7.1                    py27_0  
jinja2                    2.9.6                    py27_0  
jpeg                      9b                            0  
libffi                    3.2.1                         1  
libgcc                    5.2.0                         0  
libgfortran               3.0.0                         1  
libiconv                  1.14                          0  
libpng                    1.6.30                        1  
libxcb                    1.12                          1  
libxml2                   2.9.4                         0  
markupsafe                1.0                      py27_0  
matplotlib                2.0.2               np113py27_0  
mkl                       2017.0.3                      0  
mock                      2.0.0                    py27_0  
nose                      1.3.7                    py27_1  
numpy                     1.13.1                   py27_0  
numpydoc                  0.7.0                    py27_0  
openssl                   1.0.2l                        0  
pbr                       1.10.0                   py27_0  
pcre                      8.39                          1  
pip                       9.0.1                    py27_1  
pixman                    0.34.0                        0  
pycairo                   1.10.0                   py27_0  
pygments                  2.2.0                    py27_0  
pyparsing                 2.2.0                    py27_0  
pyqt                      5.6.0                    py27_2  
python                    2.7.13                        0  
python-dateutil           2.6.1                    py27_0  
pytz                      2017.2                   py27_0  
qt                        5.6.2                         5  
readline                  6.2                           2  
requests                  2.14.2                   py27_0  
scipy                     0.19.1              np113py27_0  
setuptools                36.4.0                   py27_1  
sip                       4.18                     py27_0  
six                       1.10.0                   py27_0  
snowballstemmer           1.2.1                    py27_0  
sphinx                    1.6.3                    py27_0  
sphinxcontrib             1.0                      py27_0  
sphinxcontrib-websupport  1.0.1                    py27_0  
sqlite                    3.13.0                        0  
subprocess32              3.2.7                    py27_0  
tk                        8.5.18                        0  
typing                    3.6.2                    py27_0  
wheel                     0.29.0                   py27_0  
zlib                      1.2.11                        0

coveralls · 2018-01-05T14:33:34Z

Coverage decreased (-0.5%) to 78.051% when pulling e24c7de on roryyorke:rory/augw into 33bebc1 on python-control:master.

coveralls · 2018-01-05T14:33:35Z

Coverage decreased (-0.5%) to 78.051% when pulling e24c7de on roryyorke:rory/augw into 33bebc1 on python-control:master.

murrayrm · 2018-01-05T17:07:37Z

@roryyorke I had the same problem with sphinx on my local machine. For some reason it seems to work on readthedocs.org. I think the issue may be with the way that version string is being reported (somehow comes in as a float). We should probably chase this down, but perhaps via a separate issue.

I'll try to make sure things look OK on readthedocs when I get a chance (probably tonight or this weekend).

murrayrm · 2018-01-06T18:36:29Z

Sent @roryyorke a PR that creates the documentation on readthedocs.

coveralls · 2018-01-06T19:03:56Z

Coverage decreased (-0.4%) to 78.073% when pulling 36e0b0f on roryyorke:rory/augw into 33bebc1 on python-control:master.

murrayrm self-requested a review December 27, 2017 16:35

murrayrm added this to the 0.8.0 milestone Dec 27, 2017

murrayrm self-assigned this Jan 2, 2018

roryyorke added 3 commits January 2, 2018 19:12

BUG: allow empty (no input or output) StateSpace objects

77d0ccd

Augmented plant and mixed-sythensis H-infinity design

78f0c7f

New functions augw and mixsyn modelled after Matlab and Octave's function of the same names. python-control doesn't currently support input group naming, so that is not implemented.

SISO and MIMO mixed-sensitivity synthesis examples

7b883ef

Both examples taken from Skogestad and Postlethwaite's Multivariable Feedback Control.

roryyorke force-pushed the rory/augw branch from b41fe06 to 7b883ef Compare January 2, 2018 17:13

murrayrm reviewed Jan 4, 2018

View reviewed changes

murrayrm added 2 commits January 6, 2018 20:47

DOC: add mixsyn and augw to documentation

7d70fb4

DOC: updated mixsyn docstring to use standard 'see also' format

36e0b0f

roryyorke mentioned this pull request Jan 6, 2018

Add mixsyn and augw to documentation roryyorke/python-control#1

Closed

murrayrm merged commit 1801c9a into python-control:master Jan 6, 2018

roryyorke deleted the rory/augw branch January 9, 2018 19:13

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

H-infinity mixed-synthesis #151

H-infinity mixed-synthesis #151

roryyorke commented Jul 1, 2017

coveralls commented Jul 1, 2017 •

edited

murrayrm commented Jan 2, 2018

roryyorke commented Jan 2, 2018

coveralls commented Jan 2, 2018 •

edited

murrayrm Jan 4, 2018

roryyorke Jan 5, 2018

murrayrm Jan 4, 2018

murrayrm Jan 4, 2018

roryyorke Jan 5, 2018

murrayrm Jan 4, 2018

roryyorke Jan 5, 2018

murrayrm Jan 4, 2018

roryyorke Jan 5, 2018

roryyorke commented Jan 5, 2018

coveralls commented Jan 5, 2018

coveralls commented Jan 5, 2018 •

edited

murrayrm commented Jan 5, 2018

murrayrm commented Jan 6, 2018 •

edited

coveralls commented Jan 6, 2018 •

edited



		def _size_as_needed(w,wname,n):
		"""_size_as_needed(w,wname,n) -> w2



		def augw(g,w1=None,w2=None,w3=None):
		"""augw(g,w1=None,w2=None,w3=None) -> p

H-infinity mixed-synthesis #151

H-infinity mixed-synthesis #151

Conversation

roryyorke commented Jul 1, 2017

coveralls commented Jul 1, 2017 • edited

murrayrm commented Jan 2, 2018

roryyorke commented Jan 2, 2018

coveralls commented Jan 2, 2018 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

roryyorke commented Jan 5, 2018

coveralls commented Jan 5, 2018

coveralls commented Jan 5, 2018 • edited

murrayrm commented Jan 5, 2018

murrayrm commented Jan 6, 2018 • edited

coveralls commented Jan 6, 2018 • edited

coveralls commented Jul 1, 2017 •

edited

coveralls commented Jan 2, 2018 •

edited

coveralls commented Jan 5, 2018 •

edited

murrayrm commented Jan 6, 2018 •

edited

coveralls commented Jan 6, 2018 •

edited