Re: Request for Matlab output

[Lukas Reichlin]

> On 30.10.2015, at 18:04, Steph Bredenhann <address@hidden> wrote:
> 
> 
> On 29/10/2015 21:13, Lukas Reichlin wrote:
>> On 29.10.2015, at 18:50, Lukas Reichlin <address@hidden> wrote:
>>>> On 29.10.2015, at 12:36, Lukas Reichlin <address@hidden> wrote:
>>>> 
>>>> Dear Octave community,
>>>> 
>>>> In order to improve error handling in the control package, I would like to 
>>>> know what Matlab’s output/error message looks like for each of the 
>>>> following three lines of code:
>>>> 
>>>> dss (1, 2, 3, 4, 0)
>>>> 
>>>> feedback (tf (1), tf (-1))
>>>> 
>>>> feedback (ss (1), ss (-1))
>>>> 
>>>> Could someone with access to Matlab and its Control System Toolbox please 
>>>> send me the resulting text output?
>>>> 
>>>> Thanks in advance,
>>>> Lukas
>>>> 
>>>> 
>>> Gentlemen, thank you very much for your help! I’ve just pushed a file 
>>> containing my conclusions:
>>> 
>>> http://sourceforge.net/p/octave/control/ci/default/tree/devel/algebraic_loops.m
>>> 
>>> Best regards,
>>> Lukas
>> May I ask you for another test in Matlab?
>> 
>> 
>> e = zeros (2, 2);
>> a = [-1, 1; 1, -1];
>> b = [1; 0];
>> c = [1, 0];
>> d = 0;
>> sys = dss (a, b, c, d, e)
>> G = tf (sys)
>> H = zpk (sys)
>> 
>> 
>> I think that descriptor systems with all-zero E matrix and non-invertible A 
>> matrix are invalid. Now I wonder whether Matlab accepts those systems at all 
>> or only returns NaN’s later on.
>> 
>> Best regards,
>> Lukas
>> 
>> 
>> 
>> 
> Ubuntu 14.04 x64 with Matlab 2015b
> 
> 
> sys =
> 
>  a =
>       x1  x2
>   x1  -1   1
>   x2   1  -1
> 
>  b =
>       u1
>   x1   1
>   x2   0
> 
>  c =
>       x1  x2
>   y1   1   0
> 
>  d =
>       u1
>   y1   0
> 
>  e =
>       x1  x2
>   x1   0   0
>   x2   0   0
> 
> Continuous-time state-space model.
> 
> Warning: The frequency response has poor relative accuracy. This may be 
> because the
> response is nearly zero or infinite at all frequencies, or because the 
> state-space
> realization is ill conditioned. Use the "prescale" command to investigate 
> further.
> > In ltipack.ssdata/iodynamics
>  In ltipack.ssdata/zpk (line 23)
>  In ltipack.ssdata/tf (line 20)
>  In lti/tf_ (line 283)
>  In DynamicSystem/tf (line 90)
>  In untitled (line 11)
> 
> G =
> 
>  -6369051672525772
> 
> Static gain.
> 
> Warning: The frequency response has poor relative accuracy. This may be 
> because the
> response is nearly zero or infinite at all frequencies, or because the 
> state-space
> realization is ill conditioned. Use the "prescale" command to investigate 
> further.
> > In ltipack.ssdata/iodynamics
>  In ltipack.ssdata/zpk (line 23)
>  In lti/zpk_ (line 295)
>  In DynamicSystem/zpk (line 80)
>  In untitled (line 12)
> 
> H =
> 
>  -6.3691e+15
> 
> Static gain.
> 
> >>
> 

Thank you very much! The result confirms my resolution to refuse such systems 
(matrix E zero and matrix A not invertible) in control-3.0.1. The reason for 
this refusal is the same as for refusing zero denominators in transfer 
functions: these systems are simply invalid.

Best regards,
Lukas