Re: digital differentiator using remez

[Michael Creel]

On 2/8/07, Doug Stewart <address@hidden> wrote:

for interest sake here is Matlab's result

EDU» b = remez(4, [0 0.75], [0 0.75*pi],  'differentiator')

b =   -0.2731    0.9430         0   -0.9430    0.2731


and
b = remez(4, [0 0.75 ], [0 0.75*pi],[70],  'differentiator')

b =   -0.2731    0.9430         0   -0.9430    0.2731

Octave 2.1.42 gives

>> b = remez(4, [0 0.75], [0 0.75*pi], [70 ], 'differentiator')
b =

   0.29538
  -0.94339
   0.00000
   0.94339
  -0.29538


Doug Stewart





Riccardo Corradini wrote:
>
> Hi Ron,
> I met a similar error runing a bfgs routine on a likelihood
> econometric function for spatial models
> error: invalid conversion from matrix to real vector ( i used growth
> rates with lots of zeros)
> Try to run your function with some values significantly different from
> zero, and you will probably get rid of that errors.
> I suggest to compile the last version of octave2.9.9 from cvs.
> Just type sudo apt-get build-dep octave2.9, then
> sudo apt-get install checkinstall
> After downloading cvs octave see instructions on www.octave.org
> type ./configure --prefix=/usr
> make
> sudo checkinstall
> This will create the new kubuntu package octavew2.9.9.cvs and install it
> Cheers
> Riccardo
> */Ron Crummett <address@hidden>/* ha scritto:
>
>     Hi -
>
>     I am trying to design a 4th order wideband differentiator using the
>     remez function. I type in
>     >>b = remez(4, [0 0.75], [0 0.75*pi], [], 'differentiator');
>
>     And I get this response:
>     error: invalid conversion from matrix to real vector
>     error: remez: need one weight for each band [=length(band)/2]
>     error: evaluating assignment _expression_ near line 22, column 3
>
>     My only guess is that I need something in the weighting entry, so I
>     change it to
>     >> b = remez(4, [0 0.75], [0 0.75*pi], 1, 'differentiator');
>
>     Which gives me the message:
>     error: remez: griddensity is too low; must be greater than 16
>     error: evaluating assignment _expression_ near line 22, column 3
>
>     A third attempt:
>     >> b = remez(4, [0 0.75], [0 0.75*pi], 1, 'differentiator', 64);
>     error: remez: incorrect argument list
>     error: evaluating assignment _expression_ near line 22, column 3
>
>     At this point I am out of ideas of what I need to change. Has anyone
>     had any success with this? I am running 2.9.6 on Kubuntu breezy.
>     Thanks.
>
>     -Ron
>
>     ___
>

_______________________________________________
Help-octave mailing list
address@hidden
https://www.cae.wisc.edu/mailman/listinfo/help-octave

 

Riccardo - did the problem with bfgsmin go away when you used newer versions of Octave?
Thanks, Michael