[SciPy-User] [SciPy-user] least-square filter design

[Tom K.]

Neal Becker wrote:
> 
> Anyone have code for least square (minimum mean square error) FIR filter 
> design?
> 

Could you be a little more specific?  scipy.signal.firwin almost designs a
least square low pass FIR filter if you use a rectangular window (I say
almost because like other packages the filter's response is normalized to
unity at DC so technically it is not least squares although the difference
is slight and decreases with increasing filter order).

Do you need a transition band?  What type of FIR filter: lowpass, highpass,
bandpass, bandstop, or multiband?  Are discrete samples OK, or do you need a
continuous band (or set of bands)?  Which type of filter - is symmetric OK,
or do you need antisymmetric?

Or, are you talking about an adaptive filter?

-- 
View this message in context: http://old.nabble.com/least-square-filter-design-tp26083443p26139404.html
Sent from the Scipy-User mailing list archive at Nabble.com.