--- In [email protected], "Ara Kaloustian" <[EMAIL PROTECTED]> wrote:
>
> You might consider using the built in FFT to get the dominant cycle.
It is 
> much faster than Laguerre or other similar calcs.
> 
> I have not compared the two for accuracy. I am sure they are both error 
> prone based on how pure a cycle is at any given time.

I tried that 1st and while calculating a single dominant cycle is
fast,  there are several problems I had so I used Laguerre because it
was simple and it improves the performance of the indicators.

The problems are:

1. To get an array of dominant periods, I need to use a loop with the
fft re-calculated every time through and that will slow things down
some. I tried doing this but so far have failed (not very good at
programming).

2. The difference in dominant cycle between Laguerre and fft is  not
consistent and varies widely 

3. Just choosing the cycle with the largest amplitude is unacceptable.
It jumps around instead of varying smoothly like the Laguerre method. 

I have not been able to think of a way to reconcile the differences
between fft and Laguerre. 

It's been a very long time since I learned about DSP and I have
forgotten more than I remember. I will need to do a lot of research to
learn what I need to know to resolve the above problems. 

If I can get satisfactory results with the Laguerre approximation it
is the best solution.

Bill

Reply via email to