Adaptive digital filters - page 31

[Alexey Subbotin]

He did. Proof?

I gave mine.

[[Deleted]]

to alsu "And what here contradicts my statements?"

That clever man, BQQ, who posted there quite a few original filters, wrote you there: "SMA(N+1) lags by N/2 samples" and "I tell the truth. Check it with your fingers or call the grpdelay function in MATLAB, if you feel more comfortable with it". - I join in. You could not give any proof, because there is none. You have cited your nonsense, nothing more.

[Alexey Subbotin]

mikfor:

to alsu "And what here contradicts my statements?"

That clever man, BQQ, who posted there quite a few original filters, wrote you there: "SMA(N+1) lags by N/2 samples" and "I tell the truth. Check it with your fingers or call the grpdelay function in MATLAB, if you feel more comfortable with it". - I join.

The author of this statement forgets that for a syssoid with a period of 2 or 3 bars the lag of N/2 bars, if N is say 20, includes several whole periods, which normal people are used to ignoring.

[Alexey Subbotin]

it's called "thinking". And what's in there is "putting numbers into formulas".

[[Deleted]]

I'm tired of listening to your nonsense. Build a "sine wave" with a period of 2 bars. Look at this broken line, be glad you know it's a sine wave, and apply SMA(20) to it. There will be a straight line, I suppose. in which you can take into account how many "whole periods" you want, but better not here.

[Alexey Subbotin]

By the way, grpdelay gives a group packet delay, which is irrelevant to the topic at hand

[Alexey Subbotin]

mikfor:
I'm tired of listening to your nonsense. Build a "sine wave" with a period of 2 bars. Look at that broken line, be glad you know it's a sine wave, and apply the SMA(20) to it. It will be a straight line, one would assume.

blah blah blah. No argument, no education.

[[Deleted]]

"blah blah blah. No argument, no education."

Dummy, how about I build you two graphs with VERY DIFFERENT input frequencies, and show you that the lag is the same on your fingers (picture)? Will you calm down?

[Alexey Subbotin]

with your hands on paper? Go ahead.

I'll stop by in the morning and have a look.

[[Deleted]]

here, model 1, and SMA(100) and model 2, and SMA(100).

The frequencies differ by a factor of 2.5. I think that's enough. Showed 500 bars of model sine waves. You, as an expert in determining lag "by eye" should see that it is the same. Look at the interval between 100 and 200 bars on the abscissa axis. And I will even tell you what it is equal to. 49.5 bars.