Adaptive digital filters - page 19

 
LeoV >> :

EMA is EMA.

Auntie Ema is not an adaptive filter, the exponent is just a dumb static function.

in the EMA+neuronet system, my understanding is that adaptation takes place at the post-filtering level

 
I decided not to create a new thread (although I could) and put my question here. Or rather not even a question, but a whole task for clever people. The task itself is in the archive. It is addressed to those who understand DSP. Actually from my point of view on the one hand it is a paradox, because under the conditions of the problem we KNOW the future, but cannot do anything).
Files:
zkkpke.rar  255 kb
 

There is no paradox.

You do not take into account the Gibbs effect (chattering), which is the nature of the finite window (no points on the right) of averaging when backtracking on the right edge of BP. This leads to the inevitable phase delay (only on the last counts on the right) and, as a consequence, the appearance of an imaginary paradox - we look in one place and count in another.

 
Farnsworth >>:

информации довольно много на эту тему, например, тут: http://prodav.narod.ru/dsp/index.html (раздел "адаптивная фильтрация цифровых данных")


Спросите так же Prival-а, он кажется, единственный специалист по ЦОС из нас, думаю, поможет в теории.

PS: что такое JMA?


Oh, yeah. :) Go to "my website" and read until you are enlightened.


Praval, alas, is (or may not be) in the grip of dangerous delusions. :) About Fourier, for example, and everything else connected with DSP. :)

To my regret I even once had (not by my own will :) ) to demonstrate it.


Read there Mr. Deburg's thesis. :) To understand where it is possible to apply something and where not.


To me in general it is amazing - that not already all that I told here it flew past the ears. :) Well, gentlemen "Teachers" so low faith in their abilities not to even try to find the truth and earn on this in the market. We keep saying the same thing. I'll say it again for the 200th time, damn it - it's very easy to make money on Forex. There's nothing at all understandable here. Why are you screwing around with some virtual problems?

 
SProgrammer >>:

Читайте там дисертацию господина Дебурга. :) чтобы понять то куда можно что-то применять, а куда нет.

.......

Повторю, уже в 200-раз, блин - заработать на форексе очень просто. Тут нет ну ВООБЩЕ ничего не понятного.

This dissertation ?(http://sepwww.stanford.edu/data/media/public/oldreports/sep06//)

How interesting! Can you tell me more about it? With references? You don't have to tell me everything, I have a degree from a TU (vocational school).

Thank you in advance.

 

SProgrammer's mad at me for something. No answer.

 
AlexEro >>:

Этот диссер ? (http://sepwww.stanford.edu/data/media/public/oldreports/sep06//)

Ой как интересно! А поподробнее можно? Со ссылками? Всё можно не рассказывать, у меня есть диплом ТУ (ПТУ).

Заранее спасибо.

Well, I think you've figured it out :)

I've already told you the "not all" part. :)

 
Prival >>:

Вот шаблон индикатора, как прикрутить сюда расчеты по заданному периоду (1, 5, 15, 30 и т.д). Подскажите пожалуйста.

Replace Bars with iBars(), Time[] with iTimes(). And get access to other TFs

 
Neutron >>:

Парадокса нет.

Вы при обратном проходе на правом краю ВР не учитывате эффект Гибса (болтанка), имеющий своей природой конечное окно (справа точек нет) усреднения. Это приводит к возникновению неизбежной фазовой задржки (только на послених отсчётах справа) и, как следствие, возникновение мнимого парадокса - смотрим в одном месте, а считаем в другом.


It was a question of if we already know the final value of the filter. I.e. it does not change and there is no chattering. We know the long filter, but we don't know the short one, and we could get the short one if we knew the long one, but I knew about chattering myself)))))) there is an assumption in the problem that we have no chattering on the long one, as it were.
 
So has anyone managed to finally implement an adaptive digital filter ( or make a kind of oscillator : this was discussed here http://forum.viac.ru/viewtopic.php?t=880 and here http://forum.viac.ru/viewtopic.php?t=1034&postdays=0&postorder=asc&start=500 above first post by user magos ) or a file to swap filter coefficients ???