Help with Fourier - page 15

New comment
 

Here, used to build a wrapper of FFT procedure from alglib. The sample size is arbitrary, enjoy)).

Read here.

Files:
fft.rar  249 kb
 
Sors.
Files:
fftmain.rar  1 kb
 
We are already enjoying ourselves. Algorithm from alglib has been rewritten long ago and is in codebase - https://www.mql5.com/ru/code/9696. Read its description about the sample size - it should be a degree of two, not arbitrary.
 
Integer:
Now we are having fun. The algorithm from the alglib has long been rewritten and is in the codebase - https://www.mql5.com/ru/code/9696. Read its description about sample size - it should be degree of two, not arbitrary.

This is a different algorithm. The one by klot is a classical fast real series transformation (although it can be done for any n, not just the power of two). I have made a wrapper complex series transformation for any n. So as not to argue, I quote from the original source.

1-dimensional complex FFT.Array size N may be arbitrary number (composite or prime).  Composite N's are handled with cache-oblivious variation of a Cooley-Tukey algorithm. Small prime-factors are transformed using hard coded codelets (similar to FFTW codelets, but without low-level optimization), large prime-factors are handled with Bluestein's algorithm.

Ibid. Remarks on speed

Fastests transforms are for smooth N's (prime factors are 2, 3, 5 only), most fast for powers of 2. When N have prime factors larger than these, but orders of magnitude smaller than N, computations will be about 4 times slower than for nearby highly composite N's. When N itself is prime, speed will be 6 times lower. The algorithm has O(N*logN) complexity for any N (composite or prime).


 
alsu:
Of course you can, no questions asked)

Can you tweak this indicator

https://www.mql5.com/ru/code/7359

so it doesn't overdraw?
 
There is such an indicator in codebase - it is called extrapolator, so it is interesting to follow its re-drawing from bar to bar + if we take into account changes of .... parameters. But there are problems with abrupt changes of the forecasting part, i.e. it seems that we can do analysis by changes, but abrupt changes spoil everything, it looks like the line goes to 45 degrees upwards then bang on -30 degrees down and back to 45 upwards, so we need to get rid of this jump ....
 

Has anyone tried applying Fourier decomposition not directly to price but to the change in rainbow slices from mash-ups

 

Has anyone tried applying Fourier decomposition not directly to price, but to the change in rainbow slices from

and the slices are not vertical

 

Valera, here's a problem for you.

Have you tried adding 1+3 instead of 2+2? Try it!

Your questions are really something! It's obvious you're not at all on the subject.

 
trol222:

and the slices are like that - not vertical.

I was in a hurry to draw it... These slices should be curved lines and run counter to each other from 2 x ends of the sample.

I will correct the drawing in that post.

1...8910111213141516171819
New comment