Requests & Ideas - page 41

Mladen Rakic
163386
Mladen Rakic  

Simba

Something like this (the upper ones of course) ? :):)

But I can not believe that crodzilla did not find it. It is only one line (just a part of line that needs to be changed) Line that goes like this :
z = phase * (180.0/pi)/360.0;[/php]
has to be changed to this
[php]z = phase * pi/2.0;
Anyway, check this one if it is what you had in mind
MrM
271
MrM  

Hi, as the person making the initial request I would suggest to go with Simba's comments if he can agree with his team.

William Snyder
9493
William Snyder  

xxxxxxxxxxxxxxxxxxxxxxxxxx

Mladen Rakic
163386
Mladen Rakic  

Simba

Getting back to the Goertzel browser :

From the pictures you posted, I assume one thing : that the "wave reconstruction" function used is cosine and not sine as crodzilla used in his Goertzel browser. The assumption is that for phase of 343 degrees (so almost a 2*Pi) the value shown by Goertzel 103 is almost 1 and that can happen only if cosine is used

So did some simplification in the browser itself (since it really had an error in wave reconstruction, even the change I told in previous post did not solve the problem, in this one the code doing it is as simple as it gets, so very little room for error) and in this one you can choose (with UseCosineForWavesparameter) whether you want to use sine or cosine for waves "reconstruction". It seems that cosine is more suitable for the purpose, but it is just a first impression.

One question (in my mind) still stays as of how previous bars phases are reconstructed : by addition of (2*Pi/period steps) to current phase or by subtraction from it in cycles, so I added one more parameter : UseAdditionForSteps- if set to true, it will add steps to current phase, if set to false, it will subtract steps from current phase

In any case, this one will reconstruct waves correctly (no illogical values any more) Changed the display so it shows angles instead of radians (it is more understandable to people) The amplitudes in this one are left as they are (there is no assumption that the amplitude should vary between -1 and 1, and frankly, I like that idea) Here is an example of cosine based 10 waves combination and it is rather interesting (in this example probably the most interesting is that the zero cross in this example happens when the "movement" is almost finished, but that is an observation that needs to be checked and re-checked) and bellow is the "single" period example on 1 hour chart using cosine too for reconstruction

regards

Mladen

SIMBA
2078
SIMBA  
mladen:
Simba Something like this (the upper ones of course) ? :):)

But I can not believe that crodzilla did not find it. It is only one line (just a part of line that needs to be changed) Line that goes like this :
z = phase * (180.0/pi)/360.0;[/php]
has to be changed to this
[php]z = phase * pi/2.0;
Anyway, check this one if it is what you had in mind

Mladen,

Crodzilla did not find it because it is way more complex than just changing that line of code,check the pic,comparison of G103 with Goertzel browser1,the 2 problems still persist,inversion and translation of the slope,even with same periodicity and phase,so,it is clear that browser1 is not the solution.

Regards

S

Files:
screen_4.gif 78 kb
SIMBA
2078
SIMBA  
Files:
screen2_1.gif 78 kb
screen3.gif 80 kb
screen4.gif 77 kb
screen5.gif 78 kb
SIMBA
2078
SIMBA  
Files:
Mladen Rakic
163386
Mladen Rakic  

:)

Saturday mornings. Good when the critical stuff is finished before

Anyway, if you took a look at the code you have noticed how simple the bloody solution was (the version 2, of course)

Here is the end-point version that uses cosine calculation and subtraction (the Use addition == false in the version 2 of the browser, in this end-point version I did not add it as parameters but they are used by default (to add some simplicity in this case in order to avoid to much confusion why this and not that. and so on ...) and to use it in comparison
Will make the version that shows all the data of the composite wave since it seems to be the most interesting thing we gained with the correct indicator - at last we can experiment with what Millard was talking) so by all mean I am going to work on it (at least you know for how long Millard is an interesting "theme" for me )

regards

Mladen

SIMBA
2078
SIMBA  
MrM
271
MrM  

are there any other ideas we can try here?