Generation of uniformly distributed random numbers (0,1) - page 3

New comment
 

More specifically, what does "bad" mean? Are there any obvious violations of the uniformity of the distribution?

In the formula AB = MathRand() + MathRand()*32768 (I made a mistake, corrected the error) you can make MathRand() calls not consecutive (i.e. dependent), but thin them out using loops with random length.

 
Mathemat >>:

А точнее - что значит плохо?

But everything you do with your hands is bad, and your children are good :o)

 
The C oscillator built into MT4 is a crooked oscillator - nothing good will come of it. Cyclicality, on the order of 1,000.
 
HideYourRichess >>:
Сишный генератор, встроенный в МТ4 - кривой генератор, - ничего хорошего из него не выйдет. Цикличность, порядка 1000.

Prove it. That is too bold a statement.

 
Mathemat >>:

А точнее - что значит "плохо"? Есть явные нарушения равномерности распределения?

В формуле AB = MathRand() + MathRand()*MathRand() можно сделать вызовы MathRand() не последовательными (т.е. зависимыми), а проредить их с помощью циклов со случайной длиной.


Already

MathRand()*MathRand()
bad (MathRand()1)2*(MathRand()2)5 and (MathRand()1)5*(MathRand()2)2 or is your "*" a plus


Although it's not getting any easier

 
gumgum писал(а) >>

"Or maybe you want your GSH to generate all the numbers. a segment of (0,1) ?" what kind of nonsense is that? Have you studied number theory.

Did you study - is that like "I studied"? :-)

The actual nonsense is this:

gumgum wrote >>

Between n/32768. and n+1/32768. when n e Z & n e [0.32767] is empty. This is not good.

Repeating question. How much would it take between n/32768 and n+1/32768 to be good ?
 

Nah, gumgum, check out my first post on this page, it's got the formula corrected.

Some links for reflection:

https://www.mql5.com/ru/forum/104214/page4

https://www.mql5.com/ru/forum/102961

 
Yurixx >>:

Изучал - это типа "я учил" ? :-)

Действительная глупость вот:

Повторяю вопрос. Сколько нужно чтобы было между n/32768 и n+1/32768 чтобы было хорошо ?


Where is the nonsense here. All numbers are rational (in this case finite (Q-count)). The higher the degree of reliability the better.
 

In short, until there is a clear answer to the question "how much to measure in grams exactly?", further discussion is pointless.

 
Mathemat >>:

Короче, пока не будет внятного ответа на вопрос "скока мерить в граммах точно?", дальнейшее обсуждение бессмысленно.


Can you give me any advice on how to improve the alternator I suggested on page 2.
12345678910...19
New comment