New Appoach for Random Number Generation???

[Namir]

(03-21-2019 01:53 PM)Albert Chan Wrote:  

(03-21-2019 12:06 PM)Namir Wrote:  your choices for f1(x) and f2(x) are HUGE!!!

Letting user pick f1 and f2, without any statistical analysis, is probably a bad idea.

With above setup, it is conceivable that for some seed, the period is low, possibly even 1.

Also, I think you mean L = floor(log10(Diff)), x = FRAC( Diff / 10^(L-4) )

In other words, replacing 5 sig. digits of Diff, with a decimal point.

Selecting the functions f1 and f2 (and they should be appomixations to the same target function) should be done along with testing them of course. The selection is an open wide field and can be a proverbial mine field if the choices ae not made wisely and tested. In other words, with freedom comes responsibility.

In calculating L, I used two seperate statements for the sake of clarity.