Jump to content

  • Log In with Google      Sign In   
  • Create Account

polyfrag

Member Since 20 Mar 2003
Offline Last Active Yesterday, 01:36 AM

Posts I've Made

In Topic: Local hash

05 April 2016 - 02:14 PM

A, who cares...

In Topic: Local hash

05 April 2016 - 12:21 PM

If anybody knows of a way to keep track of used integers, I would like to know. My method isn't as efficient as I thought I think. Maybe there's multiple ways to encode the same values with that way. E.g... 0m0 01m 101 = 000 010 010 011 101 = 010 011 101 000 = 01m 101 000... Mmm


In Topic: Local hash

21 March 2016 - 10:22 PM

I tried to make it work. I can now get any single mapping of any bit width and levels. I can get 3 mappings together with up to 4 bit x 8 level masks in 7 ms. It takes too long for anything like 30 mappings for 64x60. 2 levels are needed at most for each additional mapping. n n-bit levels will just feed through the input if set to 0.

 

It's a kind of neural net too because it learns by back-propagation and learns patterns. You can also tell it to avoid a certain output given certain inputs, or to try some other combination that will satisfy the inputs and outputs.

 

I found another thing which might be useful, which is, keeping track of numbers that have been tried. For an n-bit number, at most 2*n (+1?) entries with 2 n-bit numbers each are needed. One is a mask for bits that have been tried both ways, and the other is the value that's been tried once. E.g., for a 5-bit number, there will be 11 max entries, because anything else is a 1-bit change away from any of these:

 

00000

00011

00101

01001

01010

01100

10001

10010

10100

11000

11011

 

https://dl.dropboxusercontent.com/u/109630018/temp/lm/localmap.h

https://dl.dropboxusercontent.com/u/109630018/temp/lm/localmap.cpp


In Topic: Blackhole internal space

09 November 2015 - 02:51 AM

I am no physicist. I just had an idea. Only good things come from solving equations. Yes I guess its conjecturing not hypothesizing.

In Topic: Blackhole internal space

09 November 2015 - 01:55 AM

Has anybody ever tried plotting the gravitational motion along a straight path, for all acceptable values? Somebody must have already solved the Taylor series, so why am I expecting to find anything. But I want to see what I will get. So don't want to do this. I will gather the patience. Might take a few days. Help greatly appreciated.

Look at this beast!

y(x) = 0 + ( sigma n={2} to {inf} ((n-1)! * 1/n! * x^n * 1/( 2-y(n-1) )^n) )

Also, a bulb trajectory is still calculated even with absolute radius, except it is flipped upside and doesn't flow continuously.

PARTNERS