Random Number Generation on the Glasses (A Natural RNG) | /en/2008/05/random-number-generation-on-the-glasses/

yihui 2022-12-16 14:15:36

Random Number Generation on the Glasses (A Natural RNG)

https://yihui.org/en/2008/05/random-number-generation-on-the-glasses/

3 Comments

giscus-bot 2022-12-16 14:15:37

Guest *Bati* @ 2008-05-16 12:17:14 originally posted:

Nice!

yihui 2022-12-16 14:15:40

The initial value x[1] had better be changed to something like this:

x[1] = sum(as.integer(substring(Sys.time(), c(12, 15, 18), 
  c(13, 16, 19))) * c(3600, 60, 1))/(24 * 60 * 60)

so that we can guarantee the first random number will lie in the range [0, 1] (if we use x[1] = as.numeric(Sys.time()), x[1] will be larger than 1 after about 30 years)

Originally posted on 2008-05-20 09:16:33

giscus-bot 2022-12-16 14:15:38

Guest *michal* @ 2008-06-05 03:43:39 originally posted:

Hi!

Why don't you try inputing actual data from the picture and try analyzing it through some point pattern analysis? You can then test if it is really uniform :) I did not try such a thing before, but it is possible to import a graphics file to R via e.g. 'grid' package, you could then use 'locator' to get a grip on rain drop locations. That would be a funny analysis :)

cheerios!

michal

yihui 2022-12-16 14:15:41

The image of my glasses is not easy to deal with because the elements in it are complicated, I think. If we really want to perform such a test, we'd better use a "cleaner" material to collect the raindrops, and take a picture on a background with a pure color. In that case, we may either simply use locator() or other methods to obtain the location of beads, and perform the further test such as ks.test() or something else.

Originally posted on 2008-06-05 08:07:20

giscus-bot 2022-12-16 14:15:39

Guest *l* @ 2008-12-29 17:18:08 originally posted:

😡