Yes, it's good to think through these sort of things every now and again
- especially when reality conflicts with intuition (it *seems* obvious that adding together lots of random values will give you a more "random" result, but playing with a few dice will show that's not the case).
That's what makes debugging so much fun! Just try stepping through motor controller code...
mvh.,
David
If you already have a bell curve that is a good indication that you have a good random generator. Adding several will take pseudorandomness away. It is to be expected that e.g. a geiger teller counter, one of the best, will generate only numbers in a limited range, such as a million with a spread of 1000. You have to take that into account.
As I said, if the width of the bell curve is 16, take the 4 least significant bits. If this is a good generator (such as a Geiger), you can concatenate 8 samples to get
32 bits total. I would trust that to generate random primes for military grade encryption.Groetjes Albert
--
can
arge
sult
alue
nd
,Hide quoted text -
I agree that this method would work well in practice, but formally I question whether even this result is actually "random". Certainly - I cannot think of a better method than concatenating input from a Geiger, thermometer, or even a photocell.
You could sample amplitude of radio signals, or frequency, and as long as the digit you are sampling from is uniformly distributed you should be able to get something useable.
Still - I question whether even this is truly random. Useable, yes. Random, maybe.
Have something to add? Share your thoughts — no account required.
Ask the community — no account required