Randomness is all around us. an d yet it is surprisingly difficult to generate a truly random number electronically. Among the first projects I ever built were head/tails and dice, The heads tails was a flop. Doing a hundred tosses a definite bias appeared, it wasn't major but statistically noticeable. It was enough I scrapped the project. So if someone presents you with a table of random number how do you test them? distribution of number is one way, If you have a list of numbers between 1 and 6 (dice)then all of the numbers should show up 16.7 % 0f the time .This percentage is rounded, so there is an error right there. ( 100/6) Statistics can be tricky, it is an entire branch of math unto itself. But like I said, true randomness is very common and electronics can be used to take advantage of it to create a true random generator.
Of course, Chaos theory steps up to make things complicated where we were hoping for simplicity, but since I am no math whiz I choose to ignore it as inconvenient for now (just being smart enough to know my limitations.
In its simplest form a random generator cane be the simply a one or a zero (Heads/tails). You can check for bias by flipping it 1000 times, If it is truly random you wind up with 50%/50% distribution, you know it is bias free at that point, but what about other forms of distortion that means it is not truly random, There are names for them but I am too ignorant to know how to even describe them, how would you test for them and what are they? If someone gives you a list of 1 billion 1's and 0's claiming it is a truly random table you could chop it up into byte size chunks, throw away any numbers that are out of range and generate a table that can be used as you desire. How do you make sure this list is random? Distribution offers one cheap and dirty test, as I have talked about.
How to generate randomness
Electronic noise (AKA white noise) frequently used for S/N measurements. ideally it is random as it is a frequency band containing all the frequencies in a defined band. Precision notch a frequency out and feed this though an amplifier any noise introduced by an amplifier is a measurement of signal to noise. This offers a strong hint as to a source of white noise. Thermal noise is a random quantity in nature find a component that is noisy to generate whit noise and you have a potential white noise generator Yet the S/N test shows one of the distortions I am discuss int We are removing a specific component of the frequency distribution. So could some version of a spectrum analyzer be used to check our random generator source? Dang if I know.
Another natural source of randomness is radioactivity. I have repeatedly read that particle decay is statistically random in nature, There are plenty of mildly radioactive source available out there, Not so for particle detectors, Other than Geiger tubes I don't know of any solid state device that can detect a particle from a radioactive decay event. I would love to find an alternative.
So where am I going with this?
I am about to start a thread on possible ways to create a hypothetical 1Billion digit random binary number and am looking for ways to test it if I did. I was stuck on a long Taxi ride to nowhere and I was mulling this problem over in my head to keep me occupied. Yep I quite literally had too much time on my hand.
Discuss?
Of course, Chaos theory steps up to make things complicated where we were hoping for simplicity, but since I am no math whiz I choose to ignore it as inconvenient for now (just being smart enough to know my limitations.
In its simplest form a random generator cane be the simply a one or a zero (Heads/tails). You can check for bias by flipping it 1000 times, If it is truly random you wind up with 50%/50% distribution, you know it is bias free at that point, but what about other forms of distortion that means it is not truly random, There are names for them but I am too ignorant to know how to even describe them, how would you test for them and what are they? If someone gives you a list of 1 billion 1's and 0's claiming it is a truly random table you could chop it up into byte size chunks, throw away any numbers that are out of range and generate a table that can be used as you desire. How do you make sure this list is random? Distribution offers one cheap and dirty test, as I have talked about.
How to generate randomness
Electronic noise (AKA white noise) frequently used for S/N measurements. ideally it is random as it is a frequency band containing all the frequencies in a defined band. Precision notch a frequency out and feed this though an amplifier any noise introduced by an amplifier is a measurement of signal to noise. This offers a strong hint as to a source of white noise. Thermal noise is a random quantity in nature find a component that is noisy to generate whit noise and you have a potential white noise generator Yet the S/N test shows one of the distortions I am discuss int We are removing a specific component of the frequency distribution. So could some version of a spectrum analyzer be used to check our random generator source? Dang if I know.
Another natural source of randomness is radioactivity. I have repeatedly read that particle decay is statistically random in nature, There are plenty of mildly radioactive source available out there, Not so for particle detectors, Other than Geiger tubes I don't know of any solid state device that can detect a particle from a radioactive decay event. I would love to find an alternative.
So where am I going with this?
I am about to start a thread on possible ways to create a hypothetical 1Billion digit random binary number and am looking for ways to test it if I did. I was stuck on a long Taxi ride to nowhere and I was mulling this problem over in my head to keep me occupied. Yep I quite literally had too much time on my hand.
Discuss?
Last edited: