Random.org is a website that produces "true" random numbers based on atmospheric noise.
I do not know what is atmospheric noise. so i am not sure that whether those random numbers produced by Random.org could used for simple sampling? thank you!
I don't know how 'true' atmospheric noise generators are, or how they are produced, but they are clearly meant to be physical rather than pseudo. I'm not sure why you would want true- rather than psuedo-randomness, but if you do, why not XOR a psuedo and true sequence? This would give a physical random number stream that has at least as good uniform and correlation properties as the better component.
Hi,
Your question is very interesting one. If you want to find out the random numbers. You can use the random numbers produced by Random.org to form for simple sampling or you use the lottery method also for your study or research.
See the following link:
https://www.ma.utexas.edu/users/mks/statmistakes/SRS.html
After looking at the FAQ at random.org, I would be fairly confident to use their numbers for random sampling. But I still wonder why you wouldn't use a good pseudo-random number generator. For selecting a single sample random.org would be fine, but for extensive simulations you could need far more numbers than random.org could supply (see the last paragraph below).
The most popular RNGs are linear congruential generators. Some are very good and some are terrible; good ones have passed comprehensive tests for randomness. Another very popular one is the Mersenne twister. This produces a very long sequence of numbers, much longer than the age of the universe in picoseconds, before the sequence repeats. Individual numbers can repeat, just as they would with 'true' generators. Linear congruential generators don't repeat (if you take them to machine precision) until the whole sequence repeats.
Atmospheric generators use random noise picked up by a radio not tuned to a station. This is generally known as 'atmospherics', but in fact, unless the radios are cooled with liquid nitrogen, most of the noise is generated by random motion of electrons in the radio itself, and only a small amount comes from the atmosphere.
The numbers in the website come from three radios tuned to different frequencies and the sequences are combined (probably the bits are XORed). Thus if someone suddenly broacasts on one of the frequencies, it would not affect the randomness, and random.org would soon realise and change the frequency.
The FAQ list many certifying authorities who have recommended the sequences, but I would be more impressed if they said what tests for randomness had been done. Look up DIEHARD for one such set of tests (more comprehensive extensions of DIEHARD are also available).
Physical generators have to be carefully adjusted. Typically, if the voltage in the detector is above a certain level, a 1 is produced, otherwise a 0 is produced. That level must be set just right. The voltage changes quickly but not instantaneously, so the generator cannot be sampled too quickly or the bits will be correlated. This limits the number of random bits that can be produced in a day.
I should add that there are outside sources of atmospheric noise such as vaccuum cleaners, electric drills, car ignition systems and lightning. But I think most of what people call 'atmospherics' is internal to the radio. If I were to build a physical RNG it would just be an amplifier and a device to turn noise into 0s and 1s. I won't build one as you can already buy one that plugs into a computer port. However I'm happy with pseudo RNGs.
- Badges
- Science topic
- Similar topics
- Biological Science
- Physiology
- Physiological Processes
- Stress