but what's the application that must require non-pseudo random number?

so it's difficult to invent some algorithm that generates non-pseudo random number under the current architecture.

I feel it's enough to use pseudo random number to simulate the randomness, with seed set randomly.

A more interesting question is this:

Does random really exist?

Think about tossing a coin, the result (head or tail) is determined by the initial speed, gravity, wind direction, table material etc. If the modeling of the enviornment is complete, the result will always be the same under the same parameters.

So the randomness comes from the inability of describing the world completely, is that true?

"So the randomness comes from the inability of describing the world completely, is that true?"

I think you are right. However, if the "inability of describing the world completely" is certain to us, random exist, to us.
Maybe, randomness is not universal, but relative to the object concerned.


In a computer, although every instruction is determined, environment out of the world is not. eg. consider we are writing an application that throw a dice when the user click a button. We may use clock() to generate a random number.

Here the randomness is determined by time. Assuming a user (or device) can not distinguish time from 1ms. And our computer's clock() has resolution of 1/6 ms. Shall we treat it as a true random (to the player).

In linux, the interface is /dev/(u)random. Many crypto software use it. check http://en.wikipedia.org/wiki//dev/random

the following command gives 64 bytes of random data:

[atp@risci1 ~]$ dd if=/dev/urandom bs=64 count=1 | od
1+0 records in
1+0 records out
64 bytes (64 B) copied, 6.0784e-05 seconds, 1.1 MB/s
0000000 125047 155010 151240 044101 134716 055204 027564 056065
0000020 000400 167744 111504 136012 136674 175425 045721 005465
0000040 111165 146462 106526 175406 153705 053636 101666 057700
0000060 175272 161005 105256 053347 171741 072074 127474 042150

ing as follows -- One server stores zillions of uniform random numbers between 0 and 1, which are prepared and tested to be "random". When a client request a list of random number, it will first be allocated to a position in the data on server, then the client will generate a psudo-random sequence, and pick the "random" numbers from data according to that sequence.

Like the case of particles.

Nowadays, the chip is built in almost all motherboard. This link is better. http://en.wikipedia.org/wiki/Hardware_random_number_generator

I remember there are also RNG using photon or other quantum stuff. They are not very expensive (price comparable to harddisk, I think), easy to use and not very big in size. Dr Chang Ee Chien mentioned it on his crypto class.

only god throws dies; or even he does not.

we dont talk about religion here

consider tossing a coin. if your model includes the force of tossing it, trajectory, air dynamics, weight of the coin, height, etc, then you can determine which side it will be. So "randomness" is an ideal concept, everything in real-life is, to some extent, psudo-random... which is deterministic.