How Can We Generate High-Quality Random Seeds for Pseudo-Random Number Generators?

Generating High-Quality Random Seeds for Pseudo-Random Number Generators

The initialization of the pseudo-random number generator (PRNG) using srand() is crucial for generating high-quality random numbers. The article you referenced suggests using the output of the time() function as a seed, considering its distinctive value each second. However, for applications that run multiple times per second, this approach may result in collisions.

A more reliable alternative is to combine multiple sources of entropy to generate a robust seed. One recommended approach for portable Linux applications is to use the following formula:

unsigned long seed = mix(clock(), time(NULL), getpid());

Here, mix() is a function that combines three sources of entropy:

• clock() returns the CPU time consumed by the program, which varies with each execution.
• time(NULL) returns the Unix time, which is unique within each second.
• getpid() returns the process ID, which identifies a unique process instance.

By combining these sources, we generate a seed that is highly distinctive and provides a superior foundation for the PRNG.

The mix() function is a modified version of Robert Jenkins' 96-bit Mix Function, which effectively scrambles the input values to create a highly randomized output seed.

The above is the detailed content of How Can We Generate High-Quality Random Seeds for Pseudo-Random Number Generators?. For more information, please follow other related articles on the PHP Chinese website!