1 Chaotic functions are often successfully embedded
2 into pseudorandom sequence generators.
3 However even if these generators
4 can effectively face statistical tests
5 (thanks to optimized parameters), they may lose
6 their chaos properties since
7 they are designed using floating point numbers instead of real ones
9 In a previous work, we had constructed chaos-based PRNGs
10 defined on a discrete domain. In this context, proofs that are
11 formally established are preserved during the execution.
12 We had shown that these generators successfully pass
13 the most stringent statistical tests
14 and we had proven that some of them are chaotic.
15 This work completes this task and shows that
16 this whole class of PRNGS is chaotic
17 if and only if some given graphs of iteration are strongly connected.
