Problem:
Currently, when performing random distribution verification, we determine
the probability of each element occurring in the sum, but the probability is
only an estimate, these tests had rare sporadic failures, and we cannot verify
what the probability of failure will be.
Solution:
Using the chi-square distribution instead of the original random distribution
validation makes the test more reasonable and easier to find problems.