I have multiple realizations of the same function contract. Some are naive and straightforward, some are more complex and optimized. I'd like to run them over randomly picked points from input domain using PropSpec
.
The question is how to run all redundant realizations and compare output pair-wise. The test should be marked as failed in case computed values differ from one realization to another. If there are more than two realizations it should be possible to decide which one failed based on voting, like in the TMR system