16th
New post: verifying complex properties by software
Proving all the theorems lead us to assume that our problems satisfy a
consistency property. That is a nasty part of the paper, in the second
appendix. I’ve coded a verification procedure that shouldn’t be that
complicated. The problem is that the answer is true/false. Consistent
or inconsistent. How you get sure that your code is right or,
even worst, that there are more subtle details to be considered?.
I found a consequence of the definition that was more
informative than true/false and found an inconsistency. Now the problem
can be in our theory, in my code, or in the benchmarks were are testing
on. That’s not good.
At least this seems to be the last experiment for the journal paper.
Something has to move in some direction, beyond all the stuff we are
packing for moving back to Caracas.
Posted by email from Conformant thesis (posterous) | Comment »
