Live data from Hacker News

Software Verification and Analysis Using Z3

research.nccgroup.com

1–10 of 43 posts

Re: Software Verification and Analysis Using Z3

#7
post #2

This may be a dumb question, but how do I know that the model is accurate?

The age old problem of verification vs. validation. Verification is about whether or not you built something right. Validation is about whether or not you built the right thing. This article is about using Z3 to pursue the former, the latter is an entirely different issue.

Re: Software Verification and Analysis Using Z3

#8
post #2

This may be a dumb question, but how do I know that the model is accurate?

Do the opposite: generate the program from the model

Which, in turn, requires trusting or proving the soundness of your program generator and only proves that your incorrect original program is exactly as incorrect as the model that verifies it.

Automating the stronger proof (that the model is exactly as correct as the original program, and that the model is correct) hasn't been solved in the general case, to the best of my knowledge.

Re: Software Verification and Analysis Using Z3

#10
post #2

This may be a dumb question, but how do I know that the model is accurate?

That's not a dumb question! The accuracy of one's underlying models is an outstanding problem in verification.

Let me ask another question,

how do we evaluate the accuracy of a model?

Thank you for your time.

Post reply on HN