From the course: Exploring Deterministic LLM Programming
Unlock this course with a free trial
Join today to access over 25,500 courses taught by industry experts.
Provability score
From the course: Exploring Deterministic LLM Programming
Provability score
- [Instructor] Traditional testing can run code a thousand times and still miss an edge case that crashes in production. But what if you could mathematically prove your code would never fail? And that's what the PMAT Provability Analysis and other similar tools do. They use an abstract interpretation to verify that every possible execution path runs within a single test. In this case, the provability test shows that the visualization of a simple Rust divide function achieves this 100% provability because it's a mathematical proof. It's not just a statistic. And if you look at how, for example, if statements could be transformed into unprovable code, this is one of the ways that we want to look at provability. Essentially, we want to make sure that, when we write any kind of a code example, that that code example has a way of being proven in a mathematical sense, just like in math class. You want to write the equation…
Contents
-
-
-
-
Illusion of control agents3m 31s
-
(Locked)
Increasing determinism in deep context3m 32s
-
(Locked)
Understanding AST agentic coding3m 32s
-
(Locked)
Understanding complexity3m 4s
-
(Locked)
Understanding entropy3m 27s
-
(Locked)
Understanding code churn3m 24s
-
(Locked)
Roadmap sprints and technical debt3m 9s
-
(Locked)
SATD predicts technical debt2m 40s
-
(Locked)
Provability score3m 13s
-
-