Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Ghost functions have a model supporting their erasure #10

Merged
merged 20 commits into from
Feb 29, 2024
Merged

Conversation

nikswamy
Copy link
Contributor

In PulseCore, Ghost functions were previously not semantically distinguished from other atomic .actions, although they were marked as erased. This meant that we needed an axiom to lift ghost functions (with non-informative results) to regular actions. With this PR, we no longer need that axiom.

The technical mechanism enabling this is to move to a bipartite heap PulseCore.Heap2, consisting of a concrete heap and a ghost erased heap. Ghost functions are proven to only mutate the erased heap, and ghost functions now have a model starting from Heap2 and going up through Memory, Action, and Atomic layers.

Note, although this led to removing the axiom, the proofs require quite a lot of boilerplate and duplication between the concrete and ghost heaps. Expect a follow-up PR trying to add some abstraction there to reduce the boilerplate.

@@ -187,7 +187,7 @@ let lift_ghost_neutral
(e:stt_ghost a pre post)
(reveal_a:non_informative_witness a)
: stt_atomic a #Neutral emp_inames pre post
= admit() //This is the main axiom about ghost computations; in Steel, this axiom is implemented within the effect system
= Action.lift_erased reveal_a e
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is the main payoff

#Unobservable emp_inames
(ghost_pts_to r v)
(fun v1 -> ghost_pts_to r v ** pure (fact v1))
// Unused?
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

These are unused, but @mtzguido tells me that they are expected to be required soon. I will restore them in a follow-up PR.

@nikswamy nikswamy merged commit 5b66fee into main Feb 29, 2024
2 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant