Commit graph

486 commits

Author SHA1 Message Date
Leonardo de Moura
1d10953da4 fix(library/elaborator): add hack for experimenting with algebraic hierarchy
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-02-25 11:20:40 -08:00
Leonardo de Moura
309e7ba880 fix(library/elaborator): temporary fix for bug reported by Jeremy
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-02-24 15:26:21 -08:00
Leonardo de Moura
bfe64a7031 fix(library/elaborator): hack for fixing a bug due to pairs/projs, this is temporary fix until we build a new elaborator
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-02-11 21:30:26 -08:00
Leonardo de Moura
b7b868de85 fix(library/elaborator): bug reported by Jeremy Avigad
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-02-10 14:01:22 -08:00
Leonardo de Moura
57982135d9 fix(library/simplifier): bug using congr1 theorem
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-02-09 16:24:29 -08:00
Leonardo de Moura
c45c1748d8 refactor(builtin/kernel): reorder congr1 arguments
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-02-09 16:15:44 -08:00
Leonardo de Moura
2d70e2f4f2 fix(library/tactic/goal): bug in the proof builder
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-02-09 15:02:36 -08:00
Leonardo de Moura
1c43020fc9 fix(library/tactic/goal): bug creating main proof builder
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-02-08 09:11:07 -08:00
Leonardo de Moura
24528ff685 fix(library/elaborator): fix glitches in the elaborator that were forcing us to provide parameters explicitly
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-02-07 18:02:08 -08:00
Leonardo de Moura
1ec01f5757 refactor(builtin): merge pair.lean with kernel.lean, and add basic theorems
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-02-07 16:04:44 -08:00
Leonardo de Moura
ad7b13104f feat(*): add support for heterogeneous equality in the parser, elaborator and simplifier, adjusts unit test to reflect changes
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-02-07 15:03:16 -08:00
Leonardo de Moura
6d7ec9d7b6 refactor(kernel): add heterogeneous equality back to expr
The main motivation is that we will be able to move equalities between universes.

For example, suppose we have
    A : (Type i)
    B : (Type i)
    H : @eq (Type j) A B
where j > i

We didn't find any trick for deducing (@eq (Type i) A B) from H.
Before this commit, heterogeneous equality as a constant with type

   heq : {A B : (Type U)} : A -> B -> Bool

So, from H, we would only be able to deduce

   (@heq (Type j) (Type j) A B)

Not being able to move the equality back to a smaller universe is
problematic in several cases. I list some instances in the end of the commit message.

With this commit, Heterogeneous equality is a special kind of expression.
It is not a constant anymore. From H, we can deduce

   H1 : A == B

That is, we are essentially "erasing" the universes when we move to heterogeneous equality.
Now, since A and B have (Type i), we can deduce (@eq (Type i) A B) from H1. The proof term is

  (to_eq (Type i) A B (to_heq (Type j) A B H))  :  (@eq (Type i) A B)

So, it remains to explain why we need this feature.

For example, suppose we want to state the Pi extensionality axiom.

axiom hpiext {A A' : (Type U)} {B : A → (Type U)} {B' : A' → (Type U)} :
      A = A' → (∀ x x', x == x' → B x == B' x') → (∀ x, B x) == (∀ x, B' x)

This axiom produces an "inflated" equality at (Type U) when we treat heterogeneous
equality as a constant. The conclusion

     (∀ x, B x) == (∀ x, B' x)

is syntax sugar for

   (@heq (Type U) (Type U) (∀ x : A, B x) (∀ x : A', B' x))

Even if A, A', B, B' live in a much smaller universe.

As I described above, it doesn't seem to be a way to move this equality back to a smaller universe.

So, if we wanted to keep the heterogeneous equality as a constant, it seems we would
have to support axiom schemas. That is, hpiext would be parametrized by the universes where
A, A', B and B'. Another possibility would be to have universe polymorphism like Agda.
None of the solutions seem attractive.

So, we decided to have heterogeneous equality as a special kind of expression.
And use the trick above to move equalities back to the right universe.

BTW, the parser is not creating the new heterogeneous equalities yet.
Moreover, kernel.lean still contains a constant name heq2 that is the heterogeneous
equality as a constant.

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-02-07 10:28:10 -08:00
Leonardo de Moura
d4b08fcf96 feat(library/elaborator): be 'lazy' when normalizing terms in the elaborator
Unification constraints of the form

         ctx |- ?m[inst:i v] == T

         and

         ctx |- (?m a1 ... an) == T

are delayed by elaborator because the produce case-splits.
On the other hand, the step that puts terms is head-normal form is eagerly applied.
This is a bad idea for constraints like the two above. The elaborator will put T in head normal form
before executing process_meta_app and process_meta_inst. This is just wasted work, and creates
fully unfolded terms for solvers and provers.

The new test demonstrates the problem. In this test, we mark several terms as non-opaque.
Without this commit, the produced goal is a huge term.

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-02-06 21:08:00 -08:00
Leonardo de Moura
363c4dc5c2 feat(library/elaborator): improve support for dependent pairs in the elaborator
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-02-06 20:56:31 -08:00
Leonardo de Moura
e85b1f1ac0 feat(library/elaborator): expose elaborator configuration options
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-02-06 17:26:38 -08:00
Leonardo de Moura
e4579b93e4 fix(library/elaborator): try first projection before imitation in the higher-order unifier
Projections build more general solutions. This commit also adds a test that demonstrates the issue. Before this commit, the elaborator would produce the "constant" predicate (fun x, a + b = b + a).

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-02-06 13:35:05 -08:00
Leonardo de Moura
c01f82aeb7 feat(builtin): add sum types
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-02-05 23:04:44 -08:00
Leonardo de Moura
f4ec874c6e refactor(builtin): remove dead module heq
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-02-04 14:42:28 -08:00
Leonardo de Moura
0283887ee9 refactor(builtin/kernel): move the heq axioms into kernel.lean
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-02-04 14:17:34 -08:00
Leonardo de Moura
f03c09c10b feat(library/elaborator): add support for proj/pair/sigma in the the higher-order unification procedure
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-02-04 13:53:30 -08:00
Leonardo de Moura
cc96b50644 feat(frontends/lean): support for nary-tuples, improve notation for non-dependent tuples, add support in the elaborator for sigma types
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-02-03 20:53:11 -08:00
Leonardo de Moura
8eec289ce1 feat(kernel): add dependent pairs
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-02-03 16:52:49 -08:00
Leonardo de Moura
6be50f0133 refactor(builtin/heq): merge cast and heq modules
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-02-02 15:01:48 -08:00
Leonardo de Moura
c56df132b8 refactor(kernel): remove semantic attachments from the kernel
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-02-02 14:48:27 -08:00
Leonardo de Moura
e3dc552c39 fix(library/simplifier): nontermination
The example tests/lua/simp1.lua demonstrates the issue.
The higher-order matcher matches closed terms that are definitionally equal.
So, given a definition

    definition a := 1

it will match 'a' with '1' since they are definitionally equal.

Then, if we have a theorem

    theorem a_eq_1 : a = 1

as a rewrite rule, it was triggering the following infinite loop when simplifying the expression "a"

   a --> 1 --> 1 --> 1 ...

The first simplification is expected. The other ones are not.
The problem is that "1" is definitionally equal to "a", and they match.
The rewrite_rule_set manager accepts the rule a --> 1 since the left-hand-side does not occur in the right-hand-side.

To avoid this loop, we test if the new expression is not equal to the previous one.

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-31 15:55:21 -08:00
Leonardo de Moura
1d85267d26 fix(library/simplifier): assumptions/context may contain equations where the left-hand-side is a metavariable or semantic attachment
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-31 15:38:22 -08:00
Leonardo de Moura
110ca84984 feat(library/simplifier): allow the user to associate a simplifier monitor with the lua_State object
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-31 13:49:24 -08:00
Leonardo de Moura
b444756d20 fix(library/simplifier): missing condition in implication simplification
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-30 22:14:01 -08:00
Leonardo de Moura
759aa61f70 refactor(builtin/kernel): define if-then-else using Hilbert's operator
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-30 19:28:42 -08:00
Leonardo de Moura
b45ab9dc30 feat(library/elaborator): use equality constraints instead of convertability constraints on definitions
Convertability constraints are harder to solve than equality constraints, and it seems they don't buy us anything definitions. They are just increasing the search space for the elaborator.

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-30 14:13:21 -08:00
Leonardo de Moura
41f5e2a067 feat(library/simplifier): statically check (conditional) equations (aka rewrite rules) to verify whether we can skip type checking when using them in the simplifier
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-29 18:32:40 -08:00
Leonardo de Moura
4f3127d3d5 fix(library/simplifier): check if the given types are convertible to ceq expected types
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-29 15:15:55 -08:00
Leonardo de Moura
a19f9d4846 feat(library/simplifier): discard conditional equations that are clearly non-terminating
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-29 13:56:45 -08:00
Leonardo de Moura
dd6aae378f fix(library/simplifier): must use metavar_env in is_ceq, otherwise it may ceqs that contain metavariables
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-29 13:29:20 -08:00
Leonardo de Moura
069e5edf6b fix(library/simplifier): include flag indicating if the proof generated by simplifier is a homogenous or heterogenous equality, use flag to fix bug in the simp_tactic
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-29 12:01:01 -08:00
Leonardo de Moura
92ba4e8b2d feat(library/simplifier): add support for metavariables in conditional rewrite rules
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-29 08:34:04 -08:00
Leonardo de Moura
24452289dd feat(library/simplifier): make sure the simplifier can handle meta-variables
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-28 20:30:47 -08:00
Leonardo de Moura
ee4344076e feat(library/simplifier): improve error message when simplifier is looping
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-28 19:36:31 -08:00
Leonardo de Moura
b31ef34787 feat(library/simplifier): preserve binder names when applying higher-order rewrite rules
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-28 00:50:27 -08:00
Leonardo de Moura
6da1b447f0 fix(library/hop_match): do not match iff with =
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-28 00:21:05 -08:00
Leonardo de Moura
dbdbd211e3 fix(library/simplifier): compilation warning
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-27 23:29:53 -08:00
Leonardo de Moura
160dc71cb5 refactor(kernel/type_checker): use read-only metavariable environment in methods that do not require write access to the metavariable environment
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-27 17:38:06 -08:00
Leonardo de Moura
3b152d1a9e refactor(kernel): use ro_metavar_env instead of metavar_env in places where we only need to read the metavariable environment
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-27 16:44:43 -08:00
Leonardo de Moura
8bccfb947a feat(library/simplifier): expose simplier and simplifier_monitor objects in the Lua API
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-27 15:02:05 -08:00
Leonardo de Moura
c088825ef0 feat(library/simplifier): add simplifier_monitor interface
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-27 11:46:28 -08:00
Leonardo de Moura
579b751e01 fix(library/simplifier): compilation warning
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-26 23:16:24 -08:00
Leonardo de Moura
4d25cb7f47 feat(library/tactic): add simplify_tactic based on the simplifier
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-26 18:53:18 -08:00
Leonardo de Moura
5e6c1d4904 refactor(builtin/heq): remove axiom hpiext since we don't use it anymore
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-26 13:11:17 -08:00
Leonardo de Moura
50df761d90 refactor(library/simplifier): remove the is_typem hack, it is not needed anymore now that we don't use hpiext anymore
Now, we are again using the following invariant for simplifier_fn::result
The type of in the equality of the result is definitionally equal to the
type of the resultant expression.

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-26 12:21:23 -08:00
Leonardo de Moura
29e448f034 fix(library/simplifier): remove support in the simplifier for (forall x : A, B x) when it is not a proposition, the problem is that hpiext axiom produces an equality in a too big universe
For example, in the hpiext axiom, the resultant equality if for (Type M+1)

axiom hpiext {A A' : TypeM} {B : A -> TypeM} {B' : A' -> TypeM} :
      A = A' -> (∀ x x', x == x' -> B x = B' x') -> (∀ x, B x) = (∀ x, B' x)

even if the actual arguments A, A’, B, B’ "live" in a much smaller universe (e.g., Type).

So, it would be great if we could move the resultant equality back to the right universe.
I don't see how to do it right now.

The other solution would require a major rewrite of the code base.
We would have to support universe level arguments like Agda, and write the axiom hpiext as:

axiom hpiext {l : level} {A A' : (Type l)} {B : A -> (Type l)} {B' : A' -> (Type l)} :
      A = A' -> (∀ x x', x == x' -> B x = B' x') -> (∀ x, B x) = (∀ x, B' x)

This is the first instance I found where it is really handy to have this feature.
I think this would be a super clean solution, but it would require a big rewrite in the code base.
Another problem is that the actual semantics that Agda has for this kind of construction is not clear to me.
For instance, sometimes Agda reports that the type of an expression is (Set omega).

An easier to implement hack is to support "axiom templates".
We create instances of hipext "on-demand" for different universe levels.
This is essentially what Coq does, since the universe levels are implicit in Coq.
This is not as clean as the Agda approach, but it is much easier to implement.

A super dirty trick is to include some instances of hpiext for commonly used universes
(e.g., Type and (Type 1)).

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-01-26 12:10:34 -08:00