michael/lean2
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
2087 commits 3 branches 0 tags 28 MiB
Commit graph

371 commits

Author SHA1 Message Date
Leonardo de Moura
8df7c7b02d feat(kernel/type_checker): remove fallback that expands opaque definitions in the type checker 
We should not rely on this feature. It can be quite expensive.
We invoke is_convertible in several places, in particular, if we are using overloading. For example, the frontend uses is_convertible to check which overload should be used. Thus, it will make several calls such as

   is_convertible(num, Nat)

If is_convertible starts unfolding opaque definitions, we would keep expanding num.

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-02-09 11:31:24 -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
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
593f1f2ebd fix(frontends/lean): allow user set constants defined in other namespaces as opaque 
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-02-06 20:56:31 -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
ea06bb2885 feat(frontends/lean/pp): change how lift local entries are pretty printed 
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-02-06 17:26:38 -08:00
Leonardo de Moura
a51139e63b feat(frontends/lean): position information in error messages 
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
ef321e730f feat(builtin/tactic): add the 'skip' (bogus) tactic for ignoring a proof hole in a big proof 
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-02-06 12:47:23 -08:00
Leonardo de Moura
eab0456b27 test(tests/lean): test super opaque style 
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-02-06 11:33:27 -08:00
Leonardo de Moura
fdc4c9b53c test(tests/lean): add nested 'have' expression test 
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-02-06 08:13:04 -08:00
Leonardo de Moura
1d23d93e60 feat(frontends/lean): new 'have' expression 
Add 'have' notation suggested by Jeremy Avigad.
Add his example to the test suite.

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-02-06 08:03:59 -08:00
Leonardo de Moura
ba9a8f9d98 feat(frontends/lean): add 'show' expression syntax sugar 
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-02-06 07:50:22 -08:00
Leonardo de Moura
419fb7464e fix(tests/lean): adjust tests to reflect recent changes 
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-02-05 21:04:06 -08:00
Leonardo de Moura
aec9c84d0d fix(util/lua): deadlock 
Errors in the Lua library produce longjmps.
The longjmp will not unwind the C++ stack.
In the new test, the lock was not being released, and the system was deadlocking in the next call that tried to lock the environment

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-02-04 15:16:29 -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
413391b2b4 chore(tests/lean/sig2): remove unnecessary parenthesis from test 
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-02-04 11:37:08 -08:00
Leonardo de Moura
c9b72df34b fix(frontends/lean/parser): bug when applying tactics to synthesize remaining meta-variables 
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-02-04 11:36:26 -08:00
Leonardo de Moura
96c9c7505a test(tests/lean): add another sigma-type test 
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-02-04 10:12:39 -08:00
Leonardo de Moura
9dc86e3cf5 fix(builtin/kernel): rename generalized proof_irrel axiom to hproof_irrel, and derive the restricted one 
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-02-04 10:06:29 -08:00
Leonardo de Moura
4fcc292332 feat(frontends/lean): parse and pretty print pair/tuple projection operators proj1 and proj2, fix bug in the type checker 
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-02-03 22:10:01 -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
5c991f8fbf feat(frontends/lean): parse and pretty print tuples/pairs 
This commit also fixes a bug in the type checker when processing dependent pairs.

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-02-03 20:10:30 -08:00
Leonardo de Moura
640ebcc040 test(tests/lean/exp): add example for Steve 
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-02-03 11:35:40 -08:00
Leonardo de Moura
e4afa3dc43 fix(tests/lean/map): incorrect output 
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-02-03 00:05:19 -08:00
Leonardo de Moura
87f9c9b27e fix(tests/lean/map): make sure the unit test produce the same result in different platforms 
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-02-02 20:29:34 -08:00
Leonardo de Moura
e2add5c9f2 test(tests/lean): add heterogeneous equality simplification example 
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-02-02 20:27:39 -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
2b7bc7b673 test(tests/lean/exp): simulating HOL constructions 
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-01-31 17:59:57 -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
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
8c1f6b9055 fix(kernel/typechecker): allow elaborator to infer (Type U+1) 
In the new test elab8.lean, the parameter B is in (Type U+1).
Before, this commit, the type checker was forcing all metavariables that must be types to be <= (Type U). This restriction was preventing the elaborator from succeeding in reasonable cases.

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-01-30 14:00:01 -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
01259b1e84 feat(kernel): make sure U is the maximal universe 
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-01-29 16:31:00 -08:00
Leonardo de Moura
ea6bf224e5 feat(frontends/lean): make the parser accept (Type -> ...) 
Before this commit, the parser would accept only a universe level or a ')' after '(' 'Type'

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-01-29 15:23:20 -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
0bb8fe75b3 test(tests/lean): new simplifier test 
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-01-29 14:21:18 -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
4dc3aa46c3 feat(frontends/lean): allow tactics to be used in axiom/variable declarations and in the type of definitions/theorems; add a new test showing the need for this feature 
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-01-29 12:02:12 -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
62408a6adc test(tests/lean): move simp_loop test to slow subdirectory 
This example produces a stackoverflow on Valgrind.
We don't execute Valgrind on tests in the slow subdirectory.

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-01-29 10:32:48 -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
72c607846a test(tests/lean): add Jeremy's proof to test suite 
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
 2014-01-28 16:05:46 -08:00