]>
matita.cs.unibo.it Git - fireball-separation.git/log
acondolu [Mon, 24 Jul 2017 15:19:59 +0000 (17:19 +0200)]
Fixing the bug when eating triggers the reduction of a match
acondolu [Mon, 24 Jul 2017 14:18:19 +0000 (16:18 +0200)]
Added problem failing because of order of step
acondolu [Mon, 24 Jul 2017 14:17:53 +0000 (16:17 +0200)]
Test --with-pac fails if unseparable+uncomplete
acondolu [Mon, 24 Jul 2017 14:15:53 +0000 (16:15 +0200)]
Allow comments (#) at the end of terms
acondolu [Mon, 24 Jul 2017 14:15:19 +0000 (16:15 +0200)]
Logging options to display measure of single terms in string_of_problem
acondolu [Mon, 24 Jul 2017 13:38:50 +0000 (15:38 +0200)]
Fixed bug in Scott.mk_match when no branches (Fixes problem 1 in solved)
acondolu [Mon, 24 Jul 2017 12:55:31 +0000 (14:55 +0200)]
Fixed bug in computing special k
acondolu [Mon, 24 Jul 2017 12:55:04 +0000 (14:55 +0200)]
Added exception Lambda, which may be caught when a conv becomes a lambda
acondolu [Mon, 24 Jul 2017 12:44:47 +0000 (14:44 +0200)]
Added very simple generator of tests with bombs/pacmans, run it with --with-pac
acondolu [Mon, 24 Jul 2017 11:09:05 +0000 (13:09 +0200)]
Fixed eta_compare in the case Lam vs. Bot
acondolu [Mon, 24 Jul 2017 12:05:08 +0000 (14:05 +0200)]
Removed special strategies in choose_step
acondolu [Sun, 23 Jul 2017 20:23:34 +0000 (22:23 +0200)]
Better error messages in parser
acondolu [Sun, 23 Jul 2017 20:22:41 +0000 (22:22 +0200)]
Check for absence of bombs and pacmans
acondolu [Sun, 23 Jul 2017 07:59:00 +0000 (09:59 +0200)]
Revert subst in branches of matches
Actually, it could be moved in susbt_in_problem,
hence doing it only once (on the deltas), instead
of replacing multiple times like now (useless).
acondolu [Sun, 23 Jul 2017 07:31:07 +0000 (09:31 +0200)]
Cleanup in w, added names, uncommented test for low special_k
acondolu [Sat, 22 Jul 2017 22:33:00 +0000 (00:33 +0200)]
Change to compute_special_k
Now the special_k taked into account the number or arguments (arity) of
terms. This solves problem "low special_k" in problems/w
acondolu [Sat, 22 Jul 2017 20:35:44 +0000 (22:35 +0200)]
Comments, fix indentation and
acondolu [Sat, 22 Jul 2017 19:58:31 +0000 (21:58 +0200)]
Renamed in critical_showstoppers: showstoppers -> inedible (I don't know why)
acondolu [Sat, 22 Jul 2017 19:34:23 +0000 (21:34 +0200)]
Simplified arity_of, precompute_edible_data, critical_showsteppers
acondolu [Sat, 22 Jul 2017 19:00:26 +0000 (21:00 +0200)]
Fixed computation of arity for top-level inerts
acondolu [Sat, 22 Jul 2017 18:59:55 +0000 (20:59 +0200)]
Simplified not-working example
Claudio Sacerdoti Coen [Tue, 18 Jul 2017 15:55:29 +0000 (17:55 +0200)]
Missing parenthesis fixed
Claudio Sacerdoti Coen [Tue, 18 Jul 2017 13:38:27 +0000 (15:38 +0200)]
Failing case found
Claudio Sacerdoti Coen [Tue, 18 Jul 2017 13:36:56 +0000 (15:36 +0200)]
Hack to reserve var 0 to purify it to BOM removed
Claudio Sacerdoti Coen [Tue, 18 Jul 2017 13:26:48 +0000 (15:26 +0200)]
String ==> Bytes to make recent ocaml compilers happy
acondolu [Sun, 16 Jul 2017 17:36:19 +0000 (19:36 +0200)]
Extracting solutions of 3-coloring problems
Added sample unsat problems found on the internet
acondolu [Sun, 16 Jul 2017 14:01:21 +0000 (16:01 +0200)]
subst does not replace anymore in branches of maches
It was unnecessary: a match cannot expand to a previous branch
acondolu [Sun, 16 Jul 2017 13:41:49 +0000 (15:41 +0200)]
Commented out problem, it is (still) failing because div occurs in conv
acondolu [Sun, 16 Jul 2017 13:41:16 +0000 (15:41 +0200)]
Printing now preserves the names of the original free variables
acondolu [Sat, 15 Jul 2017 18:25:07 +0000 (20:25 +0200)]
Fixed typo in problem
acondolu [Sat, 15 Jul 2017 18:23:20 +0000 (20:23 +0200)]
Improved parsing
- removed unused parse'
- fixed bug which was ignoring the first problems in some files
acondolu [Sat, 15 Jul 2017 18:21:39 +0000 (20:21 +0200)]
Improved pretty-printing
- Num does not use print_name anymore
- Original names of free variables are preserved
- bound variables' names start with x
acondolu [Sat, 15 Jul 2017 15:16:51 +0000 (17:16 +0200)]
Added problem wrongly backtracking
acondolu [Sat, 15 Jul 2017 15:00:04 +0000 (17:00 +0200)]
Forgot one "Z"
acondolu [Sat, 15 Jul 2017 14:47:50 +0000 (16:47 +0200)]
Code clean-up
acondolu [Sat, 15 Jul 2017 14:41:53 +0000 (16:41 +0200)]
Code clean-up
acondolu [Sat, 15 Jul 2017 14:41:39 +0000 (16:41 +0200)]
Important: added special variable "Z" for zero.
- Removed append_zero in Lambda4
- Now tests append "Z" at the end of ps'
- Manually appended "Z" at the end of all problems
- Removed tmp function in Lambda4, renamed to problem_of
acondolu [Sat, 15 Jul 2017 14:01:36 +0000 (16:01 +0200)]
Moved all problems in problems folder. Last ones in "w"
acondolu [Sat, 15 Jul 2017 13:31:50 +0000 (15:31 +0200)]
Encoding of SAT problems and 3-colorability of graphs
acondolu [Sat, 15 Jul 2017 13:31:04 +0000 (15:31 +0200)]
Fixes to parser
acondolu [Sat, 15 Jul 2017 13:05:59 +0000 (15:05 +0200)]
Added problems encoded from 3-colorability of graphs
acondolu [Sat, 15 Jul 2017 13:05:34 +0000 (15:05 +0200)]
Fixed "$" in parser
acondolu [Fri, 14 Jul 2017 18:34:02 +0000 (20:34 +0200)]
Command line arguments to ./a.out
For example:
./a.out problems/*
./a.out problems/p
Moved "o"-problems to problems/o
acondolu [Fri, 14 Jul 2017 18:32:20 +0000 (20:32 +0200)]
Fixed problems separator in parser: now it is $
A separable problem's label may start with '!', a not separable's with '?'
acondolu [Fri, 14 Jul 2017 18:30:55 +0000 (20:30 +0200)]
Added label to problem
acondolu [Fri, 14 Jul 2017 17:39:42 +0000 (19:39 +0200)]
Removed invalid problems
acondolu [Fri, 14 Jul 2017 17:39:21 +0000 (19:39 +0200)]
Failure in case of empty problem, to be fixed
acondolu [Fri, 14 Jul 2017 17:28:29 +0000 (19:28 +0200)]
Parser.from_file, and p* problems moved to problems/p
acondolu [Fri, 14 Jul 2017 17:25:23 +0000 (19:25 +0200)]
problems now contain a label and the names of the original free variables
acondolu [Fri, 14 Jul 2017 16:02:19 +0000 (18:02 +0200)]
First draft of Parser.problem_of_string
- problem_of_string should unify old Lambda4.problem_of and Parser.parse'
- Lambda4.problem_of was temporarily factored in problem_of_2
and problem_of_String_tmp
acondolu [Fri, 14 Jul 2017 15:20:42 +0000 (17:20 +0200)]
Changed logic of entrypoint in Lambda4
- removed exception Fail
- added "check",which checks if we are in a subproblem
of separation on which we know to be complete
- "solve" does not raise particular exceptions anymore,
and the program raise failure only if there is some bug
acondolu [Fri, 14 Jul 2017 14:44:15 +0000 (16:44 +0200)]
Moved parse' from Num to Parser
acondolu [Fri, 14 Jul 2017 12:39:10 +0000 (14:39 +0200)]
Fixed: was using compare in place of eta_eq
in particular, eta_eq ignores arities of variables
there used to be cases of duplicate entries in PS (up to arities of variables)
acondolu [Fri, 14 Jul 2017 12:38:08 +0000 (14:38 +0200)]
Re-use Util.index_of
acondolu [Fri, 14 Jul 2017 12:31:30 +0000 (14:31 +0200)]
Code clean-up
Less casts and casts in the right place
acondolu [Fri, 14 Jul 2017 08:57:22 +0000 (10:57 +0200)]
Tidying up problems.ml
acondolu [Fri, 14 Jul 2017 08:28:09 +0000 (10:28 +0200)]
New failing problem, which shows that we need to change special_k
acondolu [Thu, 13 Jul 2017 18:35:37 +0000 (20:35 +0200)]
Added assert false
acondolu [Thu, 13 Jul 2017 18:35:10 +0000 (20:35 +0200)]
Fixed bug in eta_subterm
acondolu [Thu, 13 Jul 2017 18:34:50 +0000 (20:34 +0200)]
Remove FORCE option in Makefile
acondolu [Thu, 13 Jul 2017 17:05:39 +0000 (19:05 +0200)]
Fix: missing check to avoid stepping on vars of non positive arity
acondolu [Thu, 13 Jul 2017 16:53:50 +0000 (18:53 +0200)]
More problems
acondolu [Thu, 13 Jul 2017 16:52:32 +0000 (18:52 +0200)]
return type of simplify_match improved (no `Match in output)
acondolu [Thu, 13 Jul 2017 16:24:07 +0000 (18:24 +0200)]
Added safety check for step on n-th argument with n > 1
Avoid stepping on n if there is a div/number that would be turned into
a Lambda
acondolu [Thu, 13 Jul 2017 16:19:33 +0000 (18:19 +0200)]
Improved error messages
acondolu [Thu, 13 Jul 2017 16:17:06 +0000 (18:17 +0200)]
Failing problems commented out
acondolu [Thu, 13 Jul 2017 16:08:35 +0000 (18:08 +0200)]
New tests
acondolu [Thu, 13 Jul 2017 16:08:08 +0000 (18:08 +0200)]
Fix (??): some broken tests with Div = variable used to pass
acondolu [Thu, 13 Jul 2017 15:38:23 +0000 (17:38 +0200)]
Tentative commit: tactics dropped and clean-up
acondolu [Thu, 13 Jul 2017 15:34:12 +0000 (17:34 +0200)]
Fix: max_arity_tms was using wrong comparison
acondolu [Thu, 13 Jul 2017 09:00:36 +0000 (11:00 +0200)]
Fixed try in aux_conv + Added test in pure for divergence of conv
acondolu [Wed, 12 Jul 2017 21:28:46 +0000 (23:28 +0200)]
Fixes to printing
acondolu [Wed, 12 Jul 2017 21:23:43 +0000 (23:23 +0200)]
Added more problems with bombs, all working
acondolu [Wed, 12 Jul 2017 21:12:44 +0000 (23:12 +0200)]
Disabled some chars as variable names in the parser
acondolu [Wed, 12 Jul 2017 21:11:54 +0000 (23:11 +0200)]
Removed bomb variable
In Pure:
- Added constructor B to Pure.t
- Added rule corresponding to B (bottom) in Pure.mwhd
In Lambda4:
- Instantiating the bomb now actually uses subst_in_problem
- Two new exceptions: Bottom and Pacman
- Fixed subst_in_problem by making expand_match raise these two exceptions,
which are caught when tolerated
acondolu [Wed, 12 Jul 2017 19:48:01 +0000 (21:48 +0200)]
Prettier-printing: string_of_problem outputs OCaml code
acondolu [Wed, 12 Jul 2017 16:52:57 +0000 (18:52 +0200)]
Added second problem with bombs and pacmans: it works!
acondolu [Wed, 12 Jul 2017 16:39:03 +0000 (18:39 +0200)]
Added new (passing) problem with a bomb (o1)
acondolu [Wed, 12 Jul 2017 15:24:54 +0000 (17:24 +0200)]
Towards `Bottom and `Pacman
1) Num updated to handle them
2) lambda4 compiles: safety conditions for projections hopefully
implemented; strategy code not changed, therefore it will step over
bombs and friends
acondolu [Tue, 11 Jul 2017 16:17:23 +0000 (18:17 +0200)]
New: simple backtracking with trail
auto_instantiate replaced by choose_step and first
acondolu [Tue, 11 Jul 2017 15:40:00 +0000 (17:40 +0200)]
Fixed if-then-else
acondolu [Tue, 11 Jul 2017 15:39:37 +0000 (17:39 +0200)]
Fix: nested `Lambda(false,_) were not handled correctly
acondolu [Tue, 11 Jul 2017 15:38:14 +0000 (17:38 +0200)]
-666 -> min_int
acondolu [Tue, 11 Jul 2017 14:38:26 +0000 (16:38 +0200)]
Fixed indentation and logging
acondolu [Tue, 11 Jul 2017 14:36:23 +0000 (16:36 +0200)]
instantiate takes an additional argument which is the number of the argument to the variable on which to step
acondolu [Tue, 11 Jul 2017 13:44:40 +0000 (15:44 +0200)]
clean removes *.o files
acondolu [Tue, 11 Jul 2017 13:44:16 +0000 (15:44 +0200)]
Removed old .ar files
acondolu [Tue, 11 Jul 2017 13:42:09 +0000 (15:42 +0200)]
Run tests with: make run
acondolu [Tue, 11 Jul 2017 12:00:34 +0000 (14:00 +0200)]
Fix: conv : x:min_int (.... y:1 ...) y was not considered for stepping
When no other choice is available, step on any >0 variable that occurs in
a conv.
acondolu [Tue, 11 Jul 2017 11:24:09 +0000 (13:24 +0200)]
New problems n1, n2 used to debug the problem with dangerous variables fixed in the last commits
acondolu [Tue, 11 Jul 2017 11:22:41 +0000 (13:22 +0200)]
Bug fix: x (y z) where y was conv_dangerous did not make x so
Additional fix to last commit: args' where no longer checked for dangerous
variables. Fixed by passing three arguments to dangerous_invert_conv in
place of two.
acondolu [Tue, 11 Jul 2017 09:37:18 +0000 (11:37 +0200)]
Fix bug in dangerous_conv with arity of arguments of matches.
Next bug: variable replaced twice
acondolu [Mon, 10 Jul 2017 16:44:31 +0000 (18:44 +0200)]
The measure finally works on all problems in problems.ml!
Added in logs/fail.txt a fragment of a failing problem
acondolu [Mon, 10 Jul 2017 13:48:25 +0000 (15:48 +0200)]
Experimenting with a combined measure
acondolu [Mon, 10 Jul 2017 12:43:43 +0000 (14:43 +0200)]
Instantiate now uses a global initialSpecialK (ignoring the local one)
This fixes the bug in the previous commit, and makes the problem pass
acondolu [Mon, 10 Jul 2017 12:13:36 +0000 (14:13 +0200)]
Arity inherited only in the case (true, min_int)
Problem 25 fails because a fake variable becomes
critical step.
acondolu [Mon, 10 Jul 2017 12:06:55 +0000 (14:06 +0200)]
new_arity = old_arity + 1
acondolu [Fri, 7 Jul 2017 17:16:29 +0000 (19:16 +0200)]
Fixes to how arities are assigned and propagated
- added truelam to subst (useless, can be removed)
- TODO: to measure the problem, count the `fake` variables
which are the ones with arity = min_int
acondolu [Wed, 28 Jun 2017 14:29:21 +0000 (16:29 +0200)]
Minor fixes
<andrea.condoluci@unibo.it> [Tue, 27 Jun 2017 13:16:43 +0000 (15:16 +0200)]
still stepping on negative variables