mirror of
https://github.com/Z3Prover/z3
synced 2025-08-24 03:57:51 +00:00
optimizations, fixes, TODO items
Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com>
This commit is contained in:
parent
444084f396
commit
6478e789e9
6 changed files with 87 additions and 28 deletions
|
@ -16,9 +16,18 @@ Notes:
|
|||
|
||||
TODO: try a final core reduction step or other core minimization
|
||||
|
||||
TODO: If we have e.g. 4x+y=2 and y=0, then we have a conflict no matter the value of x, so we should drop x=? from the core.
|
||||
(works currently if x is unassigned; for other cases we would need extra info from constraint::is_currently_false)
|
||||
TODO: If we have e.g. 4x+y=2 and y=0, then we have a conflict no matter the value of x, so we should drop x=? from the core.
|
||||
(works currently if x is unassigned; for other cases we would need extra info from constraint::is_currently_false)
|
||||
|
||||
TODO: keep is buggy. The assert
|
||||
SASSERT(premise.is_currently_true(s()) || premise.bvalue(s()) == l_true);
|
||||
does not necessarily hold. A saturation premise could be inserted that is a resolvent that evaluates to false
|
||||
and therefore not a current Boolean literal on the search stack.
|
||||
|
||||
TODO: revert(pvar v) is too weak.
|
||||
It should apply saturation rules currently only available for for propagated values.
|
||||
|
||||
TODO: dependency tracking for constraints evaluating to false should be minimized.
|
||||
--*/
|
||||
|
||||
#include "math/polysat/conflict.h"
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue