| Age | Commit message (Collapse) | Author |
|---|
|
OptimizationResult now contains:
- cvc5::Result
- optimal value for objective
- whether the objective is unbounded
This gets benefit from cvc5::Result (e.g., we could get explanation for UNKNOWN) and it's slightly easier to integrate to the current API.
Also refactors BV optimizer so that it uses switch statement (instead of if-then-else) to judge the checkSat results (I was planning to do this a long while ago)...
|
|
Use CDList for optimization objectives so that optimization solver supports push and pop (just use SmtEngine's push/pop).
SmtEngine::resetAssertions will also clears the optimization objectives, so no need to have the reset objectives function.
|
|
Add support for box optimization -- independently optimize each goal as if the other goals do not exist.
Single minimize() / maximize() now maintains the pushed / popped context.
Of course unit tests are here as well.
|
|
We store constants, e.g., BitVector and Rational, in our node infrastructure. As a result, we were indirectly including some headers in almost all files, e.g., the GMP headers. This commit changes that by forward-declaring the classes for the constants. As a result, we have to include headers like util/rational.h explicitly when we use Rational but it saves about 3 minutes in compile time (CPU time).
The commit changes RoundingMode from an enum to an enum class such that it can be forward declared.
|
|
add 3 helper functions
judge whether a node is optimizable
make strong improvement expression according to optimization objective
make weak improvement expression according to optimization objective
optChecker is now created by optimizationSolver instead of the minimize/maximize functions
Slightly refactors function signatures so that they are accepting OptimizationObjective instead of accepting target, type in separate parameters.
|
|
for multiple objectives (#6459)
This PR is one part of multiple.
Preliminary support means currently only supports single objective.
It supports queue-ing up objectives and it always optimizes the first.
Multiple-obj optimization algorithm will be up next.
* Refactor optimization result and objective classes
* Add preliminary support for multiple objectives
* The unit tests are now comparing node values instead of node addresses
|
|
(#6452)
If lb = ub + 1 then (lb+ub)/2 =pivot == lb since it's rounding to -infinity.
and lb <= x < pivot will always be UNSAT.
We need to handle this differently.
And this only happens in minimization.
|
|
|
|
|
|
Adds support for BitVector optimization, which is done via offline binary search. Units tests included.
Also mildly refactors the optimizer architecture.
|
