## Problem
Fixes#10133. For `QF_BV` with `parallel.enable=true`, the solver could
return `sat` with a model that violates its own assertions.
`mk_qfbv_tactic` routes SAT solving to `mk_psat_tactic`, which built the
solver via `mk_inc_sat_solver(m, p, false)` (non-incremental). The
parallel cube-and-conquer engine **reuses this single solver** across
many `check_sat(cube)` calls, but SAT variable/blocked-clause
elimination ran because the simplifier's incremental gate was disabled.
Elimination model reconstruction is only sound for a single one-shot
solve; under repeated cube assumptions it produces models where
eliminated Tseitin variables get values contradicting the original
clauses.
## Root cause (two coupled defects)
1. **Stale simplifier cache.** `inc_sat_solver`'s constructor called
`m_solver.set_incremental()` *after* `updt_params()`. The SAT simplifier
caches `m_incremental_mode` from the SAT config *during* `updt_params`,
so setting the incremental flag afterwards left a stale non-incremental
mode and elimination stayed enabled. (This is why simply passing
`incremental_mode=true` had no effect on its own.)
2. **Non-incremental parallel solver.** `mk_psat_tactic` created the
reused solver as non-incremental.
## Fix
- Move `set_incremental()` **before** `updt_params()` in the
`inc_sat_solver` constructor so the simplifier caches the correct mode.
- Create the parallel solver with `incremental_mode=true` in
`mk_psat_tactic`.
## Validation
- Variable elimination on the QF_BV parallel path drops to zero
(`sat-elim-bool-vars-res` 21664 -> 0); remaining `elim-clauses/literals`
are sound cleaner/subsumption steps.
- Consistent with ground truth: on 4.16.0, `sat.elim_vars=false` removed
all 7 `failed to verify` errors; this change disables exactly that
unsound elimination, scoped to the reused/parallel solver.
- A solvable QF_BV instance returns a **valid** model under
`parallel.enable=true model_validate=true`.
- All 92 unit tests pass (`test-z3 /a`).
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Add new parallel algorithm as a tactic (parallel_tactical2.cpp)
Don't port over old experiments from smt_parallel that we aren't using
(sls, inprocessing, failed_literal_mode for bb detection)
Fix bugs: lease cancellation/reslimit race condition, involves changing
lease epoch to simple boolean flag
Also, now there is a single shared set of params for the tactic and
smt_parallel
**Test runs for the parallel_tactical2 vs old smt_parallel version:**
run-2747-Z3-threads-4-qflia-30s-stats.md
run-2746-Z3-threads-4-qflia-30s-parallel_tactic-stats.md
run-2745-Z3-threads-1-qfbv-30s-stats.md
run-3013-Z3-threads-4-qfbv-30s-parallel_tactic-stats.md --> note this is
indeed run-3013, I reran after a bugfix in inc_sat_solver
run-2743-Z3-threads-4-qfnia-30s-stats.md
run-2742-Z3-threads-4-qfnia-30s-parallel_tactic-stats.md
**Test runs for the new smt_parallel with bugfixes:**
run-2801-Z3-threads-4-qflia-30s-smtparallel-bugfixes-stats.md,
run-2800-Z3-threads-4-qflia-30s-smtparallel-bugfixes-stats.md
run-2797-Z3-threads-4-qfnia-30s-smtparallel-bugfixes-stats.md
compare to old smt_parallel:
run-2747-Z3-threads-4-qflia-30s-stats.md
run-2743-Z3-threads-4-qfnia-30s-stats.md
Note that there is a slight regression on lia in run-2800. The source of
this appears to be the new new LP largest-cube LIA heuristic param,
which is enabled by default. disabling this param in run-2801 restored
performance (I didn't change this in this PR though, just something to
note)
http://mtzguido.tplinkdns.com:8081/z3/compare_stats.html
---------
Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com>
Co-authored-by: Ilana Shapiro <ilanashapiro@Ilanas-MacBook-Pro.local>
Co-authored-by: Ilana Shapiro <ilanashapiro@Ilanas-MBP.localdomain>
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
* Initial plan
* Refactor mk_and and mk_or call sites to use overloaded methods
Changed 130 call sites across 64 files to use vector overloads directly instead of manually passing .size() and .data()/.c_ptr()
Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com>
* Revert mk_or changes for ptr_buffer/ptr_vector (no overload exists in ast_util.h)
* Fix compilation errors from mk_and/mk_or refactoring
Fixed type mismatches by:
- Removing m parameter for expr_ref_vector (ast_util.h has mk_and/mk_or(expr_ref_vector) overloads)
- Reverting changes for ref_buffer types (no overload exists in ast_util.h, only in ast.h for m.mk_and)
- Verified build succeeds and Z3 works correctly
Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com>
* Fix test files to use correct mk_and/mk_or overloads
Changed test/doc.cpp and test/udoc_relation.cpp to use mk_and(expr_ref_vector) and mk_or(expr_ref_vector) without m parameter
Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com>
---------
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com>
the SAT model converter is applied by m_solver.get_model()
Calling m_sat_mc->flush_smc causes the SAT model converter to be inherited within m_sat_mc and then the model converter gets applied again. This time breaking the model.
flush_smc is reserved for incremental use:
* Introduce X-macro-based trace tag definition
- Created trace_tags.def to centralize TRACE tag definitions
- Each tag includes a symbolic name and description
- Set up enum class TraceTag for type-safe usage in TRACE macros
* Add script to generate Markdown documentation from trace_tags.def
- Python script parses trace_tags.def and outputs trace_tags.md
* Refactor TRACE_NEW to prepend TraceTag and pass enum to is_trace_enabled
* trace: improve trace tag handling system with hierarchical tagging
- Introduce hierarchical tag-class structure: enabling a tag class activates all child tags
- Unify TRACE, STRACE, SCTRACE, and CTRACE under enum TraceTag
- Implement initial version of trace_tag.def using X(tag, tag_class, description)
(class names and descriptions to be refined in a future update)
* trace: replace all string-based TRACE tags with enum TraceTag
- Migrated all TRACE, STRACE, SCTRACE, and CTRACE macros to use enum TraceTag values instead of raw string literals
* trace : add cstring header
* trace : Add Markdown documentation generation from trace_tags.def via mk_api_doc.py
* trace : rename macro parameter 'class' to 'tag_class' and remove Unicode comment in trace_tags.h.
* trace : Add TODO comment for future implementation of tag_class activation
* trace : Disable code related to tag_class until implementation is ready (#7663).
Add API solve_for(vars).
It takes a list of variables and returns a triangular solved form for the variables.
Currently for arithmetic. The solved form is a list with elements of the form (var, term, guard).
Variables solved in the tail of the list do not occur before in the list.
For example it can return a solution [(x, z, True), (y, x + z, True)] because first x was solved to be z,
then y was solved to be x + z which is the same as 2z.
Add congruent_explain that retuns an explanation for congruent terms.
Terms congruent in the final state after calling SimpleSolver().check() can be queried for
an explanation, i.e., a list of literals that collectively entail the equality under congruence closure.
The literals are asserted in the final state of search.
Adjust smt_context cancellation for the smt.qi.max_instantiations parameter.
It gets checked when qi-queue elements are consumed.
Prior it was checked on insertion time, which didn't allow for processing as many
instantations as there were in the queue. Moreover, it would not cancel the solver.
So it would keep adding instantations to the queue when it was full / depleted the
configuration limit.
this will allow copying the solver state within a scope.
The new solver state has its state at level 0. It is not possible to pop scopes from the new solver (you can still pop scopes from the original solver). The reason for this semantics is the relative difficulty of implementing (getting it right) of a state copy that preserves scopes.
This update includes an experimental feature to access a congruence closure data-structure after search.
It comes with several caveats as pre-processing is free to eliminate terms. It is therefore necessary to use a solver that does not eliminate the terms you want to track for congruence of. This is partially addressed by using SimpleSolver or incremental mode solving.
```python
from z3 import *
s = SimpleSolver()
x, y, z = Ints('x y z')
s.add(x == y)
s.add(y == z)
s.check()
print(s.root(x), s.root(y), s.root(z))
print(s.next(x), s.next(y), s.next(z))
```
* split sat2goal out of goal2sat
These two classes need different things out of the sat::solver class,
and separating them makes it easier to fiddle with their dependencies
independently.
I also fiddled with some headers to make it possible to include
sat_solver_core.h instead of sat_solver.h.
* limit solver_core methods to those needed by goal2sat
And switch sat2goal and sat_tactic over to relying on the derived
sat::solver class instead. There were no other uses of solver_core.
I'm hoping this makes it feasible to reuse goal2sat's CNF conversion
from places like the tseitin-cnf tactic, so they can be unified into a
single implementation.
