Add the ability to customize incremental pre-processing simplification for the SMTLIB2 front-end. The main new capability is to use pre-processing tactics in incremental mode that were previously not available. The main new capabilities are
- solve-eqs
- reduce-args
- elim-unconstrained
There are several more. Documentation and exposed simplifiers are populated incrementally. The current set of supported simplifiers can be inspected by using z3 with the --simplifiers flag or referring to https://microsoft.github.io/z3guide/docs/strategies/simplifiers
Some pending features are:
- add the ability to update parameters to simplifiers similar to how tactics can be controlled using parameters.
- expose simplification solvers over the binary API.
move sat_smt_preprocess to solver
fix bugs in model_reconstruction_trail for dependency replay
This is a preparatory step for exposing pre-processing as tactics.
- convert reduce-args to a simplifier. Currently exposed as reduce-args2 tactic until the old tactic code gets removed.
- bug fixes in model_reconstruction trail
- allow multiple defs to be added with same pool of removed formulas
- fix tracking of function symbols instead of expressions to filter replay
- add nla_divisions to track (cheap) divisibility lemmas.
-
- increase build version to 4.12.1. This prepares updated release for MacOs-11 build on x86
- move literal propagation mode in euf-egraph to a callback and traversal of equivalence class. Track antecedent by newest equality instead of root. This makes equality propagation to literals have similar behavior as in legacy solver and appears to result in a speedup (10% fewer conflicts on QF_UF/QG-classification/qg5/iso_icl478.smt2 in preliminary testing)
- fix interaction of pre-processing and assumptions. Pre-processing has to freeze assumption literals so they don't get eliminated. This is similar to dependencies that are already frozen.
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))
```
- enable sat.smt in smt_tactic that
is invoked by default on first goals
add flatten-clauses
add push-ite
have tptp5 front-end pretty print SMT2 formulas a little nicer.
rename size() to qtail() and introduce shortcuts
ensure tactic goals are not updated if they are in inconsistent state (because indices could be invalidated)
- add sat.smt option to enable the new incremental core (it is not ready for mainstream consumption as cloning and other features are not implemented and it hasn't been tested in any detail yet).
- move "name" into attribute on simplifier so it can be reused for diagnostics by the seq-simplifier.
in an iteration of inc-sat-solver introduce sat-smt-solver to allow incremental pre-processing.
The aim is to allow incrementally handling formulas while at the same time retaining the main benefits of global in/pre-processing that change models. Previous incremental solving capabilities have been limited to use pre-processing that does not require model conversion.
other updates:
- change signature of advance_qhead to simplify call sites
- have model reconstruction replay work on a tail of dependent_expr state, while adding formulas to the tail.
literals true/false are not necessarily created when the merge flag is set.
Also disable merge_tf for if-then-else expressions
Perhaps even not insert children of if expressions into congruence table?
perf fix for propagation behavior for equalities in the new core.
The old behavior was not to allow congruence closure on equalities.
The new behavior is to just not allow merging tf with equalities unless they appear somewhere in a foreign context (not under a Boolean operator)
The change re-surfaces merge_tf and enable_cgc distinction from the old core.
They can both be turned on or off.
merge_enabled renamed to cgc_enabled
The change is highly likely to introduce regressions in the new core.
Change propagation of literals from congruence:
- track antecedent enode. There are four ways to propagate
literals from the egraph.
- the literal is an equality and the two arguments are congruent
- the antecedent is merged with node n and the antecedent has a Boolean variable assignment.
- the antecedent is true or false, they are merged.
- the merge_tf flag is toggled to true but the node n has not been merged with true/false
this update breaks use cases that set sat.smt.proof to True.
As it is such a new feature and the change affects possibly at most the tutorial it is made without compatibility layers.
* Memory leak in .NET user-propagator
The user-propagator object has to be manually disposed (IDisposable), otherwise it stays in memory forever, as it cannot be garbage collected automatically
* Throw an exception if variable passed to decide is already assigned instead of running in an assertion violation
* Added limit to "visit" to allow detecting multiple visits
* Putting visit in a separate class
(Reason: We will probably need two of them in the sat::solver)
* Bugfix
Removing ternary clause optimization from sat_solver simplifies special case handling of ternary clauses throughout the sat solver and dependent solvers (pb_solver). Benchmarking on QF_BV suggests the ternary clause optimization does not have any effect. While removing ternary clause optimization two bugs in unit propagation were also uncovered: it missed propagations when the only a single undef literal remained in the non-watched literals and it did not update blocked literals in cases where it could in the watch list. These performance bugs were for general clauses, ternary clause propagation did not miss propagations (and don't use blocked literals), but fixing these issues for general clauses appear to have made ternary clause optimization irrelevant based on what was measured.
gc-ing definitions leads to unsoundness when they are not replayed.
Instead of attempting to replay definitions theory internalization is irredundant by default.
This is also the old solver behavior where TH_LEMMA is essentially never used, but is valid for top-level theory lemmas.
move self-checking functionality to inside sat/smt so it can be used on-line and not just off-line.
when self-validation fails, use vs, not clause, to check. It allows self-validation without checking and maintaining RUP validation.
new options sat.smt.proof.check_rup, sat.smt.proof.check for online validation.
z3 sat.smt.proof.check=true sat.euf=true /v:1 sat.smt.proof.check_rup=true /st file.smt2 sat.smt.proof=p.smt2
EUF proofs are checked modulo union-find.
Equalities are added to to union-find if they are assumptions or if they can be derived using congruence closure. The congruence closure assumptions are added as proof-hints.
Note that this proof format does not track equality inferences, symmetry and transitivity. Instead they are handled by assuming a union-find based checker.
#6319 - fix incompleteness in propagation of default to all array terms in the equivalence class.
Fix bug with q_mbi where domain restrictions are not using values because the current model does not evaluate certain bound variables to values. Set model completion when adding these bound variables to the model to ensure their values are not missed.
Add better propagation of diagnostics when tactics and the new solver return unknown. The reason for unknown can now be traced to what theory was culprit (currently no additional information)
using a queue for disequality propagaiton was a regression: values of numerals can change along the same stack so prior passing the filter does not mean it passes later.
literals that are replayed need to be registered with respective theories, otherwise, they will not propagate with the theories (the enode have to be attached with relevant theory variables).
delay propagation on each disequality/equality should suffice once. It adds relevant inequalities to ensure the arithmetic solver is coherent about disequalities.
This commit overhauls the proof format (in development) for the new core.
NOTE: this functionality is work in progress with a long way to go.
It is shielded by the sat.euf option, which is off by default and in pre-release state.
It is too early to fuzz or use it. It is pushed into master to shed light on road-map for certifying inferences of sat.euf.
It retires the ad-hoc extension of DRUP used by the SAT solver.
Instead it relies on SMT with ad-hoc extensions for proof terms.
It adds the following commands (consumed by proof_cmds.cpp):
- assume - for input clauses
- learn - when a clause is learned (or redundant clause is added)
- del - when a clause is deleted.
The commands take a list of expressions of type Bool and the
last argument can optionally be of type Proof.
When the last argument is of type Proof it is provided as a hint
to justify the learned clause.
Proof hints can be checked using a self-contained proof
checker. The sat/smt/euf_proof_checker.h class provides
a plugin dispatcher for checkers.
It is instantiated with a checker for arithmetic lemmas,
so far for Farkas proofs.
Use example:
```
(set-option :sat.euf true)
(set-option :tactic.default_tactic smt)
(set-option :sat.smt.proof f.proof)
(declare-const x Int)
(declare-const y Int)
(declare-const z Int)
(declare-const u Int)
(assert (< x y))
(assert (< y z))
(assert (< z x))
(check-sat)
```
Run z3 on a file with above content.
Then run z3 on f.proof
```
(verified-smt)
(verified-smt)
(verified-smt)
(verified-farkas)
(verified-smt)
```
this update addresses some perf regressions introduced when handling axioms for bv2int and a memory smash regression when decoupling bv-ackerman from in-processing. It adds a filter based on bv_eq_axioms for disabling ackerman reductions on disequalities.
previous scheme has Ackmerman module instrument main solver to backjump and simplify when reaching a threshold.
This destroys overall performance: simplification does many more things than invoking Ackerman axioms.
Having a dependency between simplification (in-processing) and depleting a priority queue of auxiliary axioms therefore hurts overall performance. It has to be decoupled. The current approach is now to empty the axiom queue on occasion.
It is still not ideal - it should be coupled with the search level - axioms don't survive higher levels where redundant clauses get garbage collected as they don't have a chance of being used.
This update allows the python bindings for user-propagator to handle functions that are declared to be registered with the user propagator plugin. It fixes a bug in UserPropagateBase.add to allow registering terms dynamically during search.
It also fixes a bug in theory_user_propagate as scopes were not fully pushed when the solver gets the callbacks for new equalities and new disequalities.
It also adds equality and disequality interfaces to the sat/smt solver version (which isn't being exercised in earnest yet)
