mirror of
https://github.com/Z3Prover/z3
synced 2025-10-08 17:01:55 +00:00
Parallel solving (#7852)
* very basic setup * ensure solve_eqs is fully disabled when smt.solve_eqs=false, #7743 Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * respect smt configuration parameter in elim_unconstrained simplifier Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * indentation * add bash files for test runs * add option to selectively disable variable solving for only ground expressions Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * remove verbose output Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * fix #7745 axioms for len(substr(...)) escaped due to nested rewriting * ensure atomic constraints are processed by arithmetic solver * #7739 optimization add simplification rule for at(x, offset) = "" Introducing j just postpones some rewrites that prevent useful simplifications. Z3 already uses common sub-expressions. The example highlights some opportunities for simplification, noteworthy at(..) = "". The example is solved in both versions after adding this simplification. * fix unsound len(substr) axiom Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * FreshConst is_sort (#7748) * #7750 add pre-processing simplification * Add parameter validation for selected API functions * updates to ac-plugin fix incrementality bugs by allowing destructive updates during saturation at the cost of redoing saturation after a pop. * enable passive, add check for bloom up-to-date * add top-k fixed-sized min-heap priority queue for top scoring literals * set up worker thread batch manager for multithreaded batch cubes paradigm, need to debug as I am getting segfault still * fix bug in parallel solving batch setup * fix bug * allow for internalize implies * disable pre-processing during cubing * debugging * remove default constructor * remove a bunch of string copies * Update euf_ac_plugin.cpp include reduction rules in forward simplification * Update euf_completion.cpp try out restricting scope of equalities added by instantation * Update smt_parallel.cpp Drop non-relevant units from shared structures. * process cubes as lists of individual lits * merge * Add support for Algebraic Datatypes in JavaScript/TypeScript bindings (#7734) * Initial plan * Add datatype type definitions to types.ts (work in progress) Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Complete datatype type definitions with working TypeScript compilation Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Implement core datatype functionality with TypeScript compilation success Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Complete datatype implementation with full Context integration and tests 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> * chipping away at the new code structure * comments * debug infinite recursion and split cubes on existing split atoms that aren't in the cube * share lemmas, learn from unsat core, try to debug a couple of things, there was a subtle bug that i have a hard time repro'ing * merge * fix #7603: race condition in Ctrl-C handling (#7755) * fix #7603: race condition in Ctrl-C handling * fix race in cancel_eh * fix build * add arithemtic saturation * add an option to register callback on quantifier instantiation Suppose a user propagator encodes axioms using quantifiers and uses E-matching for instantiation. If it wants to implement a custom priority scheme or drop some instances based on internal checks it can register a callback with quantifier instantiation * missing new closure Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * add Z3_solver_propagate_on_binding to ml callback declarations Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * add python file Signed-off-by: Lev Nachmanson <levnach@Levs-MacBook-Pro.local> * debug under defined calls Signed-off-by: Lev Nachmanson <levnach@hotmail.com> * more untangle params Signed-off-by: Lev Nachmanson <levnach@hotmail.com> * precalc parameters to define the eval order Signed-off-by: Lev Nachmanson <levnach@hotmail.com> * remove a printout Signed-off-by: Lev Nachmanson <levnach@hotmail.com> * rename a Python file Signed-off-by: Lev Nachmanson <levnach@hotmail.com> * add on_binding callbacks across APIs update release notes, add to Java, .Net, C++ * use jboolean in Native interface Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * register on_binding attribute Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * fix java build for java bindings Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * avoid interferring side-effects in function calls Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * remove theory_str and classes that are only used by it * remove automata from python build Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * remove ref to theory_str Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * get the finest factorizations before project Signed-off-by: Lev Nachmanson <levnach@hotmail.com> * rename add_lcs to add_lc Signed-off-by: Lev Nachmanson <levnach@hotmail.com> * resolve bad bug about l2g and g2l translators using wrong global context. add some debug prints * initial attempt at dynamically switching from greedy to frugal splitting strategy in return_cubes. need to test. also there is some bug where the threads take forever to cancel? * Update RELEASE_NOTES.md * resolve bug about not translating managers correctly for the second phase of the greedy cubing, and the frugal fallback * remove unused square-free check Signed-off-by: Lev Nachmanson <levnach@hotmail.com> * add some debug prints and impelement internal polynomial fix * restore the square-free check Signed-off-by: Lev Nachmanson <levnach@hotmail.com> * add some comments and debug m_assumptions_used * redo greedy->frugal strategy so we don't split on existing cubes in frugal at all (eliminate the incorrect/wasteful step by processing current batch first) * set up initial scaffolding for sharing clauses between threads and batch manager. needs some reworking/debug still * Add .github/copilot-instructions.md with comprehensive Z3 development guide (#7766) * Initial plan * Add comprehensive .github/copilot-instructions.md with validated build commands and timing Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Remove test_example binary file from repository 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> * Bump actions/checkout from 4 to 5 (#7773) Bumps [actions/checkout](https://github.com/actions/checkout) from 4 to 5. - [Release notes](https://github.com/actions/checkout/releases) - [Changelog](https://github.com/actions/checkout/blob/main/CHANGELOG.md) - [Commits](https://github.com/actions/checkout/compare/v4...v5) --- updated-dependencies: - dependency-name: actions/checkout dependency-version: '5' dependency-type: direct:production update-type: version-update:semver-major ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> * turn off logging at level 0 for testing * add max thread conflicts backoff * Parallel solving (#7775) * very basic setup * very basic setup (#7741) * add score access and reset Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * added notes Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * Update PARALLEL_PROJECT_NOTES.md * Update PARALLEL_PROJECT_NOTES.md * Update PARALLEL_PROJECT_NOTES.md * add bash files for test runs * fix compilation Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * more notes Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * Update PARALLEL_PROJECT_NOTES.md * Update PARALLEL_PROJECT_NOTES.md * Update PARALLEL_PROJECT_NOTES.md * Update PARALLEL_PROJECT_NOTES.md * Update PARALLEL_PROJECT_NOTES.md * Update PARALLEL_PROJECT_NOTES.md * Update PARALLEL_PROJECT_NOTES.md * Update PARALLEL_PROJECT_NOTES.md * Update PARALLEL_PROJECT_NOTES.md * add top-k fixed-sized min-heap priority queue for top scoring literals * fixed-size min-heap for tracking top-k literals (#7752) * very basic setup * ensure solve_eqs is fully disabled when smt.solve_eqs=false, #7743 Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * respect smt configuration parameter in elim_unconstrained simplifier Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * indentation * add bash files for test runs * add option to selectively disable variable solving for only ground expressions Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * remove verbose output Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * fix #7745 axioms for len(substr(...)) escaped due to nested rewriting * ensure atomic constraints are processed by arithmetic solver * #7739 optimization add simplification rule for at(x, offset) = "" Introducing j just postpones some rewrites that prevent useful simplifications. Z3 already uses common sub-expressions. The example highlights some opportunities for simplification, noteworthy at(..) = "". The example is solved in both versions after adding this simplification. * fix unsound len(substr) axiom Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * FreshConst is_sort (#7748) * #7750 add pre-processing simplification * Add parameter validation for selected API functions * updates to ac-plugin fix incrementality bugs by allowing destructive updates during saturation at the cost of redoing saturation after a pop. * enable passive, add check for bloom up-to-date * add top-k fixed-sized min-heap priority queue for top scoring literals --------- Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> Co-authored-by: Nikolaj Bjorner <nbjorner@microsoft.com> Co-authored-by: humnrdble <83878671+humnrdble@users.noreply.github.com> * set up worker thread batch manager for multithreaded batch cubes paradigm, need to debug as I am getting segfault still * fix bug in parallel solving batch setup * fix bug * debugging * process cubes as lists of individual lits * Parallel solving (#7756) * very basic setup * ensure solve_eqs is fully disabled when smt.solve_eqs=false, #7743 Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * respect smt configuration parameter in elim_unconstrained simplifier Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * indentation * add bash files for test runs * add option to selectively disable variable solving for only ground expressions Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * remove verbose output Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * fix #7745 axioms for len(substr(...)) escaped due to nested rewriting * ensure atomic constraints are processed by arithmetic solver * #7739 optimization add simplification rule for at(x, offset) = "" Introducing j just postpones some rewrites that prevent useful simplifications. Z3 already uses common sub-expressions. The example highlights some opportunities for simplification, noteworthy at(..) = "". The example is solved in both versions after adding this simplification. * fix unsound len(substr) axiom Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * FreshConst is_sort (#7748) * #7750 add pre-processing simplification * Add parameter validation for selected API functions * updates to ac-plugin fix incrementality bugs by allowing destructive updates during saturation at the cost of redoing saturation after a pop. * enable passive, add check for bloom up-to-date * add top-k fixed-sized min-heap priority queue for top scoring literals * set up worker thread batch manager for multithreaded batch cubes paradigm, need to debug as I am getting segfault still * fix bug in parallel solving batch setup * fix bug * allow for internalize implies * disable pre-processing during cubing * debugging * process cubes as lists of individual lits --------- Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> Co-authored-by: Nikolaj Bjorner <nbjorner@microsoft.com> Co-authored-by: humnrdble <83878671+humnrdble@users.noreply.github.com> * snapshot Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * pair programming Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * pair programming Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * merge * chipping away at the new code structure * Parallel solving (#7758) * very basic setup * ensure solve_eqs is fully disabled when smt.solve_eqs=false, #7743 Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * respect smt configuration parameter in elim_unconstrained simplifier Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * indentation * add bash files for test runs * add option to selectively disable variable solving for only ground expressions Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * remove verbose output Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * fix #7745 axioms for len(substr(...)) escaped due to nested rewriting * ensure atomic constraints are processed by arithmetic solver * #7739 optimization add simplification rule for at(x, offset) = "" Introducing j just postpones some rewrites that prevent useful simplifications. Z3 already uses common sub-expressions. The example highlights some opportunities for simplification, noteworthy at(..) = "". The example is solved in both versions after adding this simplification. * fix unsound len(substr) axiom Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * FreshConst is_sort (#7748) * #7750 add pre-processing simplification * Add parameter validation for selected API functions * updates to ac-plugin fix incrementality bugs by allowing destructive updates during saturation at the cost of redoing saturation after a pop. * enable passive, add check for bloom up-to-date * add top-k fixed-sized min-heap priority queue for top scoring literals * set up worker thread batch manager for multithreaded batch cubes paradigm, need to debug as I am getting segfault still * fix bug in parallel solving batch setup * fix bug * allow for internalize implies * disable pre-processing during cubing * debugging * remove default constructor * remove a bunch of string copies * Update euf_ac_plugin.cpp include reduction rules in forward simplification * Update euf_completion.cpp try out restricting scope of equalities added by instantation * Update smt_parallel.cpp Drop non-relevant units from shared structures. * process cubes as lists of individual lits * merge * Add support for Algebraic Datatypes in JavaScript/TypeScript bindings (#7734) * Initial plan * Add datatype type definitions to types.ts (work in progress) Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Complete datatype type definitions with working TypeScript compilation Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Implement core datatype functionality with TypeScript compilation success Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Complete datatype implementation with full Context integration and tests 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> * chipping away at the new code structure --------- Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> Co-authored-by: Nikolaj Bjorner <nbjorner@microsoft.com> Co-authored-by: humnrdble <83878671+humnrdble@users.noreply.github.com> Co-authored-by: Nuno Lopes <nuno.lopes@tecnico.ulisboa.pt> Co-authored-by: Copilot <198982749+Copilot@users.noreply.github.com> Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * updates Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * comments * debug infinite recursion and split cubes on existing split atoms that aren't in the cube * share lemmas, learn from unsat core, try to debug a couple of things, there was a subtle bug that i have a hard time repro'ing * Parallel solving (#7759) * very basic setup * ensure solve_eqs is fully disabled when smt.solve_eqs=false, #7743 Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * respect smt configuration parameter in elim_unconstrained simplifier Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * indentation * add bash files for test runs * add option to selectively disable variable solving for only ground expressions Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * remove verbose output Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * fix #7745 axioms for len(substr(...)) escaped due to nested rewriting * ensure atomic constraints are processed by arithmetic solver * #7739 optimization add simplification rule for at(x, offset) = "" Introducing j just postpones some rewrites that prevent useful simplifications. Z3 already uses common sub-expressions. The example highlights some opportunities for simplification, noteworthy at(..) = "". The example is solved in both versions after adding this simplification. * fix unsound len(substr) axiom Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * FreshConst is_sort (#7748) * #7750 add pre-processing simplification * Add parameter validation for selected API functions * updates to ac-plugin fix incrementality bugs by allowing destructive updates during saturation at the cost of redoing saturation after a pop. * enable passive, add check for bloom up-to-date * add top-k fixed-sized min-heap priority queue for top scoring literals * set up worker thread batch manager for multithreaded batch cubes paradigm, need to debug as I am getting segfault still * fix bug in parallel solving batch setup * fix bug * allow for internalize implies * disable pre-processing during cubing * debugging * remove default constructor * remove a bunch of string copies * Update euf_ac_plugin.cpp include reduction rules in forward simplification * Update euf_completion.cpp try out restricting scope of equalities added by instantation * Update smt_parallel.cpp Drop non-relevant units from shared structures. * process cubes as lists of individual lits * merge * Add support for Algebraic Datatypes in JavaScript/TypeScript bindings (#7734) * Initial plan * Add datatype type definitions to types.ts (work in progress) Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Complete datatype type definitions with working TypeScript compilation Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Implement core datatype functionality with TypeScript compilation success Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Complete datatype implementation with full Context integration and tests 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> * chipping away at the new code structure * comments * debug infinite recursion and split cubes on existing split atoms that aren't in the cube * share lemmas, learn from unsat core, try to debug a couple of things, there was a subtle bug that i have a hard time repro'ing --------- Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> Co-authored-by: Nikolaj Bjorner <nbjorner@microsoft.com> Co-authored-by: humnrdble <83878671+humnrdble@users.noreply.github.com> Co-authored-by: Nuno Lopes <nuno.lopes@tecnico.ulisboa.pt> Co-authored-by: Copilot <198982749+Copilot@users.noreply.github.com> Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * updates Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * simplify output Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * merge * resolve bad bug about l2g and g2l translators using wrong global context. add some debug prints * initial attempt at dynamically switching from greedy to frugal splitting strategy in return_cubes. need to test. also there is some bug where the threads take forever to cancel? * Parallel solving (#7769) * very basic setup * ensure solve_eqs is fully disabled when smt.solve_eqs=false, #7743 Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * respect smt configuration parameter in elim_unconstrained simplifier Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * indentation * add bash files for test runs * add option to selectively disable variable solving for only ground expressions Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * remove verbose output Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * fix #7745 axioms for len(substr(...)) escaped due to nested rewriting * ensure atomic constraints are processed by arithmetic solver * #7739 optimization add simplification rule for at(x, offset) = "" Introducing j just postpones some rewrites that prevent useful simplifications. Z3 already uses common sub-expressions. The example highlights some opportunities for simplification, noteworthy at(..) = "". The example is solved in both versions after adding this simplification. * fix unsound len(substr) axiom Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * FreshConst is_sort (#7748) * #7750 add pre-processing simplification * Add parameter validation for selected API functions * updates to ac-plugin fix incrementality bugs by allowing destructive updates during saturation at the cost of redoing saturation after a pop. * enable passive, add check for bloom up-to-date * add top-k fixed-sized min-heap priority queue for top scoring literals * set up worker thread batch manager for multithreaded batch cubes paradigm, need to debug as I am getting segfault still * fix bug in parallel solving batch setup * fix bug * allow for internalize implies * disable pre-processing during cubing * debugging * remove default constructor * remove a bunch of string copies * Update euf_ac_plugin.cpp include reduction rules in forward simplification * Update euf_completion.cpp try out restricting scope of equalities added by instantation * Update smt_parallel.cpp Drop non-relevant units from shared structures. * process cubes as lists of individual lits * merge * Add support for Algebraic Datatypes in JavaScript/TypeScript bindings (#7734) * Initial plan * Add datatype type definitions to types.ts (work in progress) Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Complete datatype type definitions with working TypeScript compilation Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Implement core datatype functionality with TypeScript compilation success Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Complete datatype implementation with full Context integration and tests 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> * chipping away at the new code structure * comments * debug infinite recursion and split cubes on existing split atoms that aren't in the cube * share lemmas, learn from unsat core, try to debug a couple of things, there was a subtle bug that i have a hard time repro'ing * merge * resolve bad bug about l2g and g2l translators using wrong global context. add some debug prints * initial attempt at dynamically switching from greedy to frugal splitting strategy in return_cubes. need to test. also there is some bug where the threads take forever to cancel? --------- Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> Co-authored-by: Nikolaj Bjorner <nbjorner@microsoft.com> Co-authored-by: humnrdble <83878671+humnrdble@users.noreply.github.com> Co-authored-by: Nuno Lopes <nuno.lopes@tecnico.ulisboa.pt> Co-authored-by: Copilot <198982749+Copilot@users.noreply.github.com> Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * resolve bug about not translating managers correctly for the second phase of the greedy cubing, and the frugal fallback * Parallel solving (#7771) * very basic setup * ensure solve_eqs is fully disabled when smt.solve_eqs=false, #7743 Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * respect smt configuration parameter in elim_unconstrained simplifier Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * indentation * add bash files for test runs * add option to selectively disable variable solving for only ground expressions Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * remove verbose output Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * fix #7745 axioms for len(substr(...)) escaped due to nested rewriting * ensure atomic constraints are processed by arithmetic solver * #7739 optimization add simplification rule for at(x, offset) = "" Introducing j just postpones some rewrites that prevent useful simplifications. Z3 already uses common sub-expressions. The example highlights some opportunities for simplification, noteworthy at(..) = "". The example is solved in both versions after adding this simplification. * fix unsound len(substr) axiom Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * FreshConst is_sort (#7748) * #7750 add pre-processing simplification * Add parameter validation for selected API functions * updates to ac-plugin fix incrementality bugs by allowing destructive updates during saturation at the cost of redoing saturation after a pop. * enable passive, add check for bloom up-to-date * add top-k fixed-sized min-heap priority queue for top scoring literals * set up worker thread batch manager for multithreaded batch cubes paradigm, need to debug as I am getting segfault still * fix bug in parallel solving batch setup * fix bug * allow for internalize implies * disable pre-processing during cubing * debugging * remove default constructor * remove a bunch of string copies * Update euf_ac_plugin.cpp include reduction rules in forward simplification * Update euf_completion.cpp try out restricting scope of equalities added by instantation * Update smt_parallel.cpp Drop non-relevant units from shared structures. * process cubes as lists of individual lits * merge * Add support for Algebraic Datatypes in JavaScript/TypeScript bindings (#7734) * Initial plan * Add datatype type definitions to types.ts (work in progress) Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Complete datatype type definitions with working TypeScript compilation Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Implement core datatype functionality with TypeScript compilation success Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Complete datatype implementation with full Context integration and tests 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> * chipping away at the new code structure * comments * debug infinite recursion and split cubes on existing split atoms that aren't in the cube * share lemmas, learn from unsat core, try to debug a couple of things, there was a subtle bug that i have a hard time repro'ing * merge * fix #7603: race condition in Ctrl-C handling (#7755) * fix #7603: race condition in Ctrl-C handling * fix race in cancel_eh * fix build * add arithemtic saturation * add an option to register callback on quantifier instantiation Suppose a user propagator encodes axioms using quantifiers and uses E-matching for instantiation. If it wants to implement a custom priority scheme or drop some instances based on internal checks it can register a callback with quantifier instantiation * missing new closure Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * add Z3_solver_propagate_on_binding to ml callback declarations Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * add python file Signed-off-by: Lev Nachmanson <levnach@Levs-MacBook-Pro.local> * debug under defined calls Signed-off-by: Lev Nachmanson <levnach@hotmail.com> * more untangle params Signed-off-by: Lev Nachmanson <levnach@hotmail.com> * precalc parameters to define the eval order Signed-off-by: Lev Nachmanson <levnach@hotmail.com> * remove a printout Signed-off-by: Lev Nachmanson <levnach@hotmail.com> * rename a Python file Signed-off-by: Lev Nachmanson <levnach@hotmail.com> * add on_binding callbacks across APIs update release notes, add to Java, .Net, C++ * use jboolean in Native interface Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * register on_binding attribute Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * fix java build for java bindings Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * avoid interferring side-effects in function calls Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * remove theory_str and classes that are only used by it * remove automata from python build Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * remove ref to theory_str Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * get the finest factorizations before project Signed-off-by: Lev Nachmanson <levnach@hotmail.com> * rename add_lcs to add_lc Signed-off-by: Lev Nachmanson <levnach@hotmail.com> * resolve bad bug about l2g and g2l translators using wrong global context. add some debug prints * initial attempt at dynamically switching from greedy to frugal splitting strategy in return_cubes. need to test. also there is some bug where the threads take forever to cancel? * Update RELEASE_NOTES.md * resolve bug about not translating managers correctly for the second phase of the greedy cubing, and the frugal fallback --------- Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> Signed-off-by: Lev Nachmanson <levnach@Levs-MacBook-Pro.local> Signed-off-by: Lev Nachmanson <levnach@hotmail.com> Co-authored-by: Nikolaj Bjorner <nbjorner@microsoft.com> Co-authored-by: humnrdble <83878671+humnrdble@users.noreply.github.com> Co-authored-by: Nuno Lopes <nuno.lopes@tecnico.ulisboa.pt> Co-authored-by: Copilot <198982749+Copilot@users.noreply.github.com> Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> Co-authored-by: Lev Nachmanson <levnach@hotmail.com> * code and notes * add some debug prints and impelement internal polynomial fix * add some comments and debug m_assumptions_used * redo greedy->frugal strategy so we don't split on existing cubes in frugal at all (eliminate the incorrect/wasteful step by processing current batch first) * set up initial scaffolding for sharing clauses between threads and batch manager. needs some reworking/debug still * Parallel solving (#7774) * very basic setup * ensure solve_eqs is fully disabled when smt.solve_eqs=false, #7743 Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * respect smt configuration parameter in elim_unconstrained simplifier Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * indentation * add bash files for test runs * add option to selectively disable variable solving for only ground expressions Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * remove verbose output Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * fix #7745 axioms for len(substr(...)) escaped due to nested rewriting * ensure atomic constraints are processed by arithmetic solver * #7739 optimization add simplification rule for at(x, offset) = "" Introducing j just postpones some rewrites that prevent useful simplifications. Z3 already uses common sub-expressions. The example highlights some opportunities for simplification, noteworthy at(..) = "". The example is solved in both versions after adding this simplification. * fix unsound len(substr) axiom Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * FreshConst is_sort (#7748) * #7750 add pre-processing simplification * Add parameter validation for selected API functions * updates to ac-plugin fix incrementality bugs by allowing destructive updates during saturation at the cost of redoing saturation after a pop. * enable passive, add check for bloom up-to-date * add top-k fixed-sized min-heap priority queue for top scoring literals * set up worker thread batch manager for multithreaded batch cubes paradigm, need to debug as I am getting segfault still * fix bug in parallel solving batch setup * fix bug * allow for internalize implies * disable pre-processing during cubing * debugging * remove default constructor * remove a bunch of string copies * Update euf_ac_plugin.cpp include reduction rules in forward simplification * Update euf_completion.cpp try out restricting scope of equalities added by instantation * Update smt_parallel.cpp Drop non-relevant units from shared structures. * process cubes as lists of individual lits * merge * Add support for Algebraic Datatypes in JavaScript/TypeScript bindings (#7734) * Initial plan * Add datatype type definitions to types.ts (work in progress) Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Complete datatype type definitions with working TypeScript compilation Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Implement core datatype functionality with TypeScript compilation success Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Complete datatype implementation with full Context integration and tests 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> * chipping away at the new code structure * comments * debug infinite recursion and split cubes on existing split atoms that aren't in the cube * share lemmas, learn from unsat core, try to debug a couple of things, there was a subtle bug that i have a hard time repro'ing * merge * fix #7603: race condition in Ctrl-C handling (#7755) * fix #7603: race condition in Ctrl-C handling * fix race in cancel_eh * fix build * add arithemtic saturation * add an option to register callback on quantifier instantiation Suppose a user propagator encodes axioms using quantifiers and uses E-matching for instantiation. If it wants to implement a custom priority scheme or drop some instances based on internal checks it can register a callback with quantifier instantiation * missing new closure Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * add Z3_solver_propagate_on_binding to ml callback declarations Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * add python file Signed-off-by: Lev Nachmanson <levnach@Levs-MacBook-Pro.local> * debug under defined calls Signed-off-by: Lev Nachmanson <levnach@hotmail.com> * more untangle params Signed-off-by: Lev Nachmanson <levnach@hotmail.com> * precalc parameters to define the eval order Signed-off-by: Lev Nachmanson <levnach@hotmail.com> * remove a printout Signed-off-by: Lev Nachmanson <levnach@hotmail.com> * rename a Python file Signed-off-by: Lev Nachmanson <levnach@hotmail.com> * add on_binding callbacks across APIs update release notes, add to Java, .Net, C++ * use jboolean in Native interface Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * register on_binding attribute Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * fix java build for java bindings Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * avoid interferring side-effects in function calls Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * remove theory_str and classes that are only used by it * remove automata from python build Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * remove ref to theory_str Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * get the finest factorizations before project Signed-off-by: Lev Nachmanson <levnach@hotmail.com> * rename add_lcs to add_lc Signed-off-by: Lev Nachmanson <levnach@hotmail.com> * resolve bad bug about l2g and g2l translators using wrong global context. add some debug prints * initial attempt at dynamically switching from greedy to frugal splitting strategy in return_cubes. need to test. also there is some bug where the threads take forever to cancel? * Update RELEASE_NOTES.md * resolve bug about not translating managers correctly for the second phase of the greedy cubing, and the frugal fallback * remove unused square-free check Signed-off-by: Lev Nachmanson <levnach@hotmail.com> * add some debug prints and impelement internal polynomial fix * add some comments and debug m_assumptions_used * redo greedy->frugal strategy so we don't split on existing cubes in frugal at all (eliminate the incorrect/wasteful step by processing current batch first) * set up initial scaffolding for sharing clauses between threads and batch manager. needs some reworking/debug still --------- Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> Signed-off-by: Lev Nachmanson <levnach@Levs-MacBook-Pro.local> Signed-off-by: Lev Nachmanson <levnach@hotmail.com> Co-authored-by: Nikolaj Bjorner <nbjorner@microsoft.com> Co-authored-by: humnrdble <83878671+humnrdble@users.noreply.github.com> Co-authored-by: Nuno Lopes <nuno.lopes@tecnico.ulisboa.pt> Co-authored-by: Copilot <198982749+Copilot@users.noreply.github.com> Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> Co-authored-by: Lev Nachmanson <levnach@hotmail.com> * sign of life Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * add notes on parameter tuning Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * add notes on parameter tuning Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * add notes on parameter tuning Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * add notes on parameter tuning Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * turn off logging at level 0 for testing * add max thread conflicts backoff --------- Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> Signed-off-by: Lev Nachmanson <levnach@Levs-MacBook-Pro.local> Signed-off-by: Lev Nachmanson <levnach@hotmail.com> Co-authored-by: Nikolaj Bjorner <nbjorner@microsoft.com> Co-authored-by: humnrdble <83878671+humnrdble@users.noreply.github.com> Co-authored-by: Nuno Lopes <nuno.lopes@tecnico.ulisboa.pt> Co-authored-by: Copilot <198982749+Copilot@users.noreply.github.com> Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> Co-authored-by: Lev Nachmanson <levnach@hotmail.com> * fix #7776 * add > operator as shorthand for Array * updates to euf completion * resolve bug about not popping local ctx to base level before collecting shared lits * Add virtual translate method to solver_factory class (#7780) * Initial plan * Add virtual translate method to solver_factory base class and all implementations Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Add documentation for the translate method in solver_factory 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> * put return_cubes under lock * Revert "resolve bug about not popping local ctx to base level before collecting shared lits" This reverts commitbba1111e1b. * Update seq_rewriter.cpp * fix releaseNotesSource to inline Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * Use solver factory translate method in Z3_solver_translate (#7782) * Initial plan * Fix Z3_solver_translate to use solver factory translate method 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> * Revert "Parallel solving (#7775)" (#7777) This reverts commitc8e866f568. * remove upload artifact for azure-pipeline Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * Fix compilation warning: add missing is_passive_eq case to switch statement (#7785) * Initial plan * Fix compilation warning: add missing is_passive_eq case to switch statement 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> * Remove NugetPublishNightly stage from nightly.yaml (#7787) * Initial plan * Remove NugetPublishNightly stage from nightly.yaml 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> * add more params * enable pypi public Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * Fix nullptr dereference in pp_symbol when handling null symbol names (#7790) * Initial plan * Fix nullptr dereference in pp_symbol with null symbol names 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> * add option to control epsilon #7791 #7791 reports on using model values during lex optimization that break soft constraints. This is an artifact of using optimization where optimal values can be arbitrarily close to a rational. In a way it is by design, but we give the user now an option to control the starting point for epsilon when converting infinitesimals into rationals. * update on euf * check for internalized in solve_for * fix #7792 add missing revert operations * update version number to 4.15.4 Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * fix #7753 * fix #7796 Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * Create centralized version management with VERSION.txt (#7802) * Initial plan * Create VERSION.txt and update CMakeLists.txt to read version from file Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Complete centralized version management system Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Fix version update script and finalize implementation Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Create centralized version management with VERSION.txt 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> * read version from VERSION.txt Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * fix version parse Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * fix parsing of version Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * add param tuning experiment in python * Fix Azure Pipeline PyPI package builds by including VERSION.txt in source distribution (#7808) * Initial plan * Fix Azure Pipeline PyPI package builds by including VERSION.txt in source distribution 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> * Update nightly.yaml to match release.yml NuGet tool installer changes (#7810) * Initial plan * Update nightly.yaml to match release.yml NuGet tool installer changes 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> * Attempt at adding the README to the NuGet package (#7807) * Attempt at adding README to NuGet package * Forgot to enable publishing * add resources Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * remove resources directive again Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * Document how to use system-installed Z3 with CMake projects (#7809) * Initial plan * Add documentation for using system-installed Z3 with CMake 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> * Fix Julia bindings linker errors on Windows MSVC (#7794) * Initial plan * Fix Julia bindings linker errors on Windows MSVC Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Complete Julia bindings fix validation and testing Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> * Fix Julia bindings linker errors on Windows MSVC 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> * add print for version file Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * add more logging to setup.py Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * try diferennt dirs Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * try src_dir_repo Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * try other dir Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * remove extra characters Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * more output Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * print dirs Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * copy VERSION from SRC_DIR Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * Move VERSION.txt to scripts directory and update all references (#7811) * Initial plan * Move VERSION.txt to scripts/ and update all references 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> * clean up a little of the handling of VERSION.txt Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * add implementation and toggleable param for splitting frugal + choosing deepest cubes only * remove priority queue for top-k lits and replace with simple linear scan. the PQ implementation backend still remains in case we want to switch back * Add new configurations for SMT parallel settings * Bugfix: post-build sanity check when an old version of ocaml-z3 is installed (#7815) * fix: add generating META for ocamlfind. * Patch macos. We need to keep the `@rpath` and use environment var to enable the test because we need to leave it to be fixed by package managers. * Trigger CI. * Debug. * Debug. * Debug. * Debug. * Debug. * Debug. * Hacky fix for ocaml building warning. * Fix typo and rename variables. * Fix cmake for ocaml test, using local libz3 explicit. * Rename configuration from 'shareconflicts' to 'depthsplitting' * Fix configuration for depth splitting in notes * rename variables * remove double tweak versioning Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> * attempting to add backbone code, it does not work. still debugging the error: ASSERTION VIOLATION File: /home/t-ilshapiro/z3/src/ast/ast.cpp Line: 388 UNEXPECTED CODE WAS REACHED. I left a comment on the line where it's crashing * depth splitting now applies to greedy+frugal unless specified otherwise * debug the backbone crash (it was references not being counted) * iterative deepening experiment (no PQ yet). the hardness heuristic is still naive! * fix iterative deepening bug: unsolved cube needs to get re-enqueued even if we don't split it further * debug iterative deepening some more and add first attempt at PQ (untested) * fix some bugs and the PQ approach is working for now. the depth sets approach is actually unsound, but I am going to focus on the PQ approach for now since it has more potential for SAT problems with the right hardness metric * add new attempt at hardness function * attempt to add different hardness functions including heule schur and march, need to re-examine/debug/evaluate * implement march and heule schur hardness functions based on sat_lookahead.cpp implementations. they seem to be buggy, need to revisit. also set up experimental params for running on polytest * add a lot of debug prints that need to be removed. some bugs are resolved but others remain * debug in progress * remove the incorrect preselection functions for march and heule-schur. update explicit-hardness with bugfixes. now it works but i am not sure there is a good perf increase based on my handpicked examples. i tried several variations of hardness ratios as you can see commented out. there are debug prints still commented out. also return_cubes now takes in a single cube instead of a list C_worker to align with the single-cube hardness/should_split metrics, it doesn't change anything bc we only pass in 1 cube to begin with * debug a couple of things and change the hardness function * tree version in progress * cube tree data structure version for sharing maximal solver context. it compiles but segfaults immediately so it needs a lot of debugging * slowly debugging (committing for saving progress) * debugged get_next_cube to align with how we're storing the prev_cube and active status. other things are still wrong * debug manager translation problem * don't actually prune tree for UNSAT, too risky with multithreads. instead marks all such 'removed' nodes as inactive * it runs! but then crashes after a while * add optimization (or what is hopefully an optimization) about threads only working in the frontier determined by their split atoms, and only overlapping frontiers once they've exhausted their own * debug some things. exhausting the cube tree should return unsat now, not crash. also fix a couple of things in get_next_cube. also, setting k=1 for split atoms might be better, not sure * fix small bug about making sure we're searching in the thread's frontier when we just started a new frontier (like if we're at the empty cube at the root) --------- Signed-off-by: Nikolaj Bjorner <nbjorner@microsoft.com> Signed-off-by: Lev Nachmanson <levnach@Levs-MacBook-Pro.local> Signed-off-by: Lev Nachmanson <levnach@hotmail.com> Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: Nikolaj Bjorner <nbjorner@microsoft.com> Co-authored-by: humnrdble <83878671+humnrdble@users.noreply.github.com> Co-authored-by: Nuno Lopes <nuno.lopes@tecnico.ulisboa.pt> Co-authored-by: Copilot <198982749+Copilot@users.noreply.github.com> Co-authored-by: NikolajBjorner <3085284+NikolajBjorner@users.noreply.github.com> Co-authored-by: Lev Nachmanson <levnach@hotmail.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: Solal Pirelli <SolalPirelli@users.noreply.github.com> Co-authored-by: Shiwei Weng 翁士伟 <arbipher@gmail.com>
This commit is contained in:
Ilana Shapiro
GitHub
parent
e3ef4fa3f5
commit
de900d4745
3 changed files with 87 additions and 45 deletions
|
|
@ -106,8 +106,7 @@ public:
|
|||
// get closest cube to current by getting a random sibling of current (if current was UNSAT and we removed it from the tree)
|
||||
// or by descending randomly to a leaf (if we split the current node) to get the newest cube split fromthe current
|
||||
// we descend randomly to a leaf instead of just taking a random child because it's possible another thread made more descendants
|
||||
CubeNode* get_next_cube(CubeNode* current) {
|
||||
|
||||
CubeNode* get_next_cube(CubeNode* current, std::vector<CubeNode*>& frontier_roots) {
|
||||
IF_VERBOSE(1, verbose_stream() << "CubeTree: current cube is null: " << (current == nullptr) << "\n");
|
||||
if (!current) return nullptr;
|
||||
|
||||
|
|
@ -115,44 +114,66 @@ public:
|
|||
|
||||
// lambda to find any active leaf in the subtree (explore all branches)
|
||||
std::function<CubeNode*(CubeNode*)> find_active_leaf = [&](CubeNode* node) -> CubeNode* {
|
||||
if (!node || !node->active) return nullptr;
|
||||
if (node->is_leaf()) return node;
|
||||
if (!node) return nullptr;
|
||||
if (node->is_leaf() && node->active) return node;
|
||||
for (CubeNode* child : node->children) {
|
||||
CubeNode* leaf = find_active_leaf(child);
|
||||
if (leaf) return leaf;
|
||||
CubeNode* active_leaf = find_active_leaf(child);
|
||||
if (active_leaf) return active_leaf;
|
||||
}
|
||||
return nullptr;
|
||||
};
|
||||
|
||||
CubeNode* node = current;
|
||||
std::vector<CubeNode*> remaining_frontier_roots = frontier_roots;
|
||||
bool is_unexplored_frontier = frontier_roots.size() > 0 && current->cube.size() < frontier_roots[0]->cube.size(); // i.e. current is above the frontier (which always happens when we start with the empty cube!!)
|
||||
IF_VERBOSE(1, verbose_stream() << "CubeTree: current cube is " << (is_unexplored_frontier ? "above" : "within") << " the frontier\n");
|
||||
|
||||
// if current is above the frontier, start searching from the first frontier root
|
||||
if (is_unexplored_frontier && !frontier_roots.empty()) {
|
||||
IF_VERBOSE(1, verbose_stream() << "CubeTree: starting search from first frontier root\n");
|
||||
node = frontier_roots[0];
|
||||
}
|
||||
|
||||
while (node) {
|
||||
// 1. check if current node itself is active leaf
|
||||
if (node->active && node->is_leaf()) return node;
|
||||
|
||||
// 2. check active leaf descendants
|
||||
// check active leaf descendants
|
||||
CubeNode* leaf_descendant = nullptr;
|
||||
for (CubeNode* child : node->children) {
|
||||
leaf_descendant = find_active_leaf(child);
|
||||
if (leaf_descendant) return leaf_descendant;
|
||||
}
|
||||
leaf_descendant = find_active_leaf(node);
|
||||
if (leaf_descendant) return leaf_descendant;
|
||||
|
||||
// 3 & 4. check siblings and their active leaf descendants
|
||||
if (node->parent) {
|
||||
CubeNode* parent = node->parent;
|
||||
for (CubeNode* sibling : parent->children) {
|
||||
if (sibling == node) continue;
|
||||
// DO NOT NEED to check siblings and their active leaf descendants
|
||||
// since this is handled by the recusion up the tree!!
|
||||
// and checking siblings here is unsafe if we adhere to thread frontier optimizations
|
||||
|
||||
// check if sibling itself is an active leaf
|
||||
if (sibling->active && sibling->is_leaf()) return sibling;
|
||||
// see if we're at a boundary of the frontier (i.e. we hit one of the frontier roots)
|
||||
auto it = std::find(remaining_frontier_roots.begin(), remaining_frontier_roots.end(), node);
|
||||
// get the index of the node in remaining_frontier_roots
|
||||
unsigned curr_root_idx = std::distance(remaining_frontier_roots.begin(), it);
|
||||
if (it != remaining_frontier_roots.end()) { // i.e. the node is in the list of remaining_frontier_roots
|
||||
IF_VERBOSE(1, verbose_stream() << "CubeTree: hit frontier root " << node << "\n");
|
||||
|
||||
// check for active leaf descendants of sibling
|
||||
CubeNode* leaf_in_sibling = find_active_leaf(sibling);
|
||||
if (leaf_in_sibling) return leaf_in_sibling;
|
||||
if (!remaining_frontier_roots.empty()) {
|
||||
IF_VERBOSE(1, verbose_stream() << "CubeTree: picking next frontier root to search from.\n");
|
||||
// pick the next frontier root (wrap around if at end)
|
||||
// we do this so we either move onto the next split atom in the frontier (if we just processed neg(atom))
|
||||
// or we get the negation of the atom we just processed (if we just processed pos(atom))
|
||||
// since the other the splits are added is [pos, neg, ...] for each split atom
|
||||
node = remaining_frontier_roots[curr_root_idx + 1 < remaining_frontier_roots.size() ? curr_root_idx + 1 : 0];
|
||||
|
||||
// Remove exhausted frontier root
|
||||
remaining_frontier_roots.erase(it);
|
||||
} else {
|
||||
IF_VERBOSE(1, verbose_stream() << "CubeTree: resetting frontier after exhausting\n");
|
||||
// Frontier exhausted: reset frontier_roots for next iteration
|
||||
frontier_roots.clear();
|
||||
|
||||
// Start "global" search from current node
|
||||
node = node->parent;
|
||||
}
|
||||
|
||||
continue;
|
||||
}
|
||||
|
||||
// 5. climb up to parent
|
||||
// Move up in the current frontier
|
||||
node = node->parent;
|
||||
}
|
||||
|
||||
|
|
|
|||
|
|
@ -104,8 +104,12 @@ namespace smt {
|
|||
LOG_WORKER(1, " Curr cube node is null: " << (m_curr_cube_node == nullptr) << "\n");
|
||||
if (m_config.m_cubetree) {
|
||||
// use std::tie so we don't overshadow cube_node!!!
|
||||
std::tie(cube_node, cube) = b.get_cube_from_tree(m_g2l, m_curr_cube_node); // cube node is the reference to the node in the tree, tells us how to get the next cube. "cube" is the translated cube we need for the solver
|
||||
std::tie(cube_node, cube) = b.get_cube_from_tree(m_g2l, frontier_roots, m_curr_cube_node); // cube node is the reference to the node in the tree, tells us how to get the next cube. "cube" is the translated cube we need for the solver
|
||||
LOG_WORKER(1, " Got cube node from CubeTree. Is null: " << (cube_node == nullptr) << "\n");
|
||||
if (!cube_node) { // i.e. no more cubes
|
||||
LOG_WORKER(1, " No more cubes from CubeTree, exiting\n");
|
||||
return;
|
||||
}
|
||||
m_curr_cube_node = cube_node; // store the current cube so we know how to get the next closest cube from the tree
|
||||
IF_VERBOSE(1, verbose_stream() << " Worker " << id << " got cube of size " << cube.size() << " from CubeTree\n");
|
||||
} else {
|
||||
|
|
@ -159,8 +163,7 @@ namespace smt {
|
|||
b.return_cubes(m_l2g, cube, split_atoms, should_split, cube_hardness);
|
||||
} else if (m_config.m_cubetree) {
|
||||
IF_VERBOSE(1, verbose_stream() << " returning undef cube to CubeTree. Cube node is null: " << (cube_node == nullptr) << "\n");
|
||||
|
||||
b.return_cubes_tree(m_l2g, cube_node, split_atoms);
|
||||
b.return_cubes_tree(m_l2g, cube_node, split_atoms, frontier_roots);
|
||||
} else {
|
||||
b.return_cubes(m_l2g, cube, split_atoms);
|
||||
}
|
||||
|
|
@ -382,7 +385,7 @@ namespace smt {
|
|||
return r;
|
||||
}
|
||||
|
||||
std::pair<CubeNode*, expr_ref_vector> parallel::batch_manager::get_cube_from_tree(ast_translation& g2l, CubeNode* prev_cube) {
|
||||
std::pair<CubeNode*, expr_ref_vector> parallel::batch_manager::get_cube_from_tree(ast_translation& g2l, std::vector<CubeNode*>& frontier_roots, CubeNode* prev_cube) {
|
||||
std::scoped_lock lock(mux);
|
||||
expr_ref_vector l_cube(g2l.to());
|
||||
SASSERT(m_config.m_cubetree);
|
||||
|
|
@ -402,10 +405,15 @@ namespace smt {
|
|||
|
||||
// get a cube from the CubeTree
|
||||
SASSERT(!m_cubes_tree.empty());
|
||||
CubeNode* next_cube_node = m_cubes_tree.get_next_cube(prev_cube); // get the next cube in the tree closest to the prev cube (i.e. longest common prefix)
|
||||
CubeNode* next_cube_node = m_cubes_tree.get_next_cube(prev_cube, frontier_roots); // get the next cube in the tree closest to the prev cube (i.e. longest common prefix)
|
||||
|
||||
IF_VERBOSE(1, verbose_stream() << "Batch manager giving out cube from CubeTree. Is null: " << (next_cube_node==nullptr) << "\n");
|
||||
|
||||
if (!next_cube_node) { // i.e. no more cubes
|
||||
IF_VERBOSE(1, verbose_stream() << " No more cubes from CubeTree, exiting\n");
|
||||
return {nullptr, l_cube}; // return nullptr and empty cube
|
||||
}
|
||||
|
||||
for (auto& e : next_cube_node->cube) {
|
||||
l_cube.push_back(g2l(e));
|
||||
}
|
||||
|
|
@ -781,12 +789,14 @@ namespace smt {
|
|||
}
|
||||
}
|
||||
|
||||
void parallel::batch_manager::return_cubes_tree(ast_translation& l2g, CubeNode* cube_node, expr_ref_vector const& A_worker) {
|
||||
|
||||
void parallel::batch_manager::return_cubes_tree(ast_translation& l2g, CubeNode* cube_node, expr_ref_vector const& A_worker, std::vector<CubeNode*>& frontier_roots) {
|
||||
IF_VERBOSE(1, verbose_stream() << " Returning cube to batch manager's cube tree.\n");
|
||||
expr_ref_vector const& c = cube_node->cube;
|
||||
expr_ref_vector const& l_cube = cube_node->cube;
|
||||
IF_VERBOSE(1, verbose_stream() << " Cube node null: " << (cube_node == nullptr) << "\n");
|
||||
IF_VERBOSE(1, verbose_stream() << " PROCESSING CUBE of size: " << c.size() << "\n");
|
||||
IF_VERBOSE(1, verbose_stream() << " PROCESSING CUBE of size: " << l_cube.size() << "\n");
|
||||
|
||||
bool is_new_frontier = frontier_roots.empty(); // need to save this as a bool here, bc otherwise the frontier stops being populated after a single split atom
|
||||
IF_VERBOSE(1, verbose_stream() << " Is new frontier: " << is_new_frontier << "\n");
|
||||
|
||||
auto atom_in_cube = [&](expr_ref_vector const& cube, expr* atom) {
|
||||
return any_of(cube, [&](expr* e) { return e == atom || (m.is_not(e, e) && e == atom); });
|
||||
|
|
@ -797,19 +807,29 @@ namespace smt {
|
|||
auto add_split_atom_tree = [&](expr* atom) {
|
||||
IF_VERBOSE(1, verbose_stream() << " Adding split atom to tree: " << mk_bounded_pp(atom, m, 3) << "\n");
|
||||
expr_ref_vector g_cube(l2g.to());
|
||||
for (auto& atom : c)
|
||||
for (auto& atom : l_cube)
|
||||
g_cube.push_back(l2g(atom));
|
||||
|
||||
// Positive atom branch
|
||||
expr_ref_vector cube_pos = g_cube;
|
||||
cube_pos.push_back(atom);
|
||||
expr_ref_vector g_cube_pos = g_cube;
|
||||
g_cube_pos.push_back(atom);
|
||||
|
||||
// Negative atom branch
|
||||
expr_ref_vector cube_neg = g_cube;
|
||||
cube_neg.push_back(m.mk_not(atom));
|
||||
expr_ref_vector g_cube_neg = g_cube;
|
||||
g_cube_neg.push_back(m.mk_not(atom));
|
||||
|
||||
m_cubes_tree.add_children(cube_node, g_cube_pos, g_cube_neg); // default is active
|
||||
if (is_new_frontier) { // note: calling frontier_roots.empty() here would always return false after adding the first split atom
|
||||
// add cube_pos and cube_neg to frontier roots
|
||||
CubeNode* cube_node_pos = new CubeNode(*cube_node); // deep copy of the node
|
||||
CubeNode* cube_node_neg = new CubeNode(*cube_node); // deep copy of the node
|
||||
cube_node_pos->cube.push_back(atom); // modify cube
|
||||
cube_node_neg->cube.push_back(m.mk_not(atom)); // modify cube
|
||||
|
||||
m_cubes_tree.add_children(cube_node, cube_pos, cube_neg); // default is active
|
||||
frontier_roots.push_back(cube_node_pos);
|
||||
frontier_roots.push_back(cube_node_neg);
|
||||
IF_VERBOSE(1, verbose_stream() << " Added split to frontier roots. Size now: " << frontier_roots.size() << "\n");
|
||||
}
|
||||
|
||||
m_stats.m_num_cubes += 2;
|
||||
m_stats.m_max_cube_depth = std::max(m_stats.m_max_cube_depth, g_cube.size() + 1);
|
||||
|
|
@ -817,8 +837,7 @@ namespace smt {
|
|||
|
||||
std::scoped_lock lock(mux);
|
||||
|
||||
|
||||
if (c.size() >= m_config.m_max_cube_depth) {
|
||||
if (l_cube.size() >= m_config.m_max_cube_depth) {
|
||||
IF_VERBOSE(1, verbose_stream() << " Skipping split of cube at max depth " << m_config.m_max_cube_depth << "\n";);
|
||||
cube_node->active = true; // mark the cube as active again since we didn't split it
|
||||
return;
|
||||
|
|
@ -832,7 +851,7 @@ namespace smt {
|
|||
if (!m_split_atoms.contains(g_atom))
|
||||
m_split_atoms.push_back(g_atom);
|
||||
|
||||
IF_VERBOSE(1, verbose_stream() << " splitting worker cubes on new atom for PQ " << mk_bounded_pp(g_atom, m, 3) << "\n");
|
||||
IF_VERBOSE(1, verbose_stream() << " splitting worker cube on new atom " << mk_bounded_pp(g_atom, m, 3) << "\n");
|
||||
add_split_atom_tree(g_atom);
|
||||
}
|
||||
}
|
||||
|
|
|
|||
|
|
@ -132,13 +132,13 @@ namespace smt {
|
|||
// The batch manager returns the next cube to
|
||||
//
|
||||
expr_ref_vector get_cube(ast_translation& g2l); // FOR ALL NON-TREE VERSIONS
|
||||
std::pair<CubeNode*, expr_ref_vector> get_cube_from_tree(ast_translation& g2l, CubeNode* prev_cube = nullptr);
|
||||
std::pair<CubeNode*, expr_ref_vector> get_cube_from_tree(ast_translation& g2l, std::vector<CubeNode*>& frontier_roots, CubeNode* prev_cube = nullptr);
|
||||
|
||||
//
|
||||
// worker threads return unprocessed cubes to the batch manager together with split literal candidates.
|
||||
// the batch manager re-enqueues unprocessed cubes and optionally splits them using the split_atoms returned by this and workers.
|
||||
//
|
||||
void return_cubes_tree(ast_translation& l2g, CubeNode* cube, expr_ref_vector const& split_atoms);
|
||||
void return_cubes_tree(ast_translation& l2g, CubeNode* cube, expr_ref_vector const& split_atoms, std::vector<CubeNode*>& frontier_roots);
|
||||
// FOR ALL NON-TREE VERSIONS
|
||||
void return_cubes(ast_translation& l2g, expr_ref_vector const& cube, expr_ref_vector const& split_atoms, const bool should_split=true, const double hardness=1.0);
|
||||
void report_assumption_used(ast_translation& l2g, expr* assumption);
|
||||
|
|
@ -198,6 +198,8 @@ namespace smt {
|
|||
scoped_ptr<context> ctx;
|
||||
ast_translation m_g2l, m_l2g;
|
||||
CubeNode* m_curr_cube_node = nullptr;
|
||||
std::vector<CubeNode*> frontier_roots;
|
||||
|
||||
unsigned m_num_shared_units = 0;
|
||||
unsigned m_num_initial_atoms = 0;
|
||||
unsigned m_shared_clause_limit = 0; // remembers the index into shared_clause_trail marking the boundary between "old" and "new" clauses to share
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue