Age | Commit message (Collapse) | Author |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
* improved proof production statistics
This commit creates new statistics which will help us understand what is
so expensive about proof production.
There are already statistics for:
* Net time spent building/printing the LFSC proof.
* Size of the LFSC proof.
This commit adds statistics for:
* The time cost of DRAT optimization:
* net time
* tool time (drat-trim)
* clause matching time (matching core clauses with input clauses)
* Non-trivial because drat-trim can (and does) dedup and reorder
literals
* The advantage of DRAT optimization (proof/formula size before/after)
* The time cost of DRAT translation [to LRAT or ER] (net time, tool time)
* The time cost of skeleton traversal
* The time cost of printing declatations
* The time cost of printing CNF proofs
* The time cost of printing theory lemmas
* The time cost of printing final UNSAT proof.
There is a method, toStream, which is responsible for much of proof
production. The whole method was timed, but subsections were not. The
timings above consider subsections of it.
I also wanted to better understand the cost of DRAT optimization and
translation.
* [BV Proof] Optimize DRAT optimization
tldr: I made a bad data-structure/algorithm choice when implementing
part of DRAT/CNF-optimization, which consumed **a lot** of time on some
bechmarks. This commit fixes that choice.
Long Version:
Set-Keyed Maps Considered Harmful
=================================
Algorithmic Problem
-------------------
The DRAT optimization process spits out a unsatifiable core of the CNF.
The clauses in the core are all from the original formula, but their
literals may have been reordered and deduplicated. We must match the
old clauses with new ones, so we know which old clauses are in the core.
Old (BAD) Solution
------------------
Before I didn't really think about this process very much. I built a
solution in which clauses were canonically represented by hash sets of
literals, and made a hash map from canonical clauses to clause indices
into the original CNF.
Problem With Old Solution
-------------------------
In hindsight this was a bad idea. First, it required a new hash set to
be heap-allocated for every clause in the CNF. Second, the map lookups
required set-hashes (reasonable -- linear time, once) and hash-set
equality (not reasonable -- quadratic time, multiple times) on every
lookup.
New Solution
------------
The ideal solution is probably to have the map from clauses to clause
ids be something like a trie. STL doesn't have a trie, but representing
clauses as sorted, deduped vectors of literal in a tree based on
lexicographical comparison is pretty closed to this. On randomly chosen
examples it seems to be a huge improvement over the old
map-keyed-by-sets solution, and I'm in the process of running a full set
of bechmarks.
Also, we store pointers to the clauses already stored elsewhere in the
proof, instead of allocating new memory for them.
Future Work
-----------
It may also be reasonable to do a hash map of sorted, deduped, vector
clauses. I haven't tried this, yet (there's a TODO in the code).
* Update src/proof/clausal_bitvector_proof.h
Thanks andres!
Co-Authored-By: Andres Noetzli <andres.noetzli@gmail.com>
* Respond to Andres' Review: better use of CodeTimer
* Removed commented code (Andres)
|
|
|
|
|
|
This commit enables DRAT-optimization, which consists of two sub-processes:
1. removing unnecessary instructions from DRAT-proofs and
2. not proving clauses which are not needed by DRAT proofs.
These changes have the effect of dramatically shortening some some bit-vector proofs. Specifically, proofs using lemmas in the ER, DRAT, and LRAT formats, since proofs in any of these formats are derived from a (now optimized!) DRAT proof produced by CryptoMiniSat. What follows is a description of the main parts of this PR:
## DRAT Optimization
The DRAT-optimization is done by `drat-trim`, which is bundled with `drat2er`. The (new) function `ClausalBitVectorProof::optimizeDratProof` is our interface to the optimization machinery, and most of the new logic in this PR is in that function.
## CNF Representation
The ability to not prove unused clauses requires a slight architectural change as well. In particular, we need to be able to describe **which** subset of the original clause set actually needs to be proved. To facilitate this, when the clause set for CryptoMiniSat is first formed it is represented as a (a) map from clause indices to clauses and (b) a list of indices. Then, when the CNF is optimized, we temporarily store a new list of the clauses in the optimized formula. This change in representation requires a number of small tweaks throughout the code.
## Small Fixes to Signatures
When we decided to check and accept two different kinds of DRAT, some of our DRAT-checking broke. In particular, when supporting one kind of DRAT, it is okay to `fail` (crash) when a proof fails to check. If you're supporting two kinds of DRAT, crashing in response to the first checker rejecting the proof denies the second checker an opportunity to check the proof. This PR tweaks the signatures slightly (and soundly!) to do something else instead of `fail`ing.
|
|
|
|
This is a minor fix for systems with glibc version < 2.17. In that case, we need to link with `-lrt` according to the clock_gettime man page.
|
|
|
|
|
|
This commit adds a statistic that records the total size of all proofs
generated by an instance of `SmtEngine`. The commit also moves
`SmtEngine::checkProof()` into `smt_engine.cpp` because it needs to know
the complete type of `d_stats` (and the separate file for that method
didn't seem that useful). Additionally, it changes
`smt::SmtEngine::checkProofTime` to `smt::SmtEngine::lfscCheckProofTime`
that only measures the time spent in LFSC and adds a statistic
`proof::ProofManager::proofProductionTime` that measures the proof
production time separately (also works with `get-proof`/`--dump-proof`).
|
|
This commit fixes make install, which previously copied all public header files to ${CMAKE_INSTALL_PREFIX}/ instead of ${CMAKE_INSTALL_PREFIX}/cvc4.
Further, the old build system modified all #include directives in the installed public header files to use the installed headers, e.g., #include "cvc4_public.h" was changed to #include <cvc4/cvc4_public.h>. Now, after make install the script src/fix-install-headers.sh is executed to change the #include directives accordingly (this should be obsolete with the new C++ API).
|
|
* ErProof class with LFSC output
* Created a TraceCheckProof class
* parsable from text
* Created an ErProof class
* constructible from a TraceCheckProof
* writable as LFSC
* A bunch of unit tests
* Reponded to Mathias's first set of comments.
Credits to Mathias for many of the fixes!
* Responed to Andres's first set, fixed tests
I accidentally deleted a "!" last time, causing stuff to fail.
* Use Configuration::isAssertionBuild
* Clarified comment
* Responded to Andres's 2nd review
* Gaurding against a memory error.
* Renaming a file.
* Aggressively unlinking temporary files.
|
|
Creating LRAT proofs reuqires writing SAT problems in the DIMACS format.
Before this code was in the LRAT class.
However, since creating ER proofs will also require writing DIMACS, I
decided to extract it.
At the same time I realized that my prior representation of used clauses
was unnecessarily poor. I had chosen it to align with
`CnfProof::collectAtomsForClauses`, but the format is really bad (it
requires extra allocations & manual memory management), and I discovered
that the aforementioned method is super simple, so I'm moving to a
better format.
|
|
* [DRAT] ClausalBitvectorProof
Created a class, `ClausalBitvectorProof`, which represents a bitvector
proof of UNSAT using an underlying clausal technique (DRAT, LRAT, etc)
It fits into the `BitvectorProof` class hierarchy like this:
```
BitvectorProof
/ \
/ \
ClausalBitvectorProof ResolutionBitvectorProof
```
This change is a painful one because all of the following BV subsystems
referenced ResolutionBitvectorProof (subsequently RBVP) or
BitvectorProof (subsequently BVP):
* CnfStream
* SatSolver (specifically the BvSatSolver)
* CnfProof
* TheoryProof
* TheoryBV
* Both bitblasters
And in particular, ResolutionBitvectorProof, the CnfStream, and the
SatSolvers were tightly coupled.
This means that references to and interactions with (R)BVP were
pervasive.
Nevertheless, an SMT developer must persist.
The change summary:
* Create a subclass of BVP, called ClausalBitvectorProof, which has
most methods stubbed out.
* Make a some modifications to BVP and ResolutionBitvectorProof as the
natural division of labor between the different classes becomes
clear.
* Go through all the components in the first list and try to figure
out which kind of BVP they should **actually** be interacting with,
and how. Make tweaks accordingly.
* Add a hook from CryptoMinisat which pipes the produced DRAT proof
into the new ClausalBitvectorProof.
* Add a debug statement to ClausalBitvectorProof which parses and
prints that DRAT proof, for testing purposes.
Test:
* `make check` to verify that we didn't break any old stuff, including
lazy BB, and eager BB when using bvminisat.
* `cvc4 --dump-proofs --bv-sat-solver=cryptominisat --bitblast=eager
-d bv::clausal test/regress/regress0/bv/ackermann2.smt2`, and see that
1. It crashed with "Unimplemented"
2. Right before that it prints out the (textual) DRAT proof.
* Remove 2 unneeded methods
* Missed a rename
* Typos
Thanks Andres!
Co-Authored-By: alex-ozdemir <aozdemir@hmc.edu>
* Address Andres comments
* Reorder members of TBitblaster
|
|
* Copied old DRAT data-structure files.
Next step: clean up the code, and adapt them to our current usage plans.
* Polished the DRAT class.
Notably, removed the idea of lazy-parsing, this is now just a DRAT
wrapper class.
More explicit about whether methods handle binary or text.
Better constructor patterns
* Added implementation of textual DRAT output
* reordered the DratInstruction structure.
* removed the public modifier from the above struct
* removed the operator << implementation for DratInstruction
* use emplace_back
* Addressing Yoni's first review
* Extracted "write literal in DIMACS format" idea as a function
* Replaced some spurious Debug streams with `os`. (they were left over
from an earlier refactor)
* Improved some documentation
* Removed aside about std::string
* Addressed Mathias' comments
Specifically
* SCREAMING_SNAKE_CASED enum variants.
* Extracted some common logic from two branches of a conditional.
* Cleaned out some undefined behavior from bit manipulation.
* Unit tests for binary DRAT parsing
* Added text output test
* s/white/black/ derp
|
|
* [LRAT] A C++ data structure for LRAT.
Added a data structure for storing (abstract) LRAT proofs.
The constructor will take a drat binary proof and convert it to LRAT
using drat-trim. However, this is unimplemented in this PR.
Subsequent PRs will add:
* LFSC representation of LRAT
* Bitvector Proofs based on LRAT
* Enabled tests for those proofs
* Documenting LRAT constructors
* Apply suggestions from code review
Co-Authored-By: alex-ozdemir <aozdemir@hmc.edu>
* Responding to Andres' review
Consisting of
* Naming nits
* Closed fds
* Better implementation of disjoint union for LratInstruction
* DRAT -> LRAT conversion is no longer an LratProof constructor
* include reorder
* Update src/proof/lrat/lrat_proof.h
Co-Authored-By: alex-ozdemir <aozdemir@hmc.edu>
* Addressed Andres' comments
* ANonymous namespaces and name resolution?
* Remove inlines, fix i negation
Thanks Andres!
* Use `std::abs`
Credit to Andres
Co-Authored-By: alex-ozdemir <aozdemir@hmc.edu>
* Remove uneeded public
|
|
drat2er is a C/C++ project which includes support for
* Checking DRAT proofs
* Converting DRAT proofs to LRAT proofs
* Converting DRAT proofs to ER proofs
It does the first 2 by using drat-trim under the hood.
I've modified our CMake configuration to allow drat2er to be linked into
CVC4, and I added a contrib script.
|
|
* [LRA Proof] Storage for LRA proofs
During LRA solving the `ConstraintDatabase` contains the reasoning
behind different constraints. Combinations of constraints are
periodically used to justify lemmas (conflict clauses, propegations, ...
?). `ConstraintDatabase` is SAT context-dependent.
ArithProofRecorder will be used to store concise representations of the
proof for each lemma raised by the (LR)A theory. The (LR)A theory will
write to it, and the ArithProof class will read from it to produce LFSC
proofs.
Right now, it's pretty simplistic -- it allows for only Farkas proofs.
In future PRs I'll:
1. add logic that stores proofs therein
2. add logic that retrieves and prints proofs
3. enable LRA proof production, checking, and testing
* Document ArithProofRecorder use-sites
* Update src/proof/arith_proof_recorder.cpp
Co-Authored-By: alex-ozdemir <aozdemir@hmc.edu>
* Yoni's review
* clang-format
* Response to Mathias' review.
|
|
|
|
* Split BitvectorProof into a sub/superclass
The superclass contains general printing knowledge.
The subclass contains CNF or Resolution-specific knowledge.
* Renames & code moves
* Nits cleaned in prep for PR
* Moved CNF-proof from ResolutionBitVectorProof to BitVectorProof
Since DRAT BV proofs will also contain a CNF-proof, the CNF proof should
be stored in `BitVectorProof`.
* Unique pointers, comments, and code movement.
Adjusted the distribution of code between BVP and RBVP.
Notably, put the CNF proof in BVP because it isn't
resolution-specific.
Added comments to the headers of both files -- mostly BVP.
Changed two owned pointers into unique_ptr.
BVP's pointer to a CNF proof
RBVP's pointer to a resolution proof
BVP: `BitVectorProof`
RBVP: `ResolutionBitVectorProof`
* clang-format
* Undo manual copyright modification
* s/superclass/base class/
Co-Authored-By: alex-ozdemir <aozdemir@hmc.edu>
* make LFSCBitVectorProof::printOwnedSort public
* Andres's Comments
Mostly cleaning up (or trying to clean up) includes.
* Cleaned up one header cycle
However, this only allowed me to move the forward-decl, not eliminate
it, because there were actually two underlying include cycles that the
forward-decl solved.
* Added single _s to header gaurds
* Fix Class name in debug output
Credits to Andres
Co-Authored-By: alex-ozdemir <aozdemir@hmc.edu>
* Reordered methods in BitVectorProof per original ordering
|
|
|
|
|
|
|
|
|
|
|
|
Previously, all util functions for the BV instantiator were static functions
in theory/quantifiers/cegqi/ceg_bv_instantiator.cpp. For the corresponding
unit test, it was therefore required to include this cpp file in order to
test these functions. This factors out these functions into a
theory/quantifiers/cegqi/ceg_bv_instantiator_utils.(cpp|h).
Asan reported errors for the corresponing unit test because of this.
|
|
Previously, all invertibility condition functions were static functions
in theory/quantifiers/bv_inverter.cpp. For the corresponding unit test,
it was therefore required to include this cpp file in order to test
these functions. This factors out these functions into a
theory/quantifiers/bv_inverter_utils.(cpp|h).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
The sources of all previous libraries are now added to libcvc4 and built as
libcvc4. This removes circular dependencies between libcvc4 and libexpr.
Further, we now only have one parser library and don't build additional
libraries for each language.
|
|
|
|
|
|
Further, print definitions added with add_definitions(...).
|
|
|
|
|
|
|
|
|
|
|
|
TODO: cvc4autoconfig.h
|