LydiaSyft is an open-source software framework for reasoning and synthesis of Linear Tempora Logic formulas interpreted on finite traces (LTLf), integrating efficient data structures and techniques focused on LTLf specifications.
This project can be used either as a standalone CLI tool or as a C++ library that integrates with other projects. Among the implemented Data Structures and algorithms, we have:
- DFA representation and manipulation:
- LTLf synthesis settings:
Currently, the system has been tested on Ubuntu 24.04 LTS, and should work on other Linux systems. We plan to fully support also MacOS and Windows systems.
Dependencies
The software depends on the following projects:
Compilation Instructions using CMake
System-wide dependencies
The instructions have been tested over a machine with Ubuntu 24.04 as operating system.
First, install the following system-wide dependencies:
sudo apt install -y \
automake \
build-essential \
cmake \
curl \
libtool \
wget \
unzip
Install Flex, Bison
Install flex and bison:
sudo apt-get install flex bison
Install Graphviz
For the graphical features (automata and strategy visualization), graphviz need to be installed:
sudo apt install graphviz libgraphviz-dev
Install Syfco
Building Syfco requires the Glasgow Haskell Compiler. To install the tool you can use stack
:
curl -sSL https://get.haskellstack.org/ | sh
git clone https://github.com/reactive-systems/syfco.git
cd syfco
git checkout 50585e0
stack install
The installation should install the syfco
binary in a directory in teh system path. Then, make sure the binary syfco
can be found on your system path: which syfco
.
When using the CLI, you can also provide the path to the syfco
binary manually by setting --syfco-path
.
Install Slugs (Optional)
Slugs is required for solving LTLf synthesis with GR(1) conditions.
To build Slugs, run:
cd submodules/slugs
git checkout a188d83
cd src
make -j$(nproc --ignore 1)
Install CUDD
Make sure CUDD is installed. CUDD can be found at:
https://github.com/KavrakiLab/cudd.git
Install CUDD:
autoreconf -f -i
./configure --enable-silent-rules --enable-obj --enable-dddmp --prefix=/usr/local
sudo make install
Install Mona
To install MONA system-wide:
git clone --recursive https://github.com/whitemech/MONA.git
cd MONA
git checkout v1.4-19.dev0
./configure && make -j && sudo make -j install
# copy headers manually
sudo mkdir -p /usr/local/include/mona
sudo cp Mem/mem.h Mem/gnuc.h Mem/dlmalloc.h BDD/bdd_external.h BDD/bdd_dump.h BDD/bdd_internal.h BDD/bdd.h BDD/hash.h DFA/dfa.h GTA/gta.h config.h /usr/local/include/mona
Install Lydia
The tool requires the installation of Lydia, which will be triggered by the CMake configuration.
However, if you want to install Lydia manually, you can co into submodules/lydia
and follow the installation instructions in the README.md
.
Install Z3
By default, the CMake configuration will fetch z3 automatically from the GitHub repository. There might be required other dependencies.
In order to disable this behaviour, you can configure the project by setting -DZ3_FETCH=OFF
. In that case, you have to have the library installed on your system. To link the static library of z3, you have to install z3 manually:
wget https://github.com/Z3Prover/z3/releases/download/z3-4.8.12/z3-4.8.12-x64-glibc-2.31.zip
unzip z3-4.8.12-x64-glibc-2.31.zip
cd z3-4.8.12-x64-glibc-2.31
sudo cp bin/libz3.a /usr/local/lib
sudo cp include/*.h /usr/local/include
Build LydiaSyft
- Make build folder so your directory is not flooded with build files:
- Run CMake to generate the makefile:
cmake -DCMAKE_BUILD_TYPE=Release ..
- Compile using the generated makefile:
make -j$(nproc --ignore=1) LydiaSyft
4.1. For solving LTLf synthesis with GR(1) conditions, please install slugs
following submodules/slugs/README.md
- Compile and Run tests:
make -j$(nproc --ignore=1) tests
./bin/tests
Run LydiaSyft as CLI tool
Usage:
LydiaSyft: A compositional synthesizer for Linear Temporal Logic on finite traces (LTLf)
Usage: LydiaSyft [OPTIONS] SUBCOMMAND
Options:
-h,--help Print this help message and exit
--help-all Expand all help
-p,--print-strategy Print out the synthesized strategy (default: false)
-t,--print-times Print out running times of each step (default: false)
Subcommands:
synthesis solve a classical LTLf synthesis problem
maxset solve LTLf synthesis with maximally permissive strategies
fairness solve LTLf synthesis with fairness assumptions
stability solve LTLf synthesis with stability assumptions
gr1 Solve LTLf synthesis with GR(1) conditions
To see the options of each subcommand, run:
LydiaSyft [SUBCOMMAND] --help
Examples (run commands from the root directory of the project):
./build/bin/LydiaSyft synthesis -f examples/test.tlsf # UNREALIZABLE
./build/bin/LydiaSyft synthesis -f examples/test1.tlsf # REALIZABLE
./build/bin/LydiaSyft maxset -f examples/test1.tlsf # REALIZABLE
./build/bin/LydiaSyft fairness -f examples/fair_stable_test.tlsf -a examples/fair_stable_test_assumption.txt # REALIZABLE
./build/bin/LydiaSyft stability -f examples/fair_stable_counter_test.tlsf -a examples/fair_stable_test_assumption.txt # REALIZABLE
./build/bin/LydiaSyft gr1 \
-f examples/TLSF/GR1benchmarks/finding_nemo/finding_nemo_1.tlsf \
-g examples/TLSF/GR1benchmarks/finding_nemo/finding_nemo_1_env_gr1.txt \
-e examples/TLSF/GR1benchmarks/finding_nemo/finding_nemo_1_env_safety.ltlf \
-a examples/TLSF/GR1benchmarks/finding_nemo/finding_nemo_1_agn_safety.ltlf \
--slugs-path ./submodules/slugs/ # REALIZABLE
Use LydiaSyft as a library
The software also provides C++ APIs. Here there is an example of usage:
#include <memory>
#include <sstream>
#include <string>
#include <vector>
#include <lydia/parser/ltlf/driver.hpp>
#include "automata/ExplicitStateDfa.h"
#include "automata/ExplicitStateDfaAdd.h"
#include "automata/SymbolicStateDfa.h"
#include "game/InputOutputPartition.h"
#include "Player.h"
#include "VarMgr.h"
#include "synthesizer/LTLfSynthesizer.h"
int main(int argc, char ** argv) {
std::string formula_str = "F(a | b)";
std::vector<std::string> input_vars{"a"};
std::vector<std::string> output_vars{"b"};
auto driver = std::make_shared<whitemech::lydia::parsers::ltlf::LTLfDriver>();
std::stringstream formula_stream(formula_str);
driver->parse(formula_stream);
whitemech::lydia::ltlf_ptr formula = driver->get_result();
std::shared_ptr<Syft::VarMgr> var_mgr = std::make_shared<Syft::VarMgr>();
var_mgr->create_named_variables(partition.input_variables);
var_mgr->create_named_variables(partition.output_variables);
std::move(explicit_dfa_add));
var_mgr->partition_variables(partition.input_variables, partition.output_variables);
Syft::Player starting_player = Syft::Player::Agent;
Syft::Player protagonist_player = Syft::Player::Agent;
var_mgr->cudd_mgr()->bddOne());
std::cout << (result.realizability? "" : "NOT ") << "REALIZABLE" << std::endl;
return 0;
}
A DFA with explicit states and symbolic transitions represented in ADDs.
Definition: ExplicitStateDfaAdd.h:18
static ExplicitStateDfaAdd from_dfa_mona(std::shared_ptr< VarMgr > var_mgr, const ExplicitStateDfa &explicit_dfa)
Constructs an explicit-state DFA in ADD representation from an explicit-state DFA.
Definition: ExplicitStateDfaAdd.cpp:44
A DFA with explicit states and symbolic transitions.
Definition: ExplicitStateDfa.h:22
static ExplicitStateDfa dfa_of_formula(const whitemech::lydia::LTLfFormula &formula)
Construct an explicit-state DFA from a given formula using Lydia.
Definition: ExplicitStateDfa.cpp:31
A single-strategy-synthesizer for an LTLf formula given as a symbolic-state DFA.
Definition: LTLfSynthesizer.h:15
A DFA with symbolic states and transitions.
Definition: SymbolicStateDfa.h:18
CUDD::BDD final_states() const
Returns the BDD encoding the set of final states.
Definition: SymbolicStateDfa.cpp:135
static SymbolicStateDfa from_explicit(const ExplicitStateDfaAdd &explicit_dfa)
Converts an explicit DFA to a symbolic representation.
Definition: SymbolicStateDfa.cpp:92
Definition: Synthesizer.h:16
More code examples can be found in the examples/
folder. To build them, run make examples
.
Documentation
The documentation is built using Doxygen. First, install doxygen
:
Then:
License
This project is licensed under the GNU Affero General Public License. See the [LICENSE](LICENSE) file for details.
Copyright (C) 2024, Marco Favorito, Shufang Zhu. All rights reserved.