sybila/pithya-core

Pithya is a tool for parameter synthesis of ODE-based models and properties based on a hybrid extension of CTL.

Online demo

To try Pithya online, visit pithya.fi.muni.cz. In case of any problems/questions, feel free to contact us at sybila@fi.muni.cz.

Dependencies

To run Pithya, you need to have Java 8+ and Microsoft Z3 4.5.0 installed. If your OS is supported, we strongly recommend downloading precompiled Z3 binaries from github (Pithya allows you to specify a custom Z3 location, so you don't necessarily need to have it in your PATH, but we recommend doing that anyway).

We strongly recommend to use Pithya with a GUI that is developed as a separate project and accessible at https://github.com/sybila/pithya-gui.

Download Pithya

You can download the latest version of Pithya from the releases.

Run

Pithya has one main binary bin/pithya. The bin folder also contains other executables, however, these are used only when pithya operates together with the GUI interface, so you don't need to worry about them.

Arguments

This is desription of arguments for parameter synthesis. Arguments for component analysis are described in the corresponding repo.

Input and output

[-m,--model] filePath required Path to the .bio file from which the model should be loaded. Detailed description of the .bio format can be found here TODO link
[-p,--property] filePath required Path to the .huctl file from which verified properties are loaded. Detailed descrption of the .huctl format can be found here TODO link
[-ro,--result-output] [stdout, stderr, filePath] default: stdout File/stream to which verification results should be printed. You can use this option to print log and results separately. Note: errors are always printed to stderr.
[-r,--result] [human, json] default: human Output format that is used when printing results.
- human Informal text format that should be easily readable to sentient creatures.
- json A more formal output format that can be easily parsed by other tools. Description TODO.
[-lo,--log-output] [stdout, stderr, filePath] default: stdout File/stream to which logging info should be printed. You can use this option to print log and results separately. Note: errors are always printed to stderr.
[-l,--log] [none, info, verbose, debug] default: verbose Amount of log data to print during execution.
- none No logging.
- info Print coarse verification progress and statistics (started operators, final solver throughput).
- verbose Print interactive progress with dynamic throughput statistics (roughly every 2s).
- debug Print everything.

Verification options

--parallelism integer default: runtime.avaialableProcessors The maximum number of threads that are used for parallel computation (this is an upper bound, for some specific models/properties, the desired level of parallelism might not be achievable).
--z3-path filePath default: z3 Relative or absolute path to the z3 command line executable.
--fast-approximation [true, false] default: false Uses much faster, but not necessarily optimal version of the [Piecewise linear approximation] (https://optimization.mccormick.northwestern.edu/index.php/Piecewise_linear_approximation) when evaluating the model ODEs.
--create-self-loops [true, false] default: true Creating selfloops can cause significant overhead even though they have no impact on some types of properties (mainly reachability). You can disable selfloops using this switch.

Build from source

You can naturally build Pithya from source if you're so inclined. After you clone the repository, you can run one of these commands in the root folder:

# On Windows, replace ./gradlew with ./gradlew.bat

# Build Pithya and place unpacked distribution into ./build/install/pithya
./gradlew installDist

# Build Pithya and place compressed distribution into ./build/distributions
./gradlew distZip

(if you have a local gradle installation, you can replace ./gradlew with gradle for faster build)