Periodic Reporting for period 4 - PaVeS (Parametrized Verification and Synthesis)
Berichtszeitraum: 2023-03-01 bis 2023-08-31
In parameterized systems, an arbitrary number of identical agents with limited computational power interact to achieve a common goal.
Examples include distributed algorithms (i.e. algorithms running on a network of communicating computing agents), algorithms for routing in networks, or multithreaded programs, in which different software threads communicate and interact.
In modern examples the agents are often mobile, and interact with each other in a random way, like in robot swarms, and in ``natural computers'', whose agents are molecules or cells.
The designer of multithreaded program usually expects it to work correctly for an arbitrary number of threads, and models of biological process are expected to exhibit the expected behaviour for any number of cells.
So, in fact, programmers and system biologists are not designing one system, but an infinite collection of systems, one for each number of agents.
In order to check that the system works as intended, current technology only offers limited help.
In most cases, it can only explore a small number of instances of the system. For example, current verification programs, called model checkers, can automatically prove that a leader election algorithm is correct for a small number of processes, but not for any number.
Parameterized verification is the field of computer science that designs methodologies, algorithms, and software tools for the verification of parameterized systems, and for the synthesis of correct-by-construction parameterized systems.
The mission of the PaVeS project has been to conduct basic research on the the three grand challenges in parameterized verification:
- Develop novel algorithms and tools for parameterized verification of classical parameterized systems that bypass the high complexity of current techniques.
- Develop the first algorithms and tools for parameterized verification of modern parameterized systems exhibiting stochastic behaviour.
- Develop the first algorithms and tools for synthesis of correct-by-construction parameterized systems.
Examples include distributed algorithms (i.e. algorithms running on a network of communicating computing agents), algorithms for routing in networks, or multithreaded programs, in which different software threads communicate and interact.
In modern examples the agents are often mobile, and interact with each other in a random way, like in robot swarms, and in ``natural computers'', whose agents are molecules or cells.
The designer of multithreaded program usually expects it to work correctly for an arbitrary number of threads, and models of biological process are expected to exhibit the expected behaviour for any number of cells.
So, in fact, programmers and system biologists are not designing one system, but an infinite collection of systems, one for each number of agents.
In order to check that the system works as intended, current technology only offers limited help.
In most cases, it can only explore a small number of instances of the system. For example, current verification programs, called model checkers, can automatically prove that a leader election algorithm is correct for a small number of processes, but not for any number.
Parameterized verification is the field of computer science that designs methodologies, algorithms, and software tools for the verification of parameterized systems, and for the synthesis of correct-by-construction parameterized systems.
The mission of the PaVeS project has been to conduct basic research on the the three grand challenges in parameterized verification:
- Develop novel algorithms and tools for parameterized verification of classical parameterized systems that bypass the high complexity of current techniques.
- Develop the first algorithms and tools for parameterized verification of modern parameterized systems exhibiting stochastic behaviour.
- Develop the first algorithms and tools for synthesis of correct-by-construction parameterized systems.
The main results of the project have been:
- An extensive analysis of the decidability and complexity of verifying multiple classes of parameterized systems, exhibiting different communication mechanisms, like rendez-vous, broadcast, and observation.
- Constraint-based technology for the verification of liveness properties of replicated systems.
- Verification algorithms for parameterized fault-tolerant algorithms
- Algorithms and software tools for the synthesis of correct-by-construction and provably efficient systems modelled in several formalisms, including population protocols, threshold automata, finite transducers, and first-order transition systems.
- Novel algorithms for the translation of temporal logic specifications into deterministic automata.
- Five software tools for automatic verification and synthesis.
- An extensive analysis of the decidability and complexity of verifying multiple classes of parameterized systems, exhibiting different communication mechanisms, like rendez-vous, broadcast, and observation.
- Constraint-based technology for the verification of liveness properties of replicated systems.
- Verification algorithms for parameterized fault-tolerant algorithms
- Algorithms and software tools for the synthesis of correct-by-construction and provably efficient systems modelled in several formalisms, including population protocols, threshold automata, finite transducers, and first-order transition systems.
- Novel algorithms for the translation of temporal logic specifications into deterministic automata.
- Five software tools for automatic verification and synthesis.