Abstract:We present an adequacy theorem for a concurrent extension of probabilistic GCL. The underlying denotational semantics is based on the so-called mixed powerdomains which combine non-determinism with stochasticity. The theorem itself is formulated via M. Smyth's idea of treating observable properties as open sets of a topological space.
One application of our theorem is that it entails semi-decidability w.r.t. whether a concurrent program satisfies an observable property (written in a certain form). This is intimately connected to M. Escardó's conjecture about semi-decidability w.r.t. may and must probabilistic testing.
What problem does this paper attempt to address?
The main problem that this paper attempts to solve is how to determine whether a given program satisfies a certain property in a programming language that combines probabilistic and concurrent operations. Specifically, for a concurrent imperative language with probabilistic operations, consider whether the program \( \langle P, s \rangle \) satisfies a certain property \( \varphi \) under all possible schedulers, symbolically represented as \( \langle P, s \rangle \models \varphi \).
For example, the property \( \varphi \) may mean that the probability that \( \langle P, s \rangle \) terminates under all possible schedulers is greater than \( \frac{1}{2} \). Or dually, \( \varphi \) may mean that there exists at least one scheduler such that the probability that \( \langle P, s \rangle \) terminates is greater than \( \frac{1}{2} \).
In addition, the paper also explores how to establish the observational pre - order relation \( \sqsubseteq \) of the program in state \( s \), that is, for all properties \( \varphi \), if \( \langle P, s \rangle \models \varphi \) then \( \langle Q, s \rangle \models \varphi \), and asks whether for any two given programs \( P \) and \( Q \) and state \( s \), \( \langle P, s \rangle \sqsubseteq \langle Q, s \rangle \).
These problems involve quantification over the universe of scheduling systems, which may be uncountable, and also include universal quantification over formulas. Therefore, the answers to these problems may be particularly challenging. The paper provides a method to solve these problems by introducing mixed powerdomains and combining probabilistic concurrency to extend the probabilistic guarded command language (pGCL).
### Formula Summary
- Property \( \varphi \) is represented as: \( \langle P, s \rangle \models \varphi \)
- Property of termination probability greater than \( \frac{1}{2} \): \( \mathbb{P}(\text{terminate}) > \frac{1}{2} \)
- Observational pre - order relation: \( \langle P, s \rangle \sqsubseteq \langle Q, s \rangle \)
### Main Contributions
The paper proposes a fully abstract theorem by connecting the above problems with domain - theoretic tools, and proves that logical expressions about program properties are semi - decidable under certain conditions. Specifically, for a formula \( \varphi \) in a major logical fragment, semi - decidability can be achieved by traversing the scheduler space. This includes the two problems mentioned above regarding the termination probability greater than \( \frac{1}{2} \).
### Conclusion
The paper extends pGCL by introducing mixed powerdomains and probabilistic concurrency, and establishes a computational adequacy theorem through domain - theoretic tools, solving the problem of determining program properties in probabilistic concurrent programming languages. This result not only provides theoretical support but also provides an effective algorithmic method for practical applications.