Griotte: Verified Compartmentalisation via Capabilities
本文通过引入Griotte和Griotte OS，利用形式化方法验证了CHERIoT基于能力的隔离机制的有效性和安全性。
0 citationsRead paper
Selected work
Representative Papers
Type-Directed Discretization of Probabilistic Programs (Extended Version)
This work addresses the challenge of exact discretization and inference for continuous distributions in recursive higher-order probabilistic programs by proposing the Slice transformation. Through non-local type-directed analysis, this method partitions continuous values into finite regions and rewrites sampling behavior, achieving globally semantics-preserving exact discretization via coupled logical relation proofs. Consequently, programs originally reliant on continuous sampling can now be executed by discrete engines, overcoming limitations of prior systems. The proposed approach delivers inference performance comparable to state-of-the-art methods, establishing a novel paradigm for reasoning about mixed probabilistic programs.
0 citationsRead paper
Simple Types for Polymorphic Functions
This work proposes a concise type system for combinatory logic that achieves expressive polymorphism without relying on explicit quantified types. The system assigns at most one type to each combinator, with polymorphism emerging dynamically during application and precisely capturing the structure of values. In contrast to Hindley-Milner typing, the approach supports a broader notion of polymorphism while avoiding complex type syntax. An efficient type inference algorithm is developed, preserving the formal simplicity of the system and providing a solid foundation for static program analysis.
0 citationsRead paper
Recent publications
Latest Papers
Griotte: Verified Compartmentalisation via Capabilities
本文通过引入Griotte和Griotte OS，利用形式化方法验证了CHERIoT基于能力的隔离机制的有效性和安全性。
0 citationsRead paper
Type-Directed Discretization of Probabilistic Programs (Extended Version)
This work addresses the challenge of exact discretization and inference for continuous distributions in recursive higher-order probabilistic programs by proposing the Slice transformation. Through non-local type-directed analysis, this method partitions continuous values into finite regions and rewrites sampling behavior, achieving globally semantics-preserving exact discretization via coupled logical relation proofs. Consequently, programs originally reliant on continuous sampling can now be executed by discrete engines, overcoming limitations of prior systems. The proposed approach delivers inference performance comparable to state-of-the-art methods, establishing a novel paradigm for reasoning about mixed probabilistic programs.
0 citationsRead paper
Simple Types for Polymorphic Functions
This work proposes a concise type system for combinatory logic that achieves expressive polymorphism without relying on explicit quantified types. The system assigns at most one type to each combinator, with polymorphism emerging dynamically during application and precisely capturing the structure of values. In contrast to Hindley-Milner typing, the approach supports a broader notion of polymorphism while avoiding complex type syntax. An efficient type inference algorithm is developed, preserving the formal simplicity of the system and providing a solid foundation for static program analysis.
0 citationsRead paper