ScaFi: A Scala DSL and Toolkit for Aggregate Programming

The Exchange Calculus (Xc): A Functional Programming Language Design for Distributed Collective Systems

“…To showcase the broad applicability of the proposed language design, we implemented a Scala and a C++ version of 𝖷𝖢. The Scala version has been developed as an extension of ScaFi (Casadei et al, 2022b), and aims at showcasing the DSL and maximise portability to different platforms, including simulators. The C++ version has been developed as an extension of FCPP (Audrito, 2020), and has consequently been integrated into the main FCPP distribution.…”

“…We provide an implementation of 𝖷𝖢 as a DSL embedded into the Scala language 9 because of its cross-platform support (Doeraene, 2018), popularity for building distributed systems (Ghosh et al, 2012), and advanced support for internal DSLs (Artho et al, 2015). This implementation has been developed as an extension of ScaFi (Casadei et al, 2022b). The DSL is organised into a few core 𝖷𝖢 constructs and a library of reusable functions.…”

“…𝖷𝖢 can be framed in the context of a long-term research thread on programming languages and tools for programming collective adaptive systems, known under the umbrella terms of field-based coordination (Mamei and Zambonelli, 2006;Viroli et al, 2019) and aggregate computing (Beal et al, 2015;Viroli et al, 2019). This research area is characterised by works on formal calculi (Audrito et al, 2019(Audrito et al, , 2023a, new constructs (Audrito et al, 2020;Casadei et al, 2019), formal properties of programs and computations (Viroli et al, 2018;Beal et al, 2017;Audrito et al, 2018a), programming language implementations of formal calculi as DSLs (Casadei et al, 2022b(Casadei et al, , 2021Audrito, 2020), simulators (Pianini et al, 2013;Audrito et al, 2022e), algorithms and patterns (Beal, 2009;Audrito et al, 2017b,a;Pianini et al, 2021b;Audrito et al, 2021a;Pianini et al, 2022), execution models (Pianini et al, 2021a), distributed platforms and deployments (Casadei et al, 2020(Casadei et al, , 2022a, and libraries for application domains such as swarm robotics (Aguzzi et al, 2023) and distributed monitoring (Audrito et al, 2021b). In a nutshell, this work proposes a new calculus, 𝖷𝖢, inspired by previous calculi, that subsumes them and is strictly more expressive in its ability to model messages differentiated on a neighbour basis (see Section 8.3 for a more detailed comparison).…”

“…In the following we compare 𝖷𝖢 with the field calculus (FC) (Viroli et al, 2018;Audrito et al, 2020), which is the reference model for computational fields (Viroli et al, 2019), also implemented by DSLs like ScaFi (Casadei et al, 2022b(Casadei et al, , 2021 and FCPP (Audrito, 2020;Audrito et al, 2022e) ). Notice that the converse translation is not possible, as 𝚗𝚋𝚛, 𝚛𝚎𝚙 or 𝚜𝚑𝚊𝚛𝚎 expressions with arguments of neighbouring type have no defined behaviour in FC.…”

Field-informed Reinforcement Learning of Collective Tasks with Graph Neural Networks

“…In other words, it maps a Boolean field (where "true" indicates the presence of a source agent and "false" indicates its absence) to a distance field that indicates the proximity to the nearest source. In the ScaFi [25] implementation of field calculus is defined by function: def gradient(source: Boolean): Double…”