Recent Changes - Search:

CSE dept.


This site contains wiki-pages for the department of computer science and engineering at Chalmers and GU.


edit SideBar

ST /

GRACeFUL

Global systems Rapid Assessment tools through Constraint FUnctional Languages

Inofficial work page for (the Chalmers part of) an EU project proposal sent to the FETPROACT1-2014 call 2014-04-01 and funded with budget 2.4M EUR over three years. Chalmers is one of seven sites.

Patrik Jansson, Alex Gerdes, Solrun Einarsdottir and Maximilian Algehed. (Earlier also Cezar Ionescu, now at Oxford Uni. and Irene Lobo Valbuena.)

Acronym: GRACeFUL

Official project page: http://graceful-project.eu/

Admin (only relevant locally):

  • KST: 3725
  • PNr: 3720485

News and links

WP4: the DSL work package

The Chalmers site is mainly responsible for the DSL workpackage. Below as an edited version of the WP description.

[The task is] To build a DSL for translating the concept maps developed and manipulated during GMB sessions to system dynamics models adequate for the CFP layer.

One of the most successful mechanisms for managing complexity is linguistic abstraction: the creation of DSLs, especially designed for expressing problems and their solutions in a given domain.

The overall purpose of WP4 is 
to use a DSL for policy concept maps, logic and relations 
  to bridge between 
     the complexity of the CRUD case study from WP2 (visualised in WP3) 
  and 
     the underlying science and technology of WP4. 

In the longer term this will lead to a DSL aimed at building scalable RATs for collective policy making in Global Systems.

During the project we will work with embedded DSLs to improve scalability, verifiability and correctness of the models.

This WP builds on several years of research on DSLs for modelling global systems (FP7 CA GSDP).

Description of work

  • T4.1 identify key underlying concepts needed for the CRUD case study
  • T4.2 develop a DSL to describe the concept maps developed during GMB sessions
  • T4.3 provide a formal semantics for the elements of the DSL
  • T4.4 implement a middleware for connecting the DSL to the CFP layer
  • T4.5 build a testing and verification framework for RATs

Deliverables

  • D4.1 Formal description of concept map elements needed for CRUD case study (m6)
  • D4.2 DSL for description of concept maps, with formal semantics (m18)
  • D4.3 Translation of concept map descriptions to system dynamics models for the CFP layer (m30)
  • D4.4 Testing and verification framework for RATs with applications to the CRUD case study (m36)

D4.1 Formal description of concept map elements needed for CRUD case study (m6)

In the first phase we will do initial domain modelling: collect concept map examples relevant for the CRUD case study, examine and decompose the concept maps into the key underlying elements and provide a formal description of these concept map elements in terms of types and relations.

D4.2 DSL for description of concept maps, with formal semantics (m18)

In the next phase we plan to 1) investigate how simpler concepts can be combined and connected to form more complex concept maps, 2) identify the "rules of the game" necessary to rule out meaningless combinations and 3) provide a formal semantics for key elements and combinators in the form of a DSL with an evaluator.

D4.3 Translation of concept map descriptions to system dynamics models for the CFP layer (m30)

For the third deliverable the focus is on connecting the DSL and its evaluator to the underlying layer of functional and constraint programming "middleware". The requires translating and refining concept map descriptions to system dynamics models. We aim for a compositional translation with simple model components corresponding to the key underlying elements of the DSL and "model combinators" corresponding to the combinators in the DSL.

D4.4 Testing and verification framework for RATs with applications to the CRUD case study (m36)

The final deliverable is a property-based framework for semi-automatic testing and verification of the rapid assessment tools described by the DSL. This framework will evolve during the project so that a preliminary version can be used from the second year on and we expect it to reach a mature state by the end of the project.

Abstract (for the full GRACeFUL project)

The making of policies coping with Global Systems is a process that necessarily involves stakeholders from diverse disciplines, each with their own interests, constraints and objectives. People play a central role in such collective decision making and the quest for solutions to a problem generally intertwines its very specification. Simulators can assist in this process provided they employ adequate high-level modelling to separate the political question from the underlying scientific details. Domain-specific Languages (DSL) embedded in Functional Programming (FP) languages offer a promising way to implement scalable and verifiable simulators. But the use of simulators is essentially a trial-and-error process too tedious for execution in a group session. A paradigm shift is needed towards active problem solving where stakeholders’ objectives can be taken along from the very beginning. Constraint Programming (CP) has demonstrated to enable such a shift for e.g. managed physical systems like water and power networks. This project lays the base for a DSL aimed at building scalable Rapid Assessment Tools for collective policy making in global systems. This can be achieved through foundational scientific work at different levels: from the high-level, political modelling, adapting the social discipline of Group Model Building (as used in business organizations), through visual forms of CP as well as gamification aspects, down to the needs for a host language, combining CP and FP. Special emphasis is put on domain-specific constraints, constraint composition, and composable solvers and heuristics. Results are applied and validated for the problem case of Climate-Resilient Urban Design, but the ambition is a general framework applicable to many other systems. The case study is assessed by an external multi-disciplinary Advisory Board of Stakeholders that guides the specification process and evaluates needs and usability of the tools.

Keywords: Global Systems Science, Problem Solving, Collective Policy Making, Rapid Assessment Tools, Constraint Programming, Functional Programming, Climate-resilient Urban Design, Group Model Building

Edit - History - Print - Recent Changes - Search
Page last modified on November 17, 2017, at 06:12 PM