Batch Systems: High performance computing resources where program execution is regulated by means of a queue. Usually such systems are non-interactive and thus do not cater for visualisation.

Boundary Copy: Decomposition of the data to processes causes CA cells to lose their neighbours. This is tackled by copying cells over from neighbouring processes. A similar problem appears in the border cells of the model across all dimensions and is handled similarly. Data replication of this kind is called boundary copy and occurs between subsequent iterations of the model.

CABOTO: HPCN Esprit programme; the predecessor of COLOMBO.

CAMELot: The Cellular Automata programming environment developed by EPCC as part of the COLOMBO project.

CARPET: The Cellular Automata programming language used in CAMELot. CARPET is derived from the C programming language.

Cellular Automata (CA): A computer model made up of an array of cells that evolve in time. At each timestep of the model the state of each cell is updated depending on its previous state and the previous states of the surrounding cells of the model.

Large Scale Facilities: State-of-the-art facilities used for resource-demanding applications. In the CAMELot context, the term refers to computing platforms.

Load Balancing: Because the transition function complexity can depend on the position of the cells and because of the application of the inactive region mechanism, some processing elements may work on different loads than others. This leads to underutilisation of resources. Load balancing is the act by which these variations are minimised.

Parallel Paradigm: One of several different approaches (paradigms) to writing parallel applications in mainstream languages such as C or Fortran. The prevalent examples are:

• Message Passing - The parallel application consists of many separate programs (or multiple copies of the same program) that communicate using a library of message passing primitives (e.g. the MPI library used in CAMELot)
• Parallel directives - The application is written as a single program and directives are used to inform the compiler how to distribute the work (and maybe the data) across processors (e.g. the OpenMP or HPF directives)

Transition Function: The rule which defines the transition of a CA from one state to the next at a given time step.