During my career so far, I seem to have acquired an ear worm, or perhaps established somewhat of a leitmotif in my work: stencil computations over arrays. This idiom is pervasive in many applications, including in science at the core of models providing discrete approximations to differential equations. My early encounters with parallelising computational fluid dynamics simulations in C led me to consider functional domain-specific language (DSLs) to enable productive programming of stencil computations, which are at the heart of such fluid simulations. At the core of this DSL was the realisation that stencil computations are structured by a comonad, and that this interface provides a useful abstraction over various implementations, e.g., targetting different hardware. This work was born out of witnessing a need for more fruitful interaction between computer scientists and physical scientists in the design of programming languages, tools, and systems. Later, I developed declarative specifications for stencil computations in Fortran code, which led to a suite of tools developed alongside climate scientists. My interest in working with climate scientists grew and in 2022 I co-initiated the Institute of Computing for Climate Science at Cambridge of which I am a co-director. In this talk, I will trace a thread through the developments described above and end by looking at the challenges facing modern climate modelling research. Ideas for how the functional programming community could be involved in such work will be discussed and I will leave plenty of time for questions.

Dr Dominic Orchard is a Senior Lecturer in the Programming Languages and Systems group, School of Computing at the University of Kent, UK and is co-director of the Institute of Computing for Climate Science at the University of Cambridge. His research interests are programming language design and semantics, mathematically structured programming, effect and coeffect systems, embedded domain-specific languages (esp. for parallelisation), and applying programming language research to computational science.