Computational science is one of the wonders of the modern world. In almost all areas of science the use of computational techniques is rocketing, and software has moved from being a supporting tool to being a key site where research activities are performed. This has meant a huge increase in the importance of controlling and orchestrating computers as part of the daily routine of a scientific laboratory, from large teams making and running the computers performing global climate simulations to the individual scientist/programmer working alone. Across this spectrum, the productivity of teams and the happiness of scientists depends dramatically on their overall competency as programmers, as well as on their skills as researchers within their field. So, in the last 30 years we have seen the continued rise of that new profession: the scientific programmer. A good scientific programmer will carry both epithets with pride, knowing that programming is a key foundation for a successful publication record.

However, programming cultures differ widely, and, over time, gaping divides can emerge that can be to the detriment of all. In this book, Dr. Harrop has taken great steps forward to bridging three very different cultures: managed code programming, scientific programming and functional programming. At a technical level, each has its unique characteristics. Managed code programming, epitomized by .NET and Java, focuses on the productivity of the (primarily commercial) programmer. Scientific programmers focus on high performance computations, data manipulation, numerical computing and visualization. Functional programming focuses on crisp, declarative solutions to problems using compositional techniques. The challenge, then, is to bring these disparate worlds together in a productive way.

The language F#, which Dr. Harrop uses in this book, itself bridges two of these cultures by being a functional language for the .NET platform. F# is an incredibly powerful language: the .NET libraries give a rich and solid foundation of software functionality for many tasks, from routine programming to accessing web services and high performance graphics engines. F# brings an approach to programming that routinely makes even short programs powerful, simple, elegant and correct. However Dr. Harrop has gone a step further, showing how managed code functional programming can revolutionize the art of scientific programming itself by being a powerful workhorse tool that unifies and simplifies many of the tasks scientific programmers face.

But what of the future? The next 20 years will see great changes in scientific programming. It is customary to mention the ever-increasing challenges of parallel, concurrent, distributed and reactive programming. It is widely expected that future micro-processors will use ever-increasing transistor counts to host multiple processing cores, rather than more sophisticated microprocessor designs. If computations can be parallelized and distributed on commodity hardware then the computing resources that can be brought can be massively increased. It is well known that successful concurrent and distributed computing requires a combination of intelligent algorithm design, competent programming, and core components that abstract some details of concurrent execution, e.g. databases and task execution libraries. This needs a language that can interoperate with key technologies such as databases, and parallelism engines. Furthermore, the ability to declaratively and crisply describe solutions to concurrent programming problems is essential, and F# is admirably suited to this task.

The future, will, however, bring other challenges as well. Truly massive amounts of data are now being generated by scientific experiments. Web-based programming will become more and more routine for scientific teams: a good web application can revolutionize a scientific field. Shared databases will soon be used in almost every scientific field, and programmatic access to these will be essential. F# lends itself to these challenges: for example, it is relatively easy to perform sophisticated and high-performance analysis of these data sources by bringing them under the static type discipline of functional programming, as shown by some of the samples in this book.

You will learn much about both programming and science through this book. Dr. Harrop has chosen the style of F# programming most suited to the individual scientist: crisp, succinct and efficient, with a discursive presentation style reminiscent of Mathematica. It has been a pleasure to read, and we trust it will launch you on a long and productive career as a managed code, functional scientific programmer.