Big Practical Guide to Computer Simulations

The following sections are included:

• Dedication
• Preface to the 2nd edition
• Preface to the 1st edition
• Contents

Performing computer simulation in a sophisticated way always includes writing computer programs. Even if one builds upon an existing package, one usually has to perform minor or major modifications or write new subroutines for data analysis…

Programmers's life can be made much easier when using scripts. The first task after writing the source code is to compile the code. For large software projects, containing many modules, this can be quite an effort if compiling is done by hand. Special scripts, called Make files, help a great deal to manage large software and other projects as explained in Sec. 2.1. Next, in Sec. 2.2, shell scripts are explained, which allow to perform many tasks automatically. Here, scripts for the bash (Unix/Linux) shell will be introduced. In fact, shell-scripting languages are like small programming languages. More sophisticated are extended script languages, like Python, which is introduced in Sec. 2.3. They allow to implement rather large projects in a compact way with small effort.

Performing simulations or other large-scale software projects is not only a matter of skillful coding. A lot depends on the organization of the design and the way of programming. Also the methods used for testing and the organization of actually running the simulations and performing the data analysis have a strong influence on the success of a project. This is in particular true, if several people are involved in the project. In this chapter, strategies for performing software projects are discussed. The corresponding computer-science field is called software engineering. An introduction to software engineering is given, for example, in Ref. [Sommerville (1989)], while a more practical approach is presented in Ref. [Kernighan and Pike (1999)]…

In Chap. 3, the importance of thorough testing has already been stressed. Here, four useful tools are presented which significantly assist in the debugging process. Please note again that the tools run under UNIX/Linux operating systems. Similar programs are available for other operating systems as well. The tools covered here are gdb, a source-code debugger, ddd, a graphic front-end to gdb, valgrind, which finds bugs resulting from bad memory management, and finally gprof, which assists in finding running-time consuming parts of your program.

In recent years, object-oriented programming languages like C++, Smalltalk or Eiffel became very popular. These programming languages are usually used to implement programs which are designed in an object-oriented way. But, using an object-oriented language and designing a simulation program in an object-oriented style are not necessarily the same. For example, you can set up your whole project by applying object-oriented methods even when using a traditional procedural programming language like C, Pascal or Fortran. In general, taking an object-oriented viewpoint facilitates the analysis of problems and the development of suitable programs. An introduction to object-oriented software development can be found for example in Refs. [Rumbaugh et al. (1991); Johnsonbaugh and Kalin (1994); Skansholm (1997)]. In the first section of this chapter, an introduction to object-oriented concepts is given…

If you ever want to change, extend, or even develop a full simulation program, you could, in principle, just use the techniques you have learnt in the previous chapters, and directly implement the simulation methods using arrays and C structs as main data structures. Nevertheless, the efficiency of such a simulation program can be greatly enhanced, if you know about fundamental algorithms and data structures in computer science. There are many cases in science, where researchers first used straightforward implementations for their problem and even got some results which were published in scientific papers. Later on, it was found out in these cases that by using simple techniques as described in this chapter, the problem under investigation could be tackled much better, i.e. larger system sizes could be studied, and more data could be gathered, hence reducing the statistical error bar of the data (c.f. Chap. 8). Then, using the better approach, it turned out that the conclusions, which were previously drawn using the straightforward program, were wrong. Hence, if you want to be among those who write efficient programs which can simulate large systems, then you should read this chapter carefully…

Libraries are collections of data types and functions which can be used in other programs. There are libraries for numerical methods such as integration or solving differential equations, for storing, sorting and accessing data, for fancy data types like lists or trees, for generating colorful graphics and for thousands of other applications. Some libraries can be obtained for free, while other, usually specialized libraries have to be purchased. The use of libraries speeds up the software development process enormously, because you do not have to implement every standard method by yourself. Hence, you should always check whether someone has done the jobs for you already, before starting to write a program…

In this chapter, we are concerned with statistics in a very broad sense. This involves generation of (pseudo) random data, display/plotting of data and the statistical analysis of simulation results…

When you need the information given in this chapter, you have advanced very far, congratulations! You have almost completed a full cycle of a scientific computer simulation project. The results of your simulations are analyzed and now you want to prepare a publication or give a talk…

The following sections are included:

• Appendix A: Supplementary Materials
• Bibliography
• Index