Wes' statistics course

---

This course, no longer given, covered the use of statistics in experimental physics. It included

a brief review of probability,

an introduction to Monte Carlo methods to simulate data,

estimation of parameters of models, confidence intervals, and hypothesis testing.

Both Bayesian and classical statistics are covered. There is an emphasis on the use of computers. The exact choice of topics and the depth to which they are treated will be adapted to the background and interest of the students.

The version of the lecture notes used the last time the course was given are available as gzipped ps or as pdf. --> -->

The latest revised version of the notes is available as gzipped ps or as pdf.

Interested students should me at wes@hef.ru.nl.

*******************************************************************************

A large fraction of the exercises, both weekly and in the exam, are to be done using a computer. The student will have to write programs in order to solve these exercises. The preferred programming language is FORTRAN (77 or 90), since a library with useful subroutines and packages is available in FORTRAN. However, C may also be used. It is possible to combine C and FORTRAN routines.

The preferred platform is the UNIX system of the faculty, since FORTRAN (and C) compilers and the subroutine libraries are available there. Also, data files to be used with the exercises will be made available on this platform. Note that PASCAL does not exist on this platform. Students who wish to use some other language or platform should discuss this with me wes@hef.kun.nl.

Some information to get you started with the computer is in ~wes/StatInfo

If you are unfamiliar with UNIX, you might browse in Telefysica. Also see very basic UNIX commands.

If you are unfamiliar with FORTRAN, you might look at the short FORTRAN course from the University of Liverpool. Note, however, that the editor it mentions is not available here and that we use the Sun compiler rather than the NAG compiler. You might also look at Peter Klok's FORTRAN 77. FORTRAN 77 is a subset of FORTRAN; much of it is obsolescent, which means it still is part of the language but will eventually be dropped. If you write a routine in the fixed format of FORTRAN 77, give the file a .f extension rather than the .f90 extension of the preferred free-form format.

Sample programs illustrating the making of histograms are in ~wes/StatDemo. To make histograms you can use the HBOOK histogram package.

To perform numerical minimization of a function (for use in maximum likelihood and least squares fitting) you can use the MINUIT package.

Descriptions of some possibly useful FORTRAN routines are in ~wes/UsefulRoutines