	Jonas Jusélius Department of Chemistry University of Tromsø

Mailing address: Department of Chemistry University of Tromsø N-9037 Tromsø Norway		Phone: (+47) 77644082 Fax: (+47) 77644765 Mobile: (+47) 47419869 E-mail: firstname.lastname@chem.uit.no PGP public-key

You are here: Home > Home

Science

Quantum chemistry is quantum mechanics applied to chemical problems. My research interests are mainly within development of new computational methods for calculating molecular properties. Previously I have been working mostly on methods for calculating magnetic properties of molecules, in particular, magnetically induced currents. My thesis on the subject can be downloaded from the University of Helsinki web site.

In Tromsø I will be working on a wider set of projects, including:

Vibrational corrections to electronic g-tensors
Coupled-cluster theory
[Resonance] Vibrational Raman Optical Activity (VROA)
Calculation of electrostatic potentials

List of publications

Curriculum Vitae

For more information about Tromsø and the arctic region go to www.destinasjontromso.no.

My former scientific home was Laboratory for Instruction in Swedish at University of Helsinki.

Scientific computer codes

GIMIC: The Gauge-Including Magnetically Induced Currents (GIMIC) program will be available as soon as I have time (and inspiration) to make a clean version, including documentation. The GIMIC code implements the theory outlined in ref. Jusélius J., Sundholm D. and Gauss J., Calculation of Current Densities using Gauge-Including Atomic Orbitals, J. Chem. Phys., 121 3952 (2004) (pdf). GIMIC currently interfaces to CFOUR and Turbomole.
Download source: gimic_1.01.tgz (private)
ARCS: The Aromatic Ring-Current Sheildings (ARCS) method is outlined in ref. Jusélius J. and Sundholm D., Ab initio determination of the induced ring current in aromatic molecules, Phys. Chem. Chem. Phys., 1 3429 (1999) (pdf). A small Python program to do ARCS analysis of the appropriate nuclear magnetic shielding data is available.
Download source: arcs_1.0.tgz
DAGE: Direct Approach to Gravitation and Electrostatics (DAGE) is a new algorithm to calculate (mainly) electrostatic potentials. A paper on the method will appear in JCP soon. I have co-authored a program which implements the DAGE method (my contribution consists merely of parallelizing the code). The program may or may not become available some time in the future.
Download source: dage_1.0.tgz (private)
Hindarot: Hindarot is a program to calculate spectra of molecules trapped in rare-gas matrices. A thorough description of the theory is given in "Flygare H., J. Chem. Phys. 39 2263 (1963).
The source code for the hindarot program is available for download. Please not that this program depends on libgetkw, and furthermore on a obsolete version of libgetkw, which means that you probably have to do a bit of code fixing to make it compile and link. The upside is that it does more or less everything one wants, including transition moments and Boltzmann populations of rotational levels. I will fix the code the next time I need the program (which might be never). This was also my first real attempt at F90, so the code is not exactly neat and tidy...

Download source: hindarot_1.0.tgz

Teaching

Spring 2007

KJE-2001 Teoretisk kjemi og spektroskopi
Föreläsningsanteckningar

Autumn 2006

KJE-1004 Innføring i uorganisk kjemi (Introducition to Inorganic Chemistry)
För mera information, se min kurssida.

KJE-3103 Quantum chemical methods (Kvantekjemiske metoder)
Lecture notes:

Software

git

I have written two documents on the git version control system: A nutshell guide for end users, and a guide how to migrate from CVS (or Subversion) to git. The guides are available on my git page

Fortran 90/95

I have given a number of lectures on programming in Fortran 90/95, and the latest version of my talk can be found here. There is also a small example program, demonstrating some of the paradigms in F95. I have also collected a number of good F90/95 articles and tutorials (to obtain these you need to contact me for a password).

Contrary to popular belief (see for example http://www.fortranstatement.com for a list of highly uninformed people), Fortran 90/95 is a very nice and powerful programming language for scientific computing. My advice to all F77 programmers is to approach F95 as a totally new language. Pure F95 has very little in common with F77. C/C++ programmers can benefit greatly from F95. F95 has an almost MATLAB like syntax for dealing with arrays, and much better support for object-oriented programming than C. The upcoming Fortran 2003 standard will further enhance the OOP capabilities of Fortran 95.

Some people claim that using the advanced features of F95 makes the code run significantly slower than F77. This is not correct, but there are a few (compiler dependent) things to watch out for. I have written a small profiling program which demonstrates various calling conventions and their impact on compiler performance. The program is available.

Vim

From the man page; "Vi IMproved, a programmers text editor". Vim is a wonderful text editor, but the standard features for indentation and syntax highlighting for Fortran 95 are a bit rough. With a little bit of tweaking however, it's easy to set them straight. I have put together a small repository of modified vim macros for Fortran 90/95, together with my ".vimrc". To install the file copy 'vim95/vimrc' to '$HOME/.vimrc' (or edit your .vimrc), and copy the directory vim95/vim to '$HOME/.vim'.

Download source: vim95_1.0.tgz

stubmake

GNU Autoconf is a generic tool to configure software source code packages on different computer architectures. The standard autoconf has rudimentary support for Fortran 90/95. Setting up autoconf from scratch is somewhat tedious, even for simple software packages. stubmake is a Makefile/autoconf stub package to configure Fortran 77/90/95 projects. The package requires only rudimentary knowledge of autoconf and make to customize it for some particular software package, and is still flexible enough to handle big program packages (e.g. CFOUR, and soon also Dalton ;^).
The package contains new and updated m4 macros to configure BLAS/BLAS95, LAPACK/LAPACK95, libgetkw, mpi and various aspects of Fortran 90/95.

Download source: stubmake_1.2.tgz

The autoconf macros are also available separately:

libgetkw

libgetkw is a simple input file parser with a C-ish syntax, with language bindings for C, Fortran 90 and F77. Defining valid parsers with libgetkw is very simple, have a look at this example parser which parses the following input. Here is an example how to parse the input from C and Fortran 90 (The examples are a bit messy, since they are my debugging codes...).

This release of libgetkw is very stable, but has not undergone any really extensive testing. It's still being actively developed, with new features being added as I need them. As a result I do not guarantee that the API remains constant. The next release of libgetkw will feature definition of ranges (e.g. key=4-9 or key=[1,2,6-12,0]), and a F90 interface to getopt() for decent command line parsing in Fortran. libgetkw is released under the GNU General Public License (GPL).

Download source: libgetkw_1.23.tgz

Update: There is now a much more flexile implementation available, implemented in Python. Drop me an e-mail if you are interested.

getcite

getcite is a small BibTeX database manager. It consists of two small programs; 'mkbibidx' which creates a binary database from .bib files (only unique bib entries). The database can then be interactively searched and manipulated with 'getcite'. 'getcite' can also compile .bib files from citations in .tex files. getcite is released under the GNU General Public License (GPL).

Download source: getcite_0.1.tgz
Download Debian package: getcite_0.1-1_i386.deb [getcite_0.1-1.dsc ]

PBS and SGE

Most batch and/or queue systems are a pain to use (IMHO). I have written a small generic script to run jobs, either under PBS or SGE (can easily be extended to other batch systems). The script can run both serial and parallel jobs (supports MPICH, LAM-MPI and Scali). It has the following list of features:

Supports PBS and SGE (Sun GridEngine)
Supports MPICH, LAM-MPI and Scali parallel environments, as well as serial jobs
Can run arbitrary pre- and post jobs scripts
In case a job crashes, the script can generate a file containing a list of nodes and work directories. It can also generate a clean-up script to remove files after a crash.
Can copy and uniquely tag working files from all compute nodes back to the submit directory. A maximum size of files to copy can be specified (e.g. prevents copying inconveniently large two-electron files).

Download: runjob.sh (Latest version is 1.0).