Research Training: Using Departmental Software for Research
Connecting to UNIX/Linux Systems
- logging in securely: ssh (a secure equivalent of telnet, which has been disabled on the departmental servers)
- copying files securely between UNIX machines: scp (a secure equivalent of ftp)
- graphical connection: NX
- secure and recommended by support
- can be used for logging in from outside; if your connection is fast enough, the response time is quite reasonable
- read the installation instructions carefully!
- use version 1.5.0
- .nx/config only appears after you run the program once
- set Configure->General->Display to Fullscreen
- still not very reliable :-(
- alternative for graphical connections: VNC
- not secure and therefore not good for logging in from outside
- when installing on a Windows machine, install only the client!
- to run, start a remote server first (you need to log in using, e.g., putty) and then a client on your machine
- simple and highly reliable
- your own little UNIX on a Windows machine: cygwin
- allows you to run UNIX scripts under Windows
- using ssh -X you can connect to a Unix machine and run programmes (including those with GUI) remotely
Some Books and Links
- emacs editor: D.Cameron, Learning GNU Emacs. O'Reilly , 1991. RHUL code: 001.6425 CAM (may be difficult for UNIX beginners)
- Numerical Recipes: a series of books by H.S.Press et al with complete code
- C and FORTRAN versions are available on-line: www.nrbook.com.
- C++ version is available from the library: Numerical Recipes in C++, Cambridge, 2002. RHUL code OO1.6424 NUM
- from support:
MatLab
- to start, type matlab
- a no-GUI version: matlab -nojvm; it is more reliable and you can still draw graphics
- learning MatLab: go to Help->MATLAB->Printable Documentation and read "Getting Started"; use the other manuals for reference and further reading
- This tutorial was written for a final year course; you may find it useful to get started.
- some features of MatLab:
- intended for numerical computations; has a huge library of numerical methods
- based on matrices (MATrix LABoratory)
- has great graphics and many tools for visualising data
- good for rough and ready prototypes
- no type checking; encourages poor programming style
- efficient use of MatLab requires vectorisation
- if necessary, extra routines can be written in C
- if you have a Windows machine connected to the departmental system, you can install and run MatLab
- if you have a Linux machine connected to the departmental system, you can run MatLab; the executable code will be downloaded from a departmental server.
Mathematica
- to start, type mathematica
- before running under NX, you need to set up fonts: in Configure->General->Environment specify font server fonts.cs.rhul.ac.uk and port 7100.
- when running under NX, make sure that NumLock is off (!)
- a no-GUI version: math; you can still draw graphics
- learning Mathematica: go to Help->Help Browser->The Mathematica Book and look through "A Practical Introduction to Mathematica"
- some features of Mathematica:
- intended mainly for symbolic calculations
- can serve as a reference book for integrals, sums of series etc
- pretty good graphics capabilities
- programming is tricky and counterintuitive
- if you have a Linux machine connected to the departmental system, you can run Mathematica; the executable code will be downloaded from a departmental server.
- the departmental license only allows running 3 instances of Mathematica simultaneously; after you have finished working with Mathematica, please do not forget to exit