# Information about
software, report writing and graphs

Here is a list of useful links and documents.

If you are not from the School of Informatics, you will need a DICE computer account.

Contact the Informatics Teaching Office (ITO) in Appleton tower

Alternatively, you can bring your laptop and install the relevant software.

#### Matlab

In most courses we use Matlab. Matlab is easy to learn.

Apart from the online tutorials and demos that come with Matlab, here
are two elementary
introductions:

Matlab
primer (Florida) and a

local one
from David Barber and Chris Williams.

Octave is a decent matlab look-a-like.

It is not 100% compatible, on the other hand it has some useful things that matlab doesn't have:

- function definition within script files

- the '+=' operator

#### Reports

Although reports can be hand-written, there are ways to get nicer
reports.

Many people use

lyx ('lyx' on DICE),
which is a WYSIWIG interface to LaTeX, or else use LaTeX itself.

In either case figures can be included using the Encapsulated Postscript
'eps' format, or you can use pdflatex and use PDF or PNG figures.

Openoffice is another possibility ('ooffice on DICE').

Check on your final report that symbols and Greek fonts came out correctly.

#### Graphs

Reports with improper graphs can lead to lesser marks.

Here is a

guide
on producing graphs.
In particular one should

- not be tempted to fit data to higher order
polynomials or splines to data without reason,
as doing so implies a certain relation between x and y data.

- properly label all axis and make sure they have units
- avoid bar plots

'gnuplot' is one of the programs that can be used to produce graphs.

It can be a bit of a pain to create plots electronically which obey the guidelines stated in the above link.

Often it's easier simply to add text and change offending bits using
an EPS or PDF editor, such as Inkscape (installed on DICE).

#### Units and errors

If the outcome of a question is a quantity, don't forget the units:

For example: t =2 ms.

If errors are known or can be estimated, the proper way to write this is

t=(2.2 ± 0.5) ms.

Just adding decimals, like t=(2.2123412±1.491234) ms does not
improve accuracy.

The rule is this:

Only one digit of the error should be stated, e.g.

t= (0.10 ± 0.03) ms

unless the first digit in the error is a '1'

t= (324 ± 12) ms