HOW TO PLOT SCIENCE GRAPHS WITH MATLAB

A guide to using Matlab for scientific plots in Physics 359E

INTRODUCTION

The third year lab generates a large amount of scientific data that is best presented in the form
of graphs. However, not everyone is familiar with the graphical tools available in the lab and on
many computers elsewhere. This short note summarizes how to make simple data plots, and plots
of algebraic functions – for example, fits to the data points – using the computing language Matlab.
Matlab is available on the computers in the lab. You may also have bought a student edition of
Matlab for a previous course in applied maths or physics.

Creating a graph of a set of data points

Suppose you have a file on one of the lab computers that contains science data in the form of an
x – y table: two columns of numbers (or perhaps three if you have a column with measurement
errors). The following recipe should enable you to produce a graph of these data which you can
include in your lab report, either as a figure that you import into the word processor file of your
report, or as a stand-alone figure which you can simply print out and staple into your report.

• Start Matlab from the icon on the desktop (or search for Matlab using the Find command).
The programme will open with a screen showing three windows: the Command Window,
where you enter Matlab commands after the prompt; the Workspace window, which shows the
names and sizes of the data variables and arrays you have defined; and the Command History
window, which lists the commands you have recently typed into the Command Window.
There will also be a number of pull-down menus and buttons.

• Since you want to make a graph, your first job will be to import data that you want to
visualize. It will be very helpful if your data are in a file with a name ending in the extension
“.dat”; otherwise Matlab will really not believe that the file contains data, and will want to
treat it as a programme. Let’s suppose that your file is called [Link], and that it has
three columns of numbers, say an x value (perhaps energy), a y value (for example, number
of counts with that energy), and an error or uncertainty σ. The x, y, σ data should simply
be separated on each row by spaces.

• In the Workspace window, click on the Load Data File button at the top of the window. A
pop-up will open. Select Files of Type: All Files. Navigate to the directory (folder) where
you have your data. Select the file you want with the left mouse button, and click on Open.
The Import Wizard will open, and (hopefully) show the first part of your file. Avoid the
temptation to do anything with the Wizard except clicking on Next, then Finish. Your data
array should now show in the Workspace window.

1
 • Now the way graphing in Matlab works is that you must provide the plot routine with a list
of x values at which you want plotted point, and a corresponding list of y values, one for
each x point. You can do this in two convenient ways. The most direct is to tell the plot
programme which columns in your data file to use by typing the following command after the
prompt in the Command Window:
>> plot(mydata(:,1),mydata(:,2))
This command invokes the plotting programme, and gives it a list of x points (the first column
of your datafile mydata; the colon “:” tells Matlab to use all of the points in the column, and
the 1 tells it to use points only from the first column) and corresponding y points (the second
column in your datafile). When you enter this command, a new little window called Figure
No. 1 appears on the screen. This should have your data plotted in a reasonably chosen
box. The Figure window has a number of buttons (look at these); one useful one is the Print
Figure button, which sends your figure to the printer. Another useful one is the Save Figure
button, which stores the figure for later use in your lab report.

• The other way of making a graph, which you may find a little clearer, is to first define two
column vectors, one with the x values, and the other with the y values, as follows:
>> xdata=mydata(:,1);
>> ydata=mydata(:,2);
where the semicolon “;” after the command simply tells Matlab not to bother listing the
whole new column vector xdata on the screen. Then
>> plot(xdata,ydata)
produces the same figure as before.

• The command
>> errorbar(mydata(:,1),mydata(:,2),mydata(:,3))
produces the same plot as before, but uses the third column of mydata to plot errorbars on
each point in the figure.

• You can make your graph much fancier, of course. Melissinos & Napolitano describe a number
of the useful commands for labelling axes, etc, in Appendix B. One very useful command is
>> hold on
which allows you to plot a second graph over the first one on the same figure (that is, the
first graph is not erased when you repeat the plot command with another dataset). Then
>> hold off
releases the graph for a fresh start.

Creating a graph of an algebraic function y = y(x)

Matlab can also be used to create a graph of an algebraic expression, using a syntax very much like
that used in other computing languages such as FORTRAN or C. The following recipe will enable
you to graph functions with Matlab. You can use Matlab to plot the values of a function at all the
same x values as your experimental data, so that you can put both your data and a fitting function
onto the same graph.

• Start Matlab. If you have a data file in it, you can use the x values of the data (the column
vector xdata from the section above) to compute the values of the fitted function; otherwise,
you will need to start by creating a new column vector with a series of x values at which
Matlab will compute the y value of your expression. Let’s suppose that you don’t have a data

2
 file to start from, but simply want to plot y at a series of x values increasing from 10 to 20
with steps of 0.5.

• After the prompt in the Command Window, type

• Next you need to create a column vector with the y values corresponding to your x values.
Suppose that the equation that you want to plot is

y = 3.0 exp(−0.2x2 ) + 20.

You can compute the y values corresponding to the x values in your array xeqn by typing
after the prompt in the Command Window the following:
>> yeqn = 3.0 .* exp( -0.2 .* xeqn.^2) + 20
This equation instructs Matlab to create a column vector of y values called yeqn, with one
value evaluated for each element of the column vector xeqn.

• Matlab, like other computer programming languages, has various symbols for the algebraic
operations it needs. Four of these symbols appear in the Matlab equation above: the as-
signment operator “=” which tells Matlab to create a new variable yeqn and to compute
values for all of the values of the column vector xeqn; the “+” operator, which has the usual
meaning regardless of what kind of objects one adds (scalar number, matrices, etc); and
the multiplication operator “.*”, which tells Matlab to do element-by element multiplication
(that is, each numerical value of xeqn is taken in turn, and a corresponding value of yeqn
is computed, rather than doing some kind of matrix multiplication); and the operator “.ˆ”
which tells Matlab to raise xeqn (element-by-element) to the 2nd power.

• In general, Matlab (like other similar programmes such as the – free – Scilab) uses different
symbols for operators which act differently in matrix algebra than in simple scalar or array
arithmetic. In particular, matrix multiplication, division, and raising to a power are given
the symbols “*” , “/” , and “ˆ”, while operators for element-by-element array multiplication,
division, and raising to a power are “.*”, “./”, and “.ˆ”. Functions you may need mostly will
have familiar names such as log, exp, sqrt, cot, or abs. To get a list, type
>> help elfun

• You can now plot a graph of your function as before, or include it in a graph with your data
plot by turning hold “on” for the graph that has your data in it, and using plot to put the
theoretical curve over the data.

• The same technique can be used to create a column of data with is the difference between
your actual data and your fitting function – just create a new column vector
>> diff = xeqn - xdata
and plot it on the same graph.

There area a lot of small tricks that experienced Matlab users will be able to use in this process,
and which non-experienced users will gradually discover. Don’t be discouraged if your first couple
of tries don’t quite succeed. Try again. Experimentation will be rewarded.