Octave versus Matlab

Some small differences exist between Octave and Matlab. The following list is not meant to be exhaustive, but aims the solve practical problems in conversions

Octave	Matlab	remarks
`\` or `...`	`...`	catenate following line; Emacs don't respond nice on `...`
`#` or `%`	`%`	start comment on a line; Emacs don't respond nice on `%`
`"txt"` or `'txt'`	`'txt'`	text `txt`; Emacs don't respond nice on Matlab's version
yes	no	end-of-function required to close a function
yes	no	function definition allowed in script file
`exist("in")==0`	`in==[]`	check existence of input in: `function f = fnname(in)`
	`end`	last element in a vector: `a=0:5; a(end)`
`x`	`[t,x]`	result of ODE-solver: `x = lsode("nm",x0,t)` versus `[t,x] = ode45("nm",t,x0)`
`(x,t)`	`(t,x)`	arguments for user defined functions for ODE-solver. Matlab's output `t` and `x` are large if the input `t` has 2 elements.
`[x,err]`	`[x,flag,err]`	result of root finder: `fsolve("name",x0)`
`e^()` or `exp()`	`exp()`	exponent
`!=` or `~=`	`~=`	is not equal to
`x=y=1;`	`x=1;y=1;`	Octave's shortcut not allowed in Matlab
`i++;`	`i=i+1;`	Octave's shortcut not allowed in Matlab
`clear`	`clear all`	erase all variables
`clg`	`clf`	clear plot

Other differences:

Octave versions earlier than 2.1.50 handle variable number of inputs different from Matlab; Octave's manual (2005/02/01) describes the old way, which no longer works
Unused variables in Octave functions, that are declared global, can have numerical values. In Matlab they exist, but are empty (0 x 0); clear does not affect the values of global variables inside functions
plot text before plot in Octave, after plot in Matlab. Many uses of gnuplot don't work in Matlab. Matlab does not have a facility for two x-axes in a single plot; it does have rotate3D which allows the users to rotate a 3-dim plot with the mouse.
initiate multiple plots with multiplot in Octave, directly call subplot in Matlab
Matlab knows >2-dimensional arrays, Octave 2-dim only (i.e. matrices)
set path with LOADPATH=" "; under Octave and in the menu-bar under Matlab

The Student Edition of Matlab is sufficient for most applications, although it has the problem of an upper boundary of array sizes, which is easily exceeded when integrating a system over a longer time interval. Another problem is that the Student Edition has no routine for solving vector-valued x from 0 = f(x). If you go for the professional edition of Matlab, you will probably also need the optimization toolbox, which comes under a separate license requirement.