Matlab - introduction

Given that NeuroElf is a Matlab toolbox with several features only available via the command line (at present at least), I want to give a short overview of Matlab, its features and how the user may be able to make better use of the available functionality.

First of all, Matlab can be regarded as a development environment with features such as a built-in GUI to inspect the current state (variables in the present workspace and their contents), editor with syntax highlighting, a debugging mechanism to run code step-by-step, and a powerful command line prompt.

But Matlab also comes with its own language (which shares many concepts, syntax elements, and features with other languages), and I want to highlight some of the most important features:

• the language is easy to use for prototyping, given its forgiving and user-friendly nature
  • variables don't require a declaration but can be defined and have their datatype and dimension changed at any point within a program
  • numeric variables are, by default, all created with a standard datatype of double, requiring little (if any) knowledge of numeric datatypes
  • storage of and access to variable values is organized in workspaces
  • identifiers can be overloaded (e.g. built-in functions can be replaced by user defined ones), globally or separately in each workspace
• array indexing is available in multiple ways to simplify element selection (see datatypes and indexing page)
• array-based operations are highly optimized for speed and mostly support multiple CPU cores
• memory management is built-in (workspace-related garbage collection on variables no longer in use)
• Matlab comes with a vast repository of built-in and built-upon functions for mathematical computations, data in- and output, as well as visualization
• functions can return more than one output value/variable (without having to resort to compound/struct variables)
• the number of keywords of the basic language is extremely small (20 english words, besides operator symbols), as compared to about 60 for C++ or more than 100 for Visual Basic
• user-defined datatypes are (usually) based on a built-in datatype (struct) and allow object-oriented development
• with a built-in interface to compiled code (MEX), Matlab allows to translate computationally intensive operations (recursion, etc.) into compiled code for speedup
• built-in file format to store the content of variables in a binary format (incl. compression and platform independence)

On top of these features, Matlab comes with many, many built-in functions from a variety of fields, among them are

• mathematical functions (sum, mean, median, mode, standard deviation/variance, discrete derivatives, convolution, minimum, maximum, etc.)
• list and set related functions (sort, unique, intersect, setdiff, union, etc.)
• rudimentary statistical functions (histogram, correlation coefficient, detrending, etc.)
• matrix operations (inverse, pseudo-inverse, SVD, decomposition, etc.)
• graphical output functions (line, polygon, shape, surface, and scatter plot, annotations, etc.)
• GUI functions (figures, controls, callback handling, etc.)

These features make it usually relatively efficient to create code with complex functionality with just a few lines of code.

Mathworks additionally provides a user-defined toolbox area on their webserver, File Exchange, where users can upload their functions (or sets of functions, called toolboxes) which often allows users to solve even complex problems with a minimal effort, given that someone has already solved it before.

When Matlab is started (in its default configuration), the following “windows” (components) are available:

• command line (prompt): this allows to enter commands (expressions, assignments, function calls, etc.), call a system process, and control the debugging mechanism (workspace control, etc.)
• command history: a list of the previously executed commands (incl. those of previous Matlab sessions)
• workspace inspector: this window shows a list of variables in the current workspace (name, datatype, size, min/max value)
  • if a variable name is double-clicked, a variable-editor window is added (or selected if already visible), allowing to inspect and alter variable content graphically
• current folder widget: file/folder browser showing the content of the “present working directory” (pwd), allowing to navigate the folder structure and select files
  • a “file detail” view, showing the content of known file types (MAT files)

Additionally, Matlab comes with a built-in Editor, supporting syntax highlighting and augmenting several of the core features of Matlab, as well as a Profiler, a tool to identify bottlenecks in code, i.e. a help to improve the run-time properties (time and memory consumption) of user-written code.

The command line allows the user to enter commands and inspect returned values (text output intermixed with input). Matlab shows a prompt (>>) during normal operation and an extended prompt for debugging (K>>). The command line offers a few additional functions which are extremely helpful:

• pressing the cursor-up key allows to cycle through previous commands, and if some text is entered before, only entries matching this text are presented
• the tab-key does not produce a tab-character (indentation) but instead Matlab attempts to auto-complete a series of letters to a “known expression” (variable, field, function, or file name, depending on context)
• when entering expressions with parenthesis (precedence of operations, function arguments, etc.) or brackets/curly braces, Matlab highlights matching pairs to confirm the completeness of the expression
• if an expression is incomplete (e.g. a for loop is constructed, but the final end keyword is missing), Matlab expects additional input on the next line (after pressing return)
• whenever an error is encountered while processing a line of input, Matlab indicates the exact position of the problem with a vertical line below the input (pipe character)
• output in the command window supports hyperlinks (in help texts for instance), which gives the user to immediately select a topic of reference for additional information
• the display function (which is automatically called when an expression is not terminated with a semicolon) provides auto-formatting for all built-in datatypes (incl. arrays)
• to change the current folder or load a MAT-file, drag-and-drop operations can be performed with either Finder (Mac) or Explorer (Windows), by dragging a folder or MAT-file name into the command prompt window

The history window is particularly useful when a series of commands is to be converted into an M-file (batch of function file grouping a set of commands/function calls into a new functionality). This can be achieved by selecting several entries (line by line selection) and then using the context menu (right-click or CTRL-click). To quickly repeat one specific instruction, an entry can also be double-clicked.

Equally, the workspace window offers a context menu, which allows to rename variables, delete them from the workspace (same as clear VARIABLE), save the content of selected variables to disk, or alter their content.

The current folder window is a feature-rich file browser, also with additional functionality in the context menu.