Matlab functions and utilities

Matlab documentation: automatic generation of m-file documentation in XHTML
M-file Template: structure and conventions for a good structure m-file
prprintf - pretty output of all Matlab data types
plotmesh - 2-D and 3-D mesh plots of multi-dimensional functions ( with many useful extensions)

Automatic generation of Matlab m-file documentation in XHTML

The perl script for the automatic generation of m-file documentation in XHTML is fully described in the Perl section (Matlab documentation).

M-file Template

Template m-file: m-file-template.m
- This m-file contains two features:
  - a fully documented template for documenting m-files
  - a real template applicable to be copied into own m-files (find it at the end of this file)
  The first part is used to describe all the elements of a well-documented m-file. The second part can be copied into your own m-files and you can fill out the blanks, giving you a head-start.
- This template represents my working style (a result of 6+ years of matlab experience, including developing the GEATbx).
- Nevertheless, it reflects my personal opinion. I am always open to suggestions, enhancements and discussions (hartmut@pohlheim.com).
- This m-file template is optimally suited for the automatic generation of m-file documentation using my special perl script mtree2html2001.bat described in the Perl section at Automatic Matlab documentation.
- For a full description of all suggestions and the structure please read the m-file template.

prprintf - pretty output of all Matlab data types

M-file: prprintf.m
- Prprintf takes the given variable and produces a string containing the contents of the variable in a pretty printed format.
- If the variable contains structures or cell arrays, this function is called recursively for every element, even structure arrays. An internal function handles the special case of structures appropriately. The field names are printed as well, every element on a single line.
- All this produces a good looking string of the elements in 'VarData' without worrying to much about the format of the elements.
Syntax: [PrString, ErrMsg] = prprintf(Format, Between, VarData)

Input parameter:

    Format    - (optional) String containing formatstring, uses same
                syntax as sprintf
                   if omitted, '%g' is used
    Between   - (optional) String containing characters to put
                between elements in VarData
                   if omitted, '   ' is used
    VarData   - Variable(s) to pretty print, may be a scalar,
                a character, a string, a matrix, a cell array or
                a structure with all of this, even mixed up

    If only 2 input parameters are provided, the first input is
    Format and the second input VarData, Between is set to an empty
    string.

    If only 1 input parameters is provided, the first input is
    VarData, Format and  Between are set to default values.

Output parameter:

    PrString  - String containing the elements in VarData pretty
                printed
    ErrMsg    - String containing possible error message, then PrString
                is an empty string
                ErrMsg is an empty matrix, when no error occured

Examples:

    >>prprintf('%g', '/', [1.02 4.07 3.0532])
       1.02/4.07/3.0532

    >>prprintf('%.3g ', [1.02 4.07; 3.0532 5.875; 3.324 4.123])
       1.02 4.07   3.05 5.88   3.32 4.12

    >>prprintf('%s', ' ', ['eins';'zwei';'drei';'vier'])
       eins zwei drei vier

Try prprintf with cell arrays and structures as well! Especially the output of structures is very informativ, the whole structure with all elements is printed!

plotmesh - 2-D and 3-D mesh plots of multi-dimensional functions

M-file: plotmesh.m
- This function takes the name of an objective function and produces a variation mesh plot using 2 dimensions or an variation line plot using one dimension. This gives an impression of the objective function or of the area around a good (or selected) point. The function may be used for a variation analysis or for demos.
- See the m-file itself for all options and features.
Syntax: plotmesh(OBJ_F, Bounds, Points, varargin)

Input parameter:

    OBJ_F     - string with function name of objective function
                if omitted, 'objfun1' is assumed
    Bounds    - Matrix containing the boundaries of the objective
                function for mesh plotting, same format as in
                objective function.
                Bounds=[lower_bound_x1, lower_bound_x2;
                        upper_bound_x1, upper_bound_x2]
                if Bounds has more than 2 columns, than first one or two
                    columns with Bounds(2)-Bounds(1) are used for variation
                    plot
                if Bounds are omitted, the objective function is called with
                   first parameter empty and second parameter 1 to get the
                   default boundaries
    Points    - Scalar/vector containing number of mesh grid points.
                If 2 scalars are provided, the first is used for
                points in x1 and the second for points in x2.
                If 1 scalar is provided it is used for points in x1 and x2
                if ommitted, 25 points in both directions are assumed
                "Undocumented features of Points"
                if Points contains more than one row, each row has the
                following meaning:
                first row:  number of mesh grid points
                second row: value of optimum for first and second dimension
                            a doted line is plotted along this value
                third row: define limits for y-values of 2-D plots
                           (doesn't work at the moment for 3-D plots)
    varargin  - additional parameters for the objective function

Output parameter:
```
    none
      
```

Examples: (graphics representing the examples will be added later)

    % plot with default parameters
       plotmesh('objfun1');
    % use default boundaries, use 20 grid point in both dimensions
       plotmesh('objfun1', [], 20);
    % same boundaries for both dimensions, use 100 grid points per dimension
       plotmesh('objfun1', [-10; 10], 100);
    % define different points for every dimension
       plotmesh('objfun1', [-10, -5; 10, 5], [40, 30]);
    % define different points for every dimension, plot lines at x=0 and
    % y=0 to indiciate optimum
       plotmesh('objfun8', [-10, -5; 10, 5], [30, 30; 0, 0]);
    % visualize variable 4 and 7 of multidimensional function, set the remaining
    % variable 1,2,3,5,6 and 8 to zero
       plotmesh('objfun10', [0,0,0,-20,0,0,-10,0;0,0,0,20,0,0,10,0], [50, 50; 0, 0]);

A number of example objective functions is provided as well:
- objfun1.m
- objfun2.m
- objfun6.m
- objfun7.m
- objfun8.m
- objfun9.m
- more objective functions are part of the GEATbx, please look in m-file documentation, all objective functions are named obj*.m

Publications (long descriptions)

Perl utilities

Main page

Publications (compact) (long)

Matlab utils

Perl utils

Research