An Introduction to Matlab - uni-duesseldorf.decpclab.uni-duesseldorf.de/.../presentation_matlab.pdfAn Introduction to Matlab Why Matlab? zhigh-level scripting language => rapid prototyping

by Jo Bergs

An Introduction to Matlab

Why Matlab?high-level scripting language => rapid prototypingcompatibility with established languages (java, C++, …) allows recoding of speed-limiting modulesexcellent interface and documentationlarge userbase with well-documented, ready-to-use applicationswell-suited for analysis of complex data

Interface – Main Menu

Variable Overview

Command History

Access to Toolbars

Console

Generated Plot

Working Directory!

Interface - Help

• Access via ordered list of content, indexed list or search• Search supports boolean operators• Demos including code• try „Getting Started“ and „Programming Fundamentals“• Inline help: in console, type <help command> (self-completing)• visit www.mathworks.com: user-made scripts and further help• code of built-in commands: <edit command>

Interface - Variable Editor• in variable overview: new variable -> doubleclick -> rightclick -> paste tabular data

OOF tables are imported

Interface - Designing GUIs

Property Inspector: by doubleclick on Object

Running Instance

GUIDE: GUI Designer

• WYSWG GUI Designer• Property Inspector defines Function „Connectivity“• Allows Design of custom-tailored data analysis tools

Compatibility• interoperable for windows and linux• callable from Fortran, C, Java and Python• Interface to Fortran, C and Java• allows prototyping in Matlab and subsequent replacement with faster languages• Full access to specialized libraries

Emulate a Click with Java:

(Mouse location set by Matlab, Input is a Java Robot Object)

Data Structures - Matrices• Matrix creation: embrace in < [ ] >, seperate values with < > or < , > and rows with < ; >

• invert Matrices with < ‘ >; concatenate with < [A; B ] > (size!)

• creating ranges: < Start:Stepsize(default=1):End >

Data Structures - Matrices

• creating random values: < rand(rows, cols) > (uniform) < randn(rows, cols) > (normal)

• creating multidimensional matrices by appending dimensions( < : > sets Start = 1 and End = row/column length)

• accessing values: < (row, column) >; ranges may be used

Data Structures - Matrices

• < sum(matrix) > sums column-wise; < mean(vector) > and std(vector, 1) forcalculatin the mean resp. Standard deviation

• putting it all together: evaluating docking results

• < sort(vector) > returns sorted vector and permutation index, which can beapplied to another vector via < vector(IDX) >

Data Structures – Cell Arrays• matrices can store boolean, integer, float and alphanumeric

• asymmetric data should to be stored in cell arrays with < {data1, data2; data3} >

• < cellplot(cellarray) > displays current structure of a cell array

Data Structures – Structs• container for cell arrays, matrices and single values

• allows named subvariables -> dictionary

• creation: < s = struct(‘name1‘, ‘content1‘, …) >

• access: column-wise with parenthesis or row-wise via name space

Programming - Basics• if and switch statements:

• for and while loops:

• error handling: (explicit error classes implemented)

• function definition (no indent needed):

Advanced Programming - GA• plot a function:

• define a fitness function and execute the GA:

Advanced Programming – Speed it Up!• vectorization allows usage of fast C libraries:

Faster!

• Runtime Profiler: < profile viewer > allows identification of expensive functions

Select Script

Time consumed by single function(dark blue: excluding subfunctions)

Advanced Programming – Debugging• editor with built-in debugger: on-line error warnings;

set break points; hover over vars to see value

Miscellaneous• use < … > as linebreak in code

• < display(variable) > prints variable value

• < whos variable > displays variable properties (no parentheses!)

• < save(‘path.mat‘) > stores all workspace variables as matlab .mat file

• < load(‘path.mat‘) > retrieves them

• < save(‘path.mat‘, ‘varname1‘,…) > stores/loads only specified vars

• functions for handling pdb files at MatlabCentral

• In Editor: File -> Publish allows publication of code with output or errors

Example – Score/RMSD• import docking results from .xls (!compatibility!):

• extract rmsd and score: cast to char, convert to floats

• sort both by rmsd:

Example – Scatter Plot• finally, plot it!

• no funnel-shapedness : (

Tune the Plot

Example – Plot Modifikation• label axis, set ticks, change markers, draw lines,…

Store the Plot

Example – Store your Work• File -> Save: stores the plot as .fig• File -> Generate M-File: stores code and allows to plug in new Data easily

Toolboxes - GA• toolboxes consists of functions and easy-to-use GUIs• GA and Direct Search toolbox: contains GA, SA and PSO

• parameter tuning • evaluation of convergence• comparison of search algorithm performance on a specific task

Fitness Functionand # Parameters

GoGoGo!

Algorithm

Other Toolboxes• stochastics toolbox: Cluster Analysis, Classifiers, Markov Models, …

• optimization toolbox: linear programming, model-fitting (least square)

• symbolic math toolbox: linear algebra, equation solver, calculus

• bioinformatics: sadly, no structural bioinformatics, but sequence analysis etc.

Our Book• Content:

− Solving linear Equations

− Interpolation

− Least Square Fitting

− Differential Equations

− Fourier Transformation

− Eigenvalues and PCA

Eyes Still Open?

Thank you for your attention!