Lab 3: MATLAB Tutorial

Zhuo Wang

Department of Bioinformatics & Biostatistics

Shanghai Jiao Tong University

(Parts of slides are from Maoying Wu)

MATLAB: Functions

Matrix manipulations

Visualization of data

Implementation of algorithms

Creation of user interfaces

Maoying Wu SysBio
MATLAB: Layouts

Current Folder

Editor

Command Window

Workspace

Command History

Maoying Wu SysBio
Using help or doc

help

help command

help on toolbar

doc command

doc query

Maoying Wu SysBio
Matlab classes: primitives

The three main classes in Matlab are

double: multidimensional array of double precision floating
point numbers
char: multidimensional array of ascii characters
logical: Multidimensional array of 1-bit (0/1) numbers

Eight other built-in primitive classes are

uint8, uint16, uint32, uint64: multidimensional array of 8,
16, 32 or 64-bit, unsigned integers
int8, int16, int32, int64: multidimensional array of 8, 16, 32
or 64-bit, signed integers
Every object in Matlab is a rectangular array, and arrays are at
least 2-dimensional (no notion of 1-d array).

Maoying Wu SysBio
MATLAB variables

Variables are dynamically typed.

Variable types can be changed

Basic operations
sum, subtract, multiply, divide

power, sqrt, sin, cos, etc.

Maoying Wu SysBio
Scalars, Vectors and Arrays

1 × 1 arrays are also called scalars.

1 × N arrays are also called row vectors.

N × 1 arrays are also called column vectors

Row and column vectors are called vectors

Maoying Wu SysBio
Getting Started: Assignment
   
1 2 1
A= ,b = , Ax = b, x = A−1 b
2 1 2×2
2 2×1

a = [ 1 2; 2 1 ];
a
b = [1;2];
x = a\b
x = inv ( a ) * b
 
1×1 2×2
A. ∗ A0 = (A is symmetric)
2×2 1×1

a .* a
 
1×1+2×2 1×2+2×1
A∗A=
2×1+1×2 2×2+1×1

a*a

Maoying Wu SysBio
Exercise

 
1
Let π = be the initial probability for two states, and
0
 
0.8 0.1
T = be the transition probability matrix between
0.2 0.9
states 1 and 2.

pi = [1; 0];
t = [0.8 0.1; 0.2 0.9];
for i =1:20 , pi = t * pi , end

What do you notice? Explain the phenomenon.

Maoying Wu SysBio
Vector: initialization

x = []; % empty vector

x = [ x rand ]; % append a random element

Maoying Wu SysBio
Vector: Sorting

% generate 10 uniform random variables

x = rand (10 , 1) ;
% sorting
[y , ind ] = sort (x , ’ ascend ’) ;
% what the indices mean ?
all ( y == x ( ind ) )
% sort the indices
[~ , ind2 ] = sort ( ind , ’ ascend ’) ;
% what the ind2 mean ?
all ( x == y ( ind2 ) )

Maoying Wu SysBio
Vector: norm

n
xip )1/p
X
||x||p = (
i=1

% generate 10 uniform random variables

x = rand (10 , 1) ;
% l1 - norm
y = norm (x , 1) ;
% how to compute l1 - norm ?
sum ( abs ( x ) ) == y
% l2 - norm
z2 = norm (x , 2) ;
% how to compute the l2 - norm ?
z2 == sqrt ( sum ( x .^2) )
% l - infinity norm
zinf = norm (x , inf ) ;
max ( abs ( x ) ) == zinf

Maoying Wu SysBio
Vector: Histogram

x = rand (10 , 1) ;
x = x (:) ;
% histogram : the mid of bin and the count in each bin
[ count , mid ] = hist ( x ) ;
% how to get the bin , 10 bins
y = linspace ( min ( x ) , max ( x ) , 11)
all ( y (1:10) + ( y (2) -y (1) ) /2 == mid )
% count
count2 = [];
for i =1:10 , count2 ( i ) = sum (x >= y ( i ) & x <= y ( i +1) ) ; end
% are these two equal ?
all ( count == count2 )

Maoying Wu SysBio
Special double array

ones(n, m, p)

zeros(n, m, p, q)

rand(n, m)

randn(n, m)

cell(n, m)

Maoying Wu SysBio
Random array

s = rng ;
A = randn (5) ;
rng ( s )
B = randn (5) ;
A == B

Maoying Wu SysBio
The size command

If A is an array, then size(A) is a 1 × ndims(A) double array. And

the (1, k) entry is the length of the k th dimension of A.

Examples

A = rand (5 ,6 ,8) ;
B = size ( A ) ;
[ a1 , a2 , a3 , a4 ] = size ( A ) ;
[ a1 , a2 ] = size ( A ) ;
size (A , 2)
size (A , 6)

Maoying Wu SysBio
The class command

If A is an array, then class(A) is character string, whose value is

the class of A (e.g. double, char, cell, struct, uint32, etc.)

Examples

class ( rand (5 ,6) )

class ( ’ this is a string 1234.56 ’)
class (5==6)
class ( class ( rand (5 ,6 ,8) ) )
class ( cell (4 ,5) )

Maoying Wu SysBio
Struct array

Like cell arrays, structure arrays (class name is struct) allow

you to group different data types together.
Examples:

student . Name = ’ Fred Smith ’;

student . SID = 12345678;
student . Scores = [62 78 93 61];
size ( student )
class ( student )
fieldnames ( student )
isfield ( student , ’ Name ’)
isfield ( student , ’ Names ’)
student . Scores (3) = 99;
student . Scores ([1 end ])
c = struct ( ’ fld1 ’ , A , ’ fld2 ’ , B ) ;

Maoying Wu SysBio
File: opening, reading and writing

in_file = fopen ( ’ Name_of_File ’ , ’r ’) ; % read only

if ( in_file == -1)
error ( ’ oops , file can ’t be read ! ’) ;
end

[ number , count ] = fscanf ( file , ’%d ’ , 1) ;

while ( count == 1) % while we have read a number

fprintf ( ’ We just read % s \ n ’ , number ) ;

[ number , count ] = fscanf ( file , ’% d ’ , 1) ; % attempt to

read the next number

end

Maoying Wu SysBio
fprintf

fprintf ( ’% format \ n ’ , variable ) ;

%s: a string
%c: a single character
%d: a whole number
%f: a floating point number
\n: a newline
\t: a tab
\\: a slash
%%: a percent sign

Maoying Wu SysBio
Flow control

for loops

while loops

if/elseif/else conditionals

switch/case/otherwise

try/catch
Do NOT forget the end at the end.

Maoying Wu SysBio
while-loop and for-loop

Listing 1: while-loop
while < condition >
if < something - bad - happens >
break
else
% do something useful
end
% do more things
end

Listing 2: for-loop
for < condition >
if < something - bad - happens >
next
elseif
% do something useful
else
% do something useful
end
% do more things
end

Maoying Wu SysBio
Graphics in MATLAB

t = 0:.3:10;
y = sin ( t ) ;
plot (t , y ) ;

Maoying Wu SysBio
3D graphics in MATLAB

[x , y ] = meshgrid ( -2:.2:2 , -2:.2:2) ;

z = x .* exp ( - x .^2 - y .^2) ;
surf (x ,y , z )

Maoying Wu SysBio
Definition of Functions

function [N , M , Z ] = read_al ignment ( inputfile )

% reads alignment from inputfile , removes inserts and
converts into numbers
align_full = fastaread ( inputfile ) ;
M = length ( align_full ) ;
ind = align_full (1) . Sequence ~= ’. ’ & align_full (1) .
Sequence == upper ( align_full (1) . Sequence ) ;

N = sum ( ind ) ;
Z = zeros (M , N ) ;
for i =1: M
counter = 0;
for j =1: length ( ind )
if ( ind ( j ) )
counter = counter + 1;
Z (i , counter ) = letter2number ( align_full ( i ) .
Sequence ( j ) ) ;
end
end
end

Maoying Wu SysBio
Functions of the MATLAB Optimization

Linear and Quadratic minimization problems

linprog - Linear programming
quadprog - Quadratic programming
Nonlinear zero finding (equation solving)
fzero - Scalar nonlinear zero finding
fsolve - Nonlinear system of equations solve (function solve)
Linear least squares (of matrix problems)
lsqlin - Linear least squares with linear constraints.
lsqnonneg - Linear least squares with nonnegativity constraints.
Nonlinear minimization of functions:
fminbnd - Scalar bounded nonlinear function minimization.
fmincon - Multidimensional constrained nonlinear minimization.
fminsearch - Multidimensional unconstrained nonlinear minimization, by
Nelder-Mead direct search method.
fminunc - Multidimensional unconstrained nonlinear minimization.
fseminf - Multidimensional constrained minimization, semi-infinite
constraints.

Maoying Wu SysBio
Linear Programming
Linear programming problem

min f (x) = c T x
s.t.
Ax ≤ a
Bx = b
lb ≤ x ≤ ub;

Objective function: f (x) = c T x

Linear constraints: inequation (Ax ≤ a), equation (Bx = b),
box (lb ≤ x ≤ ub).

[x , fval , exitflag , output , lambda ] = linprog (f , A , a , B , b , lb

, ub , x0 , options ) ;

Maoying Wu SysBio
Algorithms under linprog

options = optimset ( ’ linprog ’) ;

a simplex algorithm (medium-scale)

options = optimset ( ’ LargeScale ’ , ’ off ’ , ’ Simplex ’ , ’ on ’

);

an active-set algorithm (medium-scale)

options = optimset ( ’ LargeScale ’ , ’ off ’ , ’ Simplex ’ , ’ off

’) ;

a primal-dual interior-point method (large-scale)

options = optimset ( ’ LargeScale ’ , ’ on ’) ;

Maoying Wu SysBio
Linear programming: Example 1

max 2x1 + 3x2

s.t.
x1 + 2x2 ≤8
2x1 + x2 ≤ 10
x2 ≤3
x1 , x2 ≥0

c = [ -2 , -3] ’;
A = [1 ,2;2 ,1;0 ,1];
a = [8 ,10 ,3] ’;
options = optimset ( ’ linprog ’) ;
options = optimset ( ’ LargeScale ’ , ’ off ’) ;
xsol = linprog (c , A , a , [] , [] , [0 ,0] ’ , [] , [] , options )

Maoying Wu SysBio
Linear programming: Exercise

 
1 1 1 1 1 1 10
(A|a) =
5 0 −3 0 1 0 15
 
1 2 3 0 0 0 5
 0 1 2 3 0 0 7 
max cT x (B|b) = 
 0 0 1 2 3 0 8


s.t. 0 0 0 1 2 3 8
Ax =a
 
3 0 0 0 −2 1 5
(D|d) =
Bx ≥b 0 4 0 −2 0 3 7
Dx ≤d 
−2 7 1


lb ≤ x ≤ ub  0 2 −2 
 −1 2 3 
 
(lb|ub|c) =
 −1 3 −4 
 
 −5 4 5 
1 10 −6

Maoying Wu SysBio
Linear programming: solution.m

A =[1 ,1 ,1 ,1 ,1 ,1;5 ,0 , -3 ,0 ,1 ,0]; a =[10 ,15] ’;

B1 =[1 ,2 ,3 ,0 ,0 ,0; 0 ,1 ,2 ,3 ,0 ,0;...
0 ,0 ,1 ,2 ,3 ,0;0 ,0 ,0 ,1 ,2 ,3]; b1 =[5 ,7 ,8 ,8]; b1 = b1 (:) ;
D =[3 ,0 ,0 ,0 , -2 ,1;0 ,4 ,0 , -2 ,0 ,3]; d =[5 ,7]; d = d (:) ;
lb =[ -2 ,0 , -1 , -1 , -5 ,1] ’; ub =[7 ,2 ,2 ,3 ,4 ,10] ’;
c =[1 , -2 ,3 , -4 ,5 , -6]; c = c (:) ;
B =[ - B1 ; D ]; b =[ - b1 ; d ];
[ xsol , fval , exitflag , output ]= linprog (c ,A ,a ,B ,b , lb , ub )
fprintf ( ’% s % s \ n ’ , ’ Algorithm Used : ’ , output . algorithm ) ;
disp ( ’ = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = ’) ;
disp ( ’ Press Enter to continue ’) ; pause
options = optimset ( ’ linprog ’) ;
options = optimset ( options , ’ LargeScale ’ , ’ off ’ , ’ Simplex ’ , ’ on
’ , ’ Display ’ , ’ iter ’) ;
[ xsol , fval , exitflag ]= linprog (c ,A ,a ,B ,b , lb , ub ,[] , options ) ;
fprintf ( ’% s % s \ n ’ , ’ Algorithm Used : ’ , output . algorithm )
fprintf ( ’% s ’ , ’ Reason for termination : ’)
if ( exitflag )
fprintf ( ’% s \ n ’ , ’ Convergence . ’) ;
else
fprintf ( ’% s \ n ’ , ’ No convergence . ’) ;
end
Maoying Wu SysBio
User Interface (UI)

Maoying Wu SysBio