Introduction to Computing with MATLAB

Arun Prakash
School of Civil Engineering Purdue University.

Contents
1 Introduction to Computing
1.1 1.2 1.3 2.1 Computing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Computer Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Basic Matrix Algebra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Legends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4
4 5 7

2 MATLAB Basics: Datatypes, Arrays, Input/Output, Plotting

Datatypes in MATLAB . . . . . . . . . . . . . . 2.1.1 Variables . . . . . . . . . . . . . . . . . . 2.1.2 Arrays . . . . . . . . . . . . . . . . . . . 2.1.3 Initialization of Variables and Arrays . . 2.1.4 Multi-dimensional Arrays . . . . . . . . 2.1.5 Subarrays . . . . . . . . . . . . . . . . . Matrices Operations vs. Arrays Operations . . . 2.2.1 Matrix operations . . . . . . . . . . . . . 2.2.2 Array operations . . . . . . . . . . . . . Input and Output (I/O) of Data . . . . . . . . . 2.3.1 Input the data from keyboard . . . . . . 2.3.2 Output of Data to the Screen . . . . . . 2.3.3 I/O through Data Files . . . . . . . . . . Introduction to Plotting . . . . . . . . . . . . . 2.4.1 The plot command . . . . . . . . . . . . 2.4.2 Title, Label, Grid and Text . . . . . . . 2.4.3 Multiple curves on one plot . . . . . . . 2.4.4 Line Color, Line Style, Marker Style, and 2.4.5 Controlling x- and y-axis Plotting Limits 2.4.6 Controlling Plot features using the GUI . Branching . . . . . . . . . . . . . 3.1.1 The Logical Data Type . . 3.1.2 Relational Operators . . . 3.1.3 Logical Array Masking . . 3.1.4 Logical Operators . . . . . The if branch . . . . . . . . . . . 3.2.1 The Nested if Statement . . . . . . . . . . . . . . 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2.2 2.3

2.4

8 8 9 10 11 12 13 13 14 15 15 15 17 18 18 18 20 21 22 23

3 Branching Statements
3.1

3.2

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

24

24 25 25 26 27 29 30

3.3 3.4 4.1 4.2 4.3

The swith statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . MATLAB Debugger . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Top-Down Design Techniques . . . . . . . . . . . Loops . . . . . . . . . . . . . . . . . . . . . . . . The for Loop . . . . . . . . . . . . . . . . . . . . 4.3.1 The general form of the for Loop . . . . . The while Loop . . . . . . . . . . . . . . . . . . . Simple Applications . . . . . . . . . . . . . . . . . Timing, Preallocation and Vectorization of Loops The rek and ontinue Statements . . . . . . . Nested Loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

31 32

4 Loops

33

4.4 4.5 4.6 4.7 4.8 5.1

33 35 36 36 38 39 41 42 43

5 More Plotting and Graphics

5.2

Additional Types of Two-dimensional Plots . . . . 5.1.1 Other Useful Plotting Functions . . . . . . 5.1.2 Logarithmic Plots . . . . . . . . . . . . . . 5.1.3 Subplots . . . . . . . . . . . . . . . . . . . 5.1.4 Creating Multiple Figure Windows . . . . 5.1.5 Exporting a Plot as a Graphical Image . . Three-dimensional Plots . . . . . . . . . . . . . . 5.2.1 plotQ function . . . . . . . . . . . . . . . 5.2.2 The meshgrid, mesh and surf commands 5.2.3 The Contour functions . . . . . . . . . . . 5.2.4 Generating Animations of Plots . . . . . .

45

46 46 47 47 48 49 50 50 51 52 53 54 59 60 60 61

6 User Dened Functions, Recursion

6.1 6.2 6.3 6.4 6.5 7.1 7.2 7.3 7.4 7.5

Introduction to Matlab Functions . . . . . . . . . . . . Variable Passing in Matlab: The Pass-by-Value Scheme Optional Arguments . . . . . . . . . . . . . . . . . . . Function of functions . . . . . . . . . . . . . . . . . . . Recursive Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

54

7 External File Input/Output

The textred@A Function . . . . . . . . . Introduction to MATLAB File Processing File Opening and Closing . . . . . . . . . 7.3.1 The fopen Function . . . . . . . . 7.3.2 The flose Function . . . . . . . . File Positioning and Status Functions . . . I/O Functions for Formatted Text Data . . 7.5.1 The fprintf Function . . . . . . . 7.5.2 The fsnf Function . . . . . . . . 7.5.3 The fgetl and fgets Functions . .

64

64 65 65 65 67 68 69 69 70 70

7.6

I/O Functions for Binary Data . . . . . . . . . . . . . . . . . . . . . . . . . . 7.6.1 The fwrite Function . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.6.2 The fred Function . . . . . . . . . . . . . . . . . . . . . . . . . . . Matrix Algebra . . . . . . . . Data Analysis . . . . . . . . . Polynomials . . . . . . . . . . 8.3.1 Roots . . . . . . . . . 8.3.2 Curve Fitting . . . . . Integration . . . . . . . . . . . Dierential Equations . . . . . 8.5.1 IVP Format . . . . . . 8.5.2 ODE Solvers . . . . . . 8.5.3 Basic Use . . . . . . . Advanced MATLAB Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

71 71 72

8 Numerical Methods in MATLAB

8.1 8.2 8.3 8.4 8.5

73

8.6

73 74 76 76 76 78 78 78 79 79 84

9 Application to Civil Engineering: Structural Dynamics

85

Chapter 1 Introduction to Computing

Using computers to solve (engineering) problems of our interest is called Computing. In this process, we develop computational tools that help us do our jobs better and faster. Computing is dierent from Computer Science. Computer Scientists try to design the Computer itself and develop programming languages that we, as programmers, can use for our own engineering applications.

1.1 Computing
Why do we need Computing?

Volume of data and societal needs have grown beyond human capabilities Human error, consistency of results, speed and accuracy Examples: Banking, Automotive, Manufacturing, Communication etc. Questions: Reliability, Fault tolerance, robustness, backup Caveat: Utilization vs. Dependence; we are responsible for the technology we create and use.
Types of Computing

On-site Data Analysis and response systems in real-life applications: Structural Health Monitoring and Control Water quality management Earthquake Engineering Direct simulation of physical phenomena (Scientic Computing) Analysis & design of systems such as buildings, bridges, machines etc. Verify & Validate current and future theories of physics - Simulate stu we cannot measure or observe - subatomic particles, core of stars, even origin of the universe!!
4

Components of a computer

Hardware CPU - Binary (0,1) instructions go in ; Binary output obtained Memory - ROM, RAM, Hard Disk, External Storage Input Devices - Keyboard, Mouse, Touch Screen etc. Output Devices - Monitor, Printer etc. Software

 Operating system

Windows, MacOS, Linux, Unix, Sun Solaris, Internet Explorer, Media Players, Photoshop, Adobe Acrobat Programming languages: C/C++, Fortran, Pascal etc. MATLAB Your programs

 Applications Programs

1.2 Computer Programming

What is programming

Dening the set of operations for a computer to perform (telling the computer what to do). Computers understand only certain binary instructions. So computer scientists developed more user friendly languages that translated (compiled / interpreted) into binary code. We need to learn these languages in order to communicate with the computers by writing programs.

Structure of a program:

 Get input  Compute - operate upon the input data to generate meaningful information  Output the results
Some Essentials of Programming

     

Data Structure for Memory management - Variables, Array, Pointers Conditional Branching Statements Loops File Handling Input / Output Graphics

MATLAB - Matrix Laboratory

Advantages:
Relatively easy to use and good for beginners, GUI Predened functions for a lot of Mathematical operations: Matrix Algebra, Solving system of equations, Eigenvalue computations Symbolic Mathematics: Algebra, Dierentiation, Integration Additional Toolboxes Plotting / Imaging / Visualization of Results - device independent Combine Languages, C/C++, Fortran Dierent platforms run the same MATLAB program / code Demos : Membrane, 3D peaks, Bar with notch

Disadvantages:
Interpreter based : Slower, but can be compiled Kernel overhead - not suitable for very large problems Limited advanced programming features: Pointers, Pass by Reference, Object-oriented
The MATLAB Environment

Desktop Command window bb is called the 'command prompt' Arithmetic: CD ED BD GD Line continuation (Ellipsis) FFF Command History window Workspace browser: Variables whosD lerD lD lf Path Browser - Variable, m-le in current directory, rst occurrence whih Editor window : m-les as scripts Figure windows plot

 Help helpD lookfor Getting started section Start Button Other commands CTRL-c : Cancel or Interrupt Operation (when MATLAB 'hangs') ! : execute command on MS-DOS or Unix shell prompt diry
Built-in functions in MATLAB

Elementary Math Functions:

abs( ) sqrt( ) factorial( ) exp( ) log( )

; log10( )

Trigonometric functions

sin( ) cos( ) tan( ) cot( )

; ; ; ;

asin( ) acos( ) atan( ) acot( )

Hyberbolic functions

sinh( ) tanh( )

; ;

cosh( ) coth( )

1.3 Basic Matrix Algebra

Refer to Matrices Handout.

Chapter 2 MATLAB Basics: Datatypes, Arrays, Input/Output, Plotting

Before we can write programs, it is important to understand how MATLAB uses and operates on dierent types of data.

2.1 Datatypes in MATLAB

The two most common data types in MATLAB are Numeric and character data (Refer to MATLAB help for details on other types of data). 1. Numeric Data is stored in doule precision format by default. Double precision numbers use 64 bits (binary digits - 0, 1) and can store a number with 15 to 16 signicant digits of precision (mantissa) and 10308 to 10308 as exponent. Double precision data types can be real, imaginary or complex. 2. Character data types are stored in 16-bit value representing a single character. Strings are a collection of characters where each character uses 16 bits. Example hr@TSA is 'A' and hr@WUA is 'a'.
2.1.1 Variables

A Variable is user given name that refers to a certain location in the computers memory where MATLAB stores data. The user can access that data by specifying the variable name associated with it.

Rules:

1. Variable names are case sensitive. Example: vrD rD e are all dierent. 2. Must begin with an alphabet followed by alphabets, numbers and the underscore _ character. 3. MATLAB can distinguish variable names upto 63 characters in length.

Examples

x a IH x a x C I a PH C PHi hrterI a 99 hrterP a 9I9 ghrrI a hr@WUA trrI a 9his is gii IS9 stringvrile a 9hts ool39

Predened Variables in MATLAB (not protected: can be overwritten)

pi iD j snfD xnD ns relmxD relminD eps lokD dte
are over-written.
2.1.2

Note: Choose the names of your variables so that no inbuilt predened variables or functions

Arrays

MATLAB treats all data as arrays. An array is a `collection of data' (any data - numbers, characters etc.) that is stored in continuous locations in the computers memory. All variables refer to arrays in the computers memory. Even scalars are actually treated as 1 1 array. Arrays are primarily of two types: Vectors (dimension 1) and Matrices (2 or more dimensions). The size of an array is the number of rows and columns in an array. (For higher dimensional arrays it includes the extent of all dimensions).

Example:

1 2 a = 3 4 5 6 b= 1 2 3 4 1 c = 2 3

3 2 matrix 1 4 array, row vector

3 1 array, column vector

COMMANDS: length@AD size@A size@A gives the size of a specic matrix . length@A returns the length of a vector or the longest dimension of a 2-D array.

Individual elements in an array are accessed using the row and column number of the element in parentheses. For example, in the above arrays @PDIA is 3, @PA is 2, and @QA is 3.
2.1.3 Initialization of Variables and Arrays

Variables need not be declared prior to using them (unlike C, C++, Fortran etc.). Variables can be created and stored using: 1. Assignment

vr a expression re a piB@PFQAP myrryI a I P Q Y R S T myrryI@QDPA a I (expanding an existing array) Note If a particular subscript in not in range of an array, MATLAB automatically increases the dimensions of the array to t the new element. vr a first X in X lst (default in is 1) myrryP a IXS creates a row vector myrryQ a IXSXPT Y PSXSXSH Note: Number of entries in each row must be equal. olI a IXQ9 olP a TXEIXR9 myrryR a olI olP myrryR a myrryR olP olI nme a 9wike9 9 9 9mith9 mgi@ AD zeros@ AD ones@ AD eye@ A

2. Shortcut Expressions

3. Combining arrays

4. Built-in Functions

from 1 through n2 . The integers are ordered in such a way that all the row sums and all the column sums are equal to the same number. zeros: The zeros function generates a matrix containing all zeros. ones: The ones function generates a matrix containing all ones. eye: The eye function generates an identity matrix.

magic: The mgi@nA function generates an n n matrix constructed from the integers

10

Summary of symbols related to array operations Character Description

X a @ A D Y '

Used in short-cut expressions Assignment operator Subscripts of arrays Brackets; forms arrays Separates array elements Semicolon; suppresses echo of input, ends row in array Single quote; matrix transpose, creates string

2.1.4

Multi-dimensional Arrays

Three dimensional arrays can be visualized as cuboids and can be addressed using 3 subscripts. For example

rryQd@XDXDIA a I P Q Y R S T rryQd@XDXDPA a U V W Y IH II IP is a 2 3 2 array.

rryRd@PDPDPDPA a I is a 2 2 2 2 array.

However higher dimension arrays are harder to visualize and should be thought of in terms of subscripts. For example

11

2.1.5

Subarrays

It is possible to select and use subsets of MATLAB arrays as though they were separate arrays. To select a portion of an array, just include a list of all the elements to be selected in the parentheses after the array name. For example,

rrI a IFI EPFP QFQ ERFR SFS; rrI@QA QFQ rrI@I RA IFI ERFR rrI@IXPXSA IFI QFQ SFS
For a two-dimensional array, a colon can be used in a subscript to select all of the values of that subscript. For example,

rrP a I P QY EP EQ ERY Q R S; rrP@IDXA I P Q rrP@XDIXPXQA I QY EP ERY Q S

The end function end function returns the highest value taken on by that subscript, For example

rrP@PXendDXA

EP EQ ERY Q R S

Subarrays can also be used to change the values of that portion of the main array. For example,

Assigning using subarrays

rrP@XDIXPXQA a III PPP Y QQQ RRR Y SSS TTT rrP@XDIXPXQA a IH

Elements of an array can be deleted by assigning them to the empty array .

Empty array [ ] ; Deleting elements of an array

rrQ a mgi@UA rrQ@I QDXA a

12

2.2 Matrices Operations vs. Arrays Operations

2.2.1 Matrix operations

MATLAB has all the operators of conventional matrix algebra already built in.

same size by adding or subtracting the corresponding elements of the matrices.

Addition and Subtraction of Matrices is carried out on two or more matrices of the Transpose of a Matrix The transpose of a matrix is a new matrix in which the rows of

the original matrix are the columns of the new matrix. If a matrix contains a complex value then we can have both the complex conjugate transpose (trnspose and 9) and complex nonconjugate transpose (trnspose and F9).

a dot@DA The dot product is the scalar computed from two vectors of the same size.
n

Dot Product MATLAB command:

c=
i=1

ai b i

The matrix multiplication is dened by

Matrix Multiplication MATLAB command:

n

a B

c=ab

cij =
k=1

aik bkj

For example if matrices and of dimensions m n and n p respectively are such that number of columns of are equal to number of rows in (in this case: n) then the resulting matrix will have dimensions m p according the above formula. to 2). P is equivalent to B. Similarly, R is equivalent to BBB. To raise a matrix to a power, the matrix must be a square matrix.

Matrix Powers The command for the power of a matrix is P (where, power is equal

Matrix Inverse MATLAB Command: a inv@A By denition, if is an inverse of a square matrix , then B or B are both equal to an identity matrix with only the diagonal elements being 1 and other elements being 0. Determinants MATLAB Command: Solving system of equations
det@A

The solution of a system of equations A x = b is given by x = A1 b. The direct way of calculating this solution using x a inv@eAB is expensive. Alternatively, MATLAB can solve this system using Gaussian elimination which is implemented as the backslash \ . MATLAB Command: x a e \ .

13

2.2.2

Array operations

a I P Q Y R S T a R S T Y I P Q FB denotes element-by-element multiplication of and . A normal matrix multiplication between the above matrices is not dened.

Sometimes we have to perform arithmetic operations between the elements of two arrays of the same size in an element-by-element manner. These operators are denoted by preceding the normal arithmetic operators by a dot . such as (FCD FED FBD FGD F) . For example if and are matrices of same size:

Note + & .+

- & .-

operations produce the exact same result.

Summary of Array and Matrix operators Character Description C or E Array and Matrix addition or subtraction of arrays
FB FG F\ F B G \

Element-by-element multiplication of arrays Element-by-element right division : a/b = @iDjAG@iDjA Element-by-element left division : a\b = @iDjAG@iDjA Element-by-element exponentiation Matrix multiplication Matrix right divide : G a B@A1 Matrix left divide (equation solve) : \ a @A1 B Matrix exponentiation

Precedence (higher to lower):

1. Parentheses @ A 2. transpose F9, power F, complex conjugate transpose 9, matrix power 3. unary operator: Unary plus +, unary minus -, logical negation 4. multiplication FB, right division FG, left division F\, matrix multiplication B, matrix right division G, matrix left division \ 5. addition C, subtraction E 6. colon operator X For the operators with the same precedence, the executions proceed from left to right. MATLAB Programming Basic Programming Components Operators

Refer to MATLAB help for complete precedence rules

14

2.3 Input and Output (I/O) of Data

2.3.1 Input the data from keyboard

We can ask the user to provide input data using the input@A command.

vr a input@9inter the vlue to e storedX

9A

This allows the user to enter any valid MATLAB expression, that evaluates to a numeric or character value.

stringvr a input@9inter the string to e storedX 9D9s9A

When used with the option 9s9, anything that the user enters is stored as character data.

2.3.2

Output of Data to the Screen

The formt statement

In MATLAB the decimal fractions are printed using a default format (short format) that shows 4 decimal decimal digits (eben though MATLAB internally stores double precision variables with 14-15 digits of accuracy). If we want values to be displayed in a decimal format with 14 decimal digits, we use the command formt long. The format can be returned to a decimal format with 4 decimal digits using the command formt short. formt short e command will print the values in scientic notation with 5 signicant digits and formt long e prints the same but with 15 signicant digits. formt C command is used to print the sign only. When a matrix is printed with the formt C command, the only character printed is plus and minus signs. If a value is positive a plus sign will be printed; if a value is zero, a space will be printed; if a value is negative, a minus sign will be printed.

Display formats Command Description

formt formt formt formt formt formt formt formt formt formt short short e short g long long e long g nk ompt loose C

(Default) Fixed-point format with 4 decimal digits Scientic notation with 4 decimal digits Best of 5-digit xed or oating point Fixed-point format with 14 decimal digits Scientic notation with 15 decimal digits Best of 15-digit xed or oating point Two decimal digits Eliminates empty lines Adds empty lines Only signs are printed

15

The disp and numPstr functions

The disp function accepts one array argument whether numeric or string, and displays the value of the array in the Command Window. If the array is of type hr, then the character string contained in the array is printed out. This is useful for displaying the nal result of a program. numPstr function can be used to convert numeric values to character strings and then use disp@A for displaying them. For example,

n a PHY disp@9otl numer of students in the lss a9 numPstr@nAA disp@nA

The fprintf Function

The general form of the fprintf function is:

fprintf @formtD dtA

where formt is a string that controls the way the data is to be printed, and dt is one or more scalars or arrays to be printed. The formt is a character string containing text to be printed plus special characters describing the format of the data.

Common Special Characters in fprintf Format Strings Format String Results

7d 7e 7f 7g

7 7s \n

Display value as an integer Display value in exponential format Display value in oating point format Display value in either oating point or exponential format, whichever is shorter Display a single character Display a string of characters Skip to a new line

temp a UVFPQRSTUWVWY fprintf@he temperture is 7f degreesF

Example

\n9DtempA

will print: he temperture is UVFPQRT degreesF It is also possible to specify the width of the eld in which a number will be displayed and the number of decimal places to display. This is done by specifying the width and precision after the 7 sign and before the f. For example,

fprintf@he temperture is 7RFIf degreesF

16

\n9DtempA

will print: he temperture is UVFP degreesF The output contains the value of temp printed with 4 positions, one of which will be a decimal position as shown above.

2.3.3

I/O through Data Files

Matrices can be dened from information that has been stored in a data le. MATLAB can interface to two dierent types of data les.

MAT les: Data stored in a memory-ecient binary format. They are preferable for data that are going to be generated and used by MATLAB programs only. ASCII les: Data stored in ASCII characters. They are necessary if the data are to be shared (imported or exported) with programs other than MATLAB.

MAT Files

sve filenme vrI vrP vrQ The sve command saves the values of variables vrID vrPD etc. in a le named filenme. By default, the le name will be give the extent mt, and such data les are called MAT-les. sve filenme x y z Eppend adds the variables xD yD z to an existing MAT le filenmeFmt.
The load command is the opposite of the save command. It loads data from a disk le into the current MATLAB workspace. For example,

lod filenme

ASCII Files
The ASCII les must contain only numeric information. We can use % in the ASCII le for comment lines. Each row of the ASCII le must contain the same number of data values to be read by another program in MATLAB.

sve tempFdt d Esii

By loading the ascii le, data value will be automatically stored in a matrix temp (with the same name as the data le) which will have the same size as the data. Though values of the variables c and d were stored in the temp.dat le, they can be read as a variable matrix (temp) for our case.

lod tempFdt
17

2.4 Introduction to Plotting

Plotting is useful when we have to display the output / results of our program in a graphical format.
2.4.1 The

plot

command

The plot@A command generates an x-y plot using 2 arrays. For example to plot the function y = x2 10x + 15 for values of x between 0 and 8.

x a HXIXVY y a xFP E VFBx C ISY plot@xDyAY

Figure 2.1: Plot of y = x2 8x + 15 from 0 to 8. The rst statement creates a vector of x values between 0 and 10 using the colon operator. The second statement calculates the y values from the equation. Finally, the third statement creates the plot using plot function. When the plot function is executed, Matlab opens a Figure Window and displays the plot in the window, see Figure 2.1. Note: Both vectors of x and y must have the same length.
2.4.2 Title, Label, Grid and Text

Titles and axis labels can be added to a plot with the title, xlel, and ylel functions. Each function is called with a string containing the title or label to be applied to 18

the plot. Grid lines can be added or removed from the plot with the grid command: grid on turns on the grid lines, and grid off turns o grid lines. For example, titles, labels and grid lines are applied to the previous gure, as shown in Figure 2.2.

x a HXIXVY y a xFP E VFBx C ISY plot@xDyAY title@9lot of y a xP E VBx C IS9AY xlel@9x9AY ylel@9y9AY grid onY

Figure 2.2: Plot of y = x2 8x + 15 from 0 to 8 with a title, axis labels, and grid lines. The function, text@xDyD9string9A writes the string on the graphics screen at the points specied by the coordinates (xDy) using the axes from the current plot. If x and y are vectors then the text is written in each point.

text{PFSD QD 9y@xA a xP E Vx C IS9} text{PFID QD 9leftrrow9}

19

2.4.3

Multiple curves on one plot

It is possible to plot multiple curves on the same graph. We can plot multiple curves on the same graph by using multiple arguments in the plot command, for example, plot@xIDyIDxPDyPA. Here, xI, yI, xP and yP are vectors. When the command is executed, the curve corresponding to xI and yI will be plotted, and then the curve corresponding to xP and yP will be plotted on the same graph. Another way to plot multiple curves on the same graph is with the hold command. After a hold on command is issued, all additional plots will be laid on top of the previously existing plots. A hold off command switches plotting behavior back to the default situation, in which a new plot replaces the previous one. For example, suppose that we want to plot the function f (x) = sin2x and its derivative, 2cos2x, on the same plot. We can use either of the following two ways (the result is shown in Figure 2.3):

x a HXpiGIHHXPBpiY yI a sin@PBxAY yP a PBos@PBxAY plot@xDyIDxDyPAY

or

x a HXpiGIHHXPBpiY yI a sin@PBxAY yP a PBos@PBxAY plot@xDyIAY hold onY plot@xDyPAY

Figure 2.3: Plot of f (x) = sin2x and f (x) = 2cos2x on the same graph.

20

2.4.4

Line Color, Line Style, Marker Style, and Legends

Matlab allows to select the color of a line to be plotted, the style of the line to be plotted, and the type of marker to be used for data points on the line. These traits may be selected using an attribute character string after the x and y vectors in the plot function. The attribute character string can have up to three characters, with the rst character specifying the color of the line, the second character specifying the style of the marker, and the last character specifying the style of the line. The characters for various colors, markers, and line styles are shown in the following table:

Table of Plot Colors, Marker Styles, and Line Styles Color Marker Style Line Style
y m r g w k
yellow magenta cyan red green blue white black

F o x C B s d v ` b p h `noneb

point circle x-mark plus star square diamond triangle (down) triangle (up) triangle (left) triangle (right) pentagram hexagram no marker

E X EF EE `noneb

solid dotted dash-dot dashed no line

The attribute characters may be mixed in any combination, and more than one attribute string may be specied if more than one pair of (x, y ) vectors are included in a single plot function.

Enhanced Control Plotted Lines

It is also possible to set additional properties associated with lines and markers in the gure.

vineidth species the width of each line in points. wrkeridgegolor species the color of the marker or the edge color for lled markers. wrkerpegolor species the color of the face of lled markers. wrkerize species the size of the marker in points.
These properties are specied in the plot command after the data to be plotted in the following fashion:

plot@xD yD 9ropertyxme9D vlueD FFFA

21

Adding Legends

Legends may be created with the legend function. The basic form is

legend@9stringI9D 9stringP9D FFFD 9votion9D posA or legend@9stringI9D 9stringP9D FFFA where stringI, stringP, etc. are the labels associated with the lines plotted, and pos may be a string specifying where to place the legend. Command legend off will remove an existing legend. The possible values of position are given in the following table.

Values of position in the legend Command Value Legend Location

9xorthest9 9xorth9 9xorthist9 9est9 9ist9 9outhest9 9outh9 9outhist9
Top left corner Top center Top right corner (default) Middle left edge Middle right edge Bottom left corner Bottom center Bottom right corner

2.4.5

Controlling x- and y-axis Plotting Limits

By default, a plot is displayed with x- and y -axis ranges wide enough to show every point in an input data set. However, we can use the xis command/function to control axis scaling and appearance. Some of the forms of the xis command/function are: 1. v a xisY returns a 4-element row vector containing the current limits, xminD xmxD ymin, and ymx, of the plot. 2. xis@xmin xmx ymin ymxA sets the x and y limits of the plot to the specied values. 3. xis equl sets the axis increments to be equal on both axes. 4. xis squre makes the current axis box square. 5. xis norml cancels the eect of axis equal and axis square. 6. xis off turns o all axis labelling, tick marks, and background. 7. xis on turns on all axis labelling, tick marks, and background (default case). An example of a complete plot is shown in Figure 2.4, and the statements to produce that plot are shown below.

22

x a HXpiGPSXPBpiY yI a sin@PBxAY yP a PBos@PBxAY plot@xD yID 9goE9D 9wrkerize9D TFHD 9wrkeridgegolor9D 99D 9wrkerpegolor9D 9g9AY hold onY plot@xD yPD 9rdE9D 9wrkerize9D TFHD 9wrkeridgegolor9D 9r9D 9wrkerpegolor9D 9g9AY title@9lot of f@xA a sin@PxA nd its derivtive9AY xlel@9x9AY ylel@9y9AY legend@9f@xA9D 9dGdx f@xA9D 9votion9D 9xorthest9AY

Figure 2.4: A complete plot with title, axis labels, legend, grid, and multiple line styles.

2.4.6

Controlling Plot features using the GUI

You can perform similar operations to control the settings of a plot using GUI Plotting Tools. However, this process has to be done manually and is not recommended when dealing with a large of plots. Refer to MATLAB help for details on plotting tools.

23

Chapter 3 Branching Statements

All of the MATLAB programs developed previously consist of a series of MATLAB statements that are executed one after another in a xed order. Such programs are called sequential programs. There is no way to repeat sections of the program more than once, and there is no way to selectively execute only certain portions of the program. We will introduce two broad categories of control statements: branches, which select specic sections of the code to execute, and loops, which cause specic sections of the code to be repeated. By using these control statements, we can control the order in which statements are executed in a program.

3.1 Branching
Branches are used to make decisions in your program and to run dierent pieces of code depending upon some logical condition that can be either true or false . For example: . . .

if

else statement group B end . . .

<logical condition > statement group A

If the logical condition is true, then the program executes the statements A, otherwise, the program executes the statements B. After the if-construct the program goes to the line immediately following end. The value of these logical conditions is stored in a dierent MATLAB datatype called the logical datatype.

24

3.1.1

The Logical Data Type

The logil data type can have one of only two possible values: true or flse. These values are produced by the two special functions true and flse. They are also produced by two types of MATLAB operators: relational operators and logical operators. To create a logil variable, just assign a logical value to it in an assignment statement. For example, the following statement creates a logical variable lI containing the logical value true.

lI a trueY

Automatic conversion between numeric and logical datatypes:

In MATLAB, numerical and logical data can be mixed in expressions. If a logical value is used in a place where a numerical value is expected, true values are converted to 1 and flse values are converted to 0, and then used as numbers. If a numerical value is used in a place where a logical value is expected, non-zero values are converted to true and zero values are converted to flse, and then used as logical values. The inbuilt function logil@A does this conversion explicitly. For example logil@HA is flse and logil@xA, where x is some non-zero number, gives true.
3.1.2 Relational Operators

logil result, depending on the relationship between the two operands. The general form of a relational operator is I `opb P

Relational Operators are operators with two numerical or string operands that yield a

where I and P are arithmetic expressions, variables, or strings, and `opb is one of the following relational operators:

Summary of Relational Operators `opb Operator Interpretation

aa a b ba ` `a
Equal to Not equal to Greater than Greater than or equal to Less than Less than or equal to

Note:
Relational operators may be used to compare a scalar value with an array. logrrI a rryI ` Q Relational operators may be used to compare two arrays if they have the same size. logrrP a rryI b rryP
25

 Relational operators may be used to compare two strings if they are of equal lengths. logrrQ a 9his is string I9 aa 9his is sxq P9 The aa symbol is a comparison operation that returns a logical result, while the a symbol is an assignment operation that assigns the value of the expression on the right of the equal sign to the variable on the left of the equal sign. Due to roundo errors during computer calculation, two theoretically equal numbers can dier slightly. For example,

a HY a sin@piAY a @ aa AY
The logical variable should have the value of true, while the result of MATLAB is flse. Because the result of sin@piA MATLAB calculated is 1.2246 1016 instead of exactly zero. So, instead of using exact `equal to' operation, we use the following statement

a @s@ E A ` IFHiEIRA
and if has the value of true, we consider and have the same values.

All the relational operators have the same precedence.

3.1.3 Logical Array Masking

An array of logical values can be used to conduct operations on another numeric array of the same size. This is called masking. A mask is a logical array that selects only those elements of a main array that correspond to a true value in the mask array.

For example:
bb a I P Q Y R S T Y U V W bb a bS bb @A a sqrt@@AA
The array above is the masking array (it is the same size as ) and contains only true or flse value. Thus @A is the subarray of containing only those elements that have a true value in the corresponding element of . Masking is helpful when one is trying to do some operations on a part of a bigger array based on some logical condition.

26

3.1.4

Logical Operators

On several occasions in branching, one needs to combine multiple logical conditions into a single condition. These is done by logical operators. Logical operators combine or operate on logical variables and yield a logical true or flse result. There are ve logical operators that operate on two logical variables: AND (8 and 88), OR (| and ||), and Exclusive-OR (xor), The general form of these operations is

lI `opb lP
where lI and lP are expressions or variables. In addition there is a unary operator: NOT () with the general form

lI
The NOT operator takes the value of lI and simply returns the opposite value. Thus it changes true to flse and flse to true.

Summary of Logical Operators `opb Operator Operation

8 88 | || xor
Logical Logical Logical Logical Logical Logical

AND AND with shortcut evaluation Inclusive OR Inclusive OR with shortcut evaluation Exclusive OR NOT

Results of Logical Operators

lI lP

lI 8 lP lI | lP xor@lIDlPA lI 88 lP lI || lP flse flse flse flse flse flse true flse true true true flse flse true true true true true true flse

With `shortcut evaluation' MATLAB evaluates the rst expression lI and then decides if evaluating the second expression lP is going to change the result of the operation. For example, if we consider the following expression.

Note:

lI a @R`SA || @ dot@IXIHHHDIXIHHHA b H A
Once MATLAB knows the value of the expression on the left of || is true, it does not matter what the result on the right is because the nal result is still going to be true. MAT-

27

LAB simply assigns the logical constant lI a value of true and moves to the next statement. Thus using shortcut evaluation can speed up your MATLAB program. However, shortcut evaluation cannot be used to combine two logical arrays. Then one must use the single | or 8.

Hierarchy of Operations
Logic operators are evaluated after all arithmetic operations and all relational operators have been evaluated. 1. All arithmetic operators are evaluated rst in the order previously described. 2. All relational operators are evaluated, working from left to right. 3. All operators are evaluated. 4. All 8 and 88 operators are evaluated, working from left to right. 5. All |, ||, and xor operators are evaluated, working from left to right.

28

3.2 The if branch

The simplest form of an ifEend statement allows one to decide whether or not to execute certain block of code. The syntax of an ifEend is:

if logical expression statements A end

If logical expression is true, the program executes the statements A. Otherwise, the program skips the statements in between the ifEend construct and goes to the line immediately following end. A short form of this construct can also be used:

if

logical expression ,

statement A ,

end

One may choose to execute a dierent section of code if the logical expression is not true. This can be done with the ifEelseEend construct, which has the syntax:

if logical expression statement group A else statement group B end

If the logical expression is true, then statement group A is executed. If the logical expression is flse, then statement group B is executed. A short form of this construct can also be used:

if

logical expression ,

statement A ,

else

statement B ,

end

Sometimes, one may have to check multiple conditions one after the other before deciding what section of code to execute. This can be done by adding the elseif clause to the above ifEend syntax. The general form of the ifEelseifEend construct has the following syntax:

if logical expression 1 statement group A elseif logical expression 2 statement group B elseif logical expression 3 statement group C . . . end
Here, if logical expression 1 is true, then only statement group A is executed. otherwise if logical expression 2 is true, then only statement group B is executed.

29

otherwise if

logical expression 3 is true, then only statement group C is executed.

If both the logical expressions 1 and 2 are true, then only statement group A is executed. If none of the logical expressions is true, then none of the statements within the ifEelseifEend structure is executed. Another variation of this construct includes an else clause towards the end resulting in an ifEelseifEelseEend construct:

Note:

if logical expression 1 statement group A elseif logical expression 2 statement group B elseif logical expression 3 statement group C . . . else statement group D end
In this case, if none of the logical expressions 1, 2 and 3 is true, then the statement group D is executed.

Note:

3.2.1

The Nested

if

Statement

When one if branch is completely contained within the statement group of another outer if branch, the process is called nesting. In general it takes the form:

if logical expression 1 statement group A if logical expression 2 statement group B end statement group C end
In this case, statement group B is executed only if rst, the and then if logical expression 2 is true.

logical expression 1 is true

30

3.3 The switch statement

The swith construct is used when you want to match the value of a variable or expression with a set of dierent values and then choose to execute a particular block of code. The variable or expression to be matched is called the switch expression and it is matched with dierent case expressions. Both these types of expressions can be integer, character-string, or logical expressions. The general form of a swith construct is:

swith (switch expression ) se case expression 1, statement group A se case expression 2, statement group B ... otherwise, statement group N end
If the value of switch expression is equal to case expression 1, then `statement group A' will be executed, and the program will jump to the rst statement following the end of the swith construct. Similarly, if the value of switch expression is equal to case expression 2, then `statement group B' will be executed, and the program will jump to the rst statement following the end of the swith construct. The same idea applies for any other cases in the construct. The otherwise code block is optional. If it is present, it will be executed whenever the value of switch expression is outside the range of all the case selectors. If it is not present and the value of switch expression is outside the range of all the case selectors, then none of the code blocks will be executed. If many values of the switch expression should cause the same code to execute, all of those values may be included in a single block by enclosing them in curly brackets. In the following example, if the switch expression matches any of the three case expressions in the list, then `statement group a' will be executed.

swith (switch expression ) se {case expression 1, case expression 2, case expression 3 }, statement group A ... otherwise, statement group N end
At most one code block can be executed. After a code block is executed, execution skips to the rst executable statement after the end statement. If the switch expression matches more than one case expression, only the rst one of them will be executed. 31

3.4 MATLAB Debugger

The debugger is a useful tool integrated with the MATLAB editor that helps you nd errors in your code and follow the execution of your code line by line.

Breakpoints: You can stop the execution of your program at a certain point in your code by setting a breakpoint. Once MATLAB reaches that point in the code it `freezes' execution and returns the control to the user. At this point you can check all your variables and run commands as you normally would. To proceed from a breakpoint you may press F10 to follow your code line by line or press F5 to continue execution till the next breakpoint (if any). Conditional Breakpoints: Conditional breakpoints can be set when you wish to stop execution of your code at a point depending upon certain `logical' condition being true. For example, stop only if a variable has a negative value or only for a certain value of the loop variable. Error Breakpoints: Execution stops if any errors or warnings are encountered. Check code with M-lint: M-lint checks for errors, obsolete features, unused variables etc. Proler: Shows the details of the computational time taken to execute a program. Thus one can see which parts of the program take up more time and try to make their that part of the calculation more ecient.

32

Chapter 4 Loops
As your programs start to get more complicated, you will need to follow some Program Design techniques before you can begin writing the code.

4.1 Top-Down Design Techniques

Top-down design is the process of starting with a large task and breaking it down into smaller, more easily understandable pieces (sub-tasks) which perform a portion of the desired task. Each sub-task may in turn be subdivided into smaller sub-tasks if necessary. Once the program is divided into small pieces, each piece can be coded and tested independently. The sub-tasks will be combined into a complete task after each of the sub-tasks has been veried to work properly by itself. The concept of top-town design is the basis of program design process. The details of the process are shown in Figure 3.1. Program design process
1. 2. 3.

Clearly state the problem that you are trying to solve. Dene the inputs required by the program and the outputs to be produced by the program.
An algorithm is a step-by-step procedure for nding the solution to a problem. At this stage, the top-down design techniques are used.

Design the algorithm that you intend to implement in the program.

4. 5.

Turn the algorithm into MATLAB statements. Test the resulting MATLAB program.

Algorithms are usually described using a pseudo-code (derived from pseudo meaning false or stand in, and code for program) that uses constructs such as branching and loops similar to a programming language but does not have a set syntax. We will use a mixture of MATLAB and English as pseudo-code in this class.

33

Figure 4.1: The program design process 34

4.2 Loops
One of the strongest attributes of a computer is its ability to do fast repetitive operations on a set of data. As programmers, we use this feature through loops when we want to repeat certain parts of our program over and over again. For instance, say we want to calculate the sum of all the elements of an array e. The `brute force' way of doing this would be:

bb sume a e@IA C e@PA C e@QA C F F F C e@xA

where x is the total number of elements in the array. Another way to write the same is:

bb bb bb bb

sume sume sume sume

. . .

a a a a

H sume C e@IA sume C e@PA sume C e@QA

. . .

bb sume a sume C e@xA

Clearly, this will be very inecient for large arrays. However, there is a repetitive pattern in the above expressions which can be expressed in the following formula:
N

sume =
i=1

e(i)

Having identied the pattern, the formula can be directly coded up in MATLAB as a for loop as:

sume a H for ii a I X x sume a sume C e@iiA end

The way MATLAB implements this loop is by repeatedly executing the statement between the forEend construct for each value of ii from I to x. In general, as a programmer, if you nd yourself doing repetitive operations then you should immediately try to write down an expression that denes the pattern (as in the above equation) and use a loop to implement it. In MATLAB there are two basic forms of loop constructs: for loops and while loops. The major dierence between these two types of loops is in how the repetition is controlled. The code in a for loop is repeated a specied number of times, and the number of repetitions is known before the loops starts. By contrast, the code in a while loop is repeated an indenite number of times until some user-specied condition is satised.

35

4.3 The for Loop

The general form of a for loop is

for index = expression statement group (body of the loop) end

The expression usually takes the form of a vector in shortcut notation firstXinrXlst. The index variable is assigned values from the loop expression incrementally for every pass of the loop. The number of times the loop will be executed can be computed using the following equation: floor@@lstEfirstAGinrACI. If the value is negative, the loop is not executed.

Example: Calculate the summation of I C P C FFF C IHH.

sum a HY for iiaIXIHH sum a sum C iiY end

Good Programming Practice

1. Indent the bodies of loops. 2. Do not modify the loop index ii within the body of a for loop.
4.3.1 The general form of the

for

Loop

In for loops, the loop expression can take other forms also. In general, the loop expression can be any matrix. Then, MATLAB assigns one column of the matrix to the index variable and executes the statements within the body of the forEend loop, successively for as many columns there are in the matrix. Example:

for ii a S W U ii end

for ii a I P QY R S T ii end

Example:

36

The for loop construct functions as follows: 1. At the beginning of the loop, MATLAB generates the control expression. 2. The rst time through the loop, the program assigns the rst column of the expression to the loop variable index, and the program executes the statements within the body of the loop. 3. After the statements in the body of the loop have been executed, the program assigns the next column of the expression to the loop variable index, and the program executes the statements within the body of the loop again. 4. The above step 3 is repeated over an over as long as there are additional columns in the control expression. Note:

The index of the for loop must be a variable. If the expression is a scalar, then the loop will be executed one time, with the index containing the value of the scalar. If the expression is a row vector, then each time through the loop, the index will contain the next value in the vector. If the expression is a matrix, then each time through the loop, the index will contain the next column in the matrix. Upon completion of the loop, the index contains the last value used.

37

4.4 The while Loop

The general form of a while loop is

while expression statement group end

The controlling expression produces a logical value. If the expression is true, the statement group will be executed, and then control will return to the while statement. If the expression is still true, the statements will be executed again. The process will be repeated until the expression becomes flse. When control returns to the while statement and the expression is flse, the program will execute the rst statement after the end. If the expression is always true (for example, we made an mistake in the expression ), the loop becomes an innite loop and we need to use the Ctrl-C key to abort it.

Example: Calculate the summation of I C P C FFF C IHH.

sum a HY urrent a IY while urrent <a IHH sum a sum C urrentY urrent a urrent C IY end

38

4.5 Simple Applications

1. Finding Statistical Average and Standard Deviation of some numbers:

numers a input@9inter n rry of numersX 9A n a length@numersAY sumnum a HY sqrsum a HY for ii a IXn sumnum a sumnum C numers@iiAY sqrsum a sqrsum C numers@iiAPY end vg a sumnumGnY stddev a sqrt@@nBsqrsumEsumnumPAG@nB@nEIAAAY fprintf@9he men is 7f nd the stndrd devition is 7fF n9D vgD stddevAY
2. Searching for a number in a given array:

numers a input@9inter n rry of numersX 9A srh a input@9inter the numer to serhX 9AY lo a Y for ii a IXlength@numersA if numers@iiA aa srh lo a lo iiY end end if length@loA b H fprintf@9he numer 7d ours t the indiesX 9D srhAY fprintf@9 7d 9D loAY fprintf@9in the rryF n9AY else fprintf@9he numer 7d ws not found in the rryX 9D srhAY end

39

3. Find minimum or maximum number in an array:

numers a input@9inter n rry of numersX 9A minnum a snfY mxnum a HY minlo a Y mxlo a Y for ii a IXlength@numersA if numers@iiA ` minnum minnum a numers@iiAY minlo a iiY elseif numers@iiA aa minnum minlo a minlo iiY end if numers@iiA b mxnum mxnum a numers@iiAY mxlo a iiY elseif numers@iiA aa mxnum mxlo a mxlo iiY end end fprintf@9he minimum numer isX 7d n9 fprintf@9st ws found t the following fprintf@9 7d 9D minloAY fprintf@9n9AY fprintf@9he mximum numer isX 7d n9 fprintf@9st ws found t the following fprintf@9 7d 9D mxloAY fprintf@9n9AY D minnumAY lotions in the rryX9AY D mxnumAY lotions in the rryX9AY

40

4.6 Timing, Preallocation and Vectorization of Loops

The commands ti and to can be used to calculate the the time taken for a certain MATLAB code to execute.

ler llY lY tiY for ii a IXIHHHH sqr@iiA a iiPY end to

7 he size of sqr is qysxq in the loop s ii grows 7 wevef hs to llote 8 dellote memory 7 nd trnsfer dt whih tkes lot of timeF

Preallocation of arrays before the loop begins can speed up the execution.
ler llY lY 7 ith rellotion tiY sqr a zeros@IDIHHHHAY for ii a IXIHHHH sqr@iiA a iiPY end to

7 rere we ievvygei ll the memory needed y sqr

Alternatively, you can also use Vectorization i.e. use the etc.) to speed up the execution even more.

array operators (FB FG F

ler llY lY 7 etoriztion tiY ii a IXIHHHHY sqr a iiFPY to

7 etoriztion s n lterntive to loops

Note that the for loop has been replaced entirely by the array operations. MATLAB internally implements the loop for you when you use array operations. It does so using vectorized memory operations which are faster.

41

4.7 The break and continue Statements

The rek and ontinue statements can be used to control the operation of while loops and for loops. The rek statement terminates the execution of a loop and passes control to the next statement after the end of the loop, while the ontinue statement terminates the current pass through the loop and returns control to the top of the loop.

Examples:
for ii a IXS if ii aa Q rekY end fprintf@ii a 7d \n9D iiAY end disp@ind of loop39AY for ii a IXS if ii aa Q ontinueY end fprintf@ii a 7d \n9D iiAY end disp@ind of loop39AY

42

4.8 Nested Loops

It is also possible to have one loop be completely inside another loop. If one loop is completely inside another one, the two loops are called nested loops. Nested loops are commonly used for doing computations on Matrices.

Example: Ask user for a row vector . Calculate e a 9 B

ler llY a input@9inter row vetorX 9A n a length@AY for ii a IXn fprintf@9ii a 7d n9D iiAY for jj a IXn fprintf@9 jj a 7d n9D jjAY e@iiDjjA a @iiAB@jjAY end end e
Gives the output:

P ii a I jj jj jj ii a P jj jj jj ii a Q jj jj jj e a R IH IR

S a I a P a Q a I a P a Q a I a P a Q IH PS QS

IR QS RW

43

Note:
MATLAB associates each end statement with the latest open for loop. You cannot have an end for the outer loop before the end for the inner loop. When for loops are nested, they should have independent loop index variables. When a rek or ontinue statement appears inside a set of nested loops, then it refers to the latest open loops containing it.

Example:
ler llY a input@9inter row vetorX 9A n a length@AY for ii a IXn fprintf@9ii a 7d n9D iiAY if @ii aa QA fprintf@9yuter voop ii aa Q pss usihF n9AY ontinueY end for jj a IXn fprintf@9 jj a 7d n9D jjAY if @jj aa QA fprintf@9snner voop fyuix t jj aa QF n9AY rekY end e@iiDjjA a @iiAB@jjAY end end e

44

Chapter 5 More Plotting and Graphics

Recall the basic usage of plot@A command. An example of a complete plot is shown in Figure 5.1, and the statements to produce that plot are shown below.

Figure 5.1: A complete plot with title, axis labels, legend, grid, and multiple line styles.

x a HXpiGPSXPBpiY yI a sin@PBxAY yP a PBos@PBxAY plot@xD yID 9goE9D 9wrkerize9D TFHD 9wrkeridgegolor9D 99D 9wrkerpegolor9D 9g9AY hold onY
45

plot@xD yPD 9rdE9D 9wrkerize9D TFHD 9wrkeridgegolor9D 9r9D 9wrkerpegolor9D 9g9AY title@9lot of f@xA a sin@PxA nd its derivtive9AY xlel@9x9AY ylel@9y9AY legend@9f@xA9D 9dGdx f@xA9D 9votion9D 9xorthest9AY

5.1 Additional Types of Two-dimensional Plots

Some of the additional two-dimensional plotting functions are listed here. 1. r@xD yA creates a vertical bar plot, with the values in x used to label each bar and the values in y used to determine the height of the bar. 2. rh@xD yA creates a horizontal bar plot, with the values in x used to label each bar and the values in y used to determine the horizontal of the bar. 3. ompss@xD yA creates a polar plot, with an arrow drawn from the origin to the location of each (xD y) point. 4. pie@xA creates a pie plot. This function determines the percentage of the total pie corresponding to each value of x and plots pie slices of that size. 5. stirs@xD yA creates a stair plot, with each stair step centered on an (xD y) point. 6. stem@xD yA creates a stem plot, with a marker at each (xD y) point and a stem drawn vertically from that point to the x axis. 7. errorr@DDvDA plots versus with error bars. For each point dened by @@iAD@iAA, the vectors v, and represents the distance v@iA below and @iA above the point for the error bar.
5.1.1 Other Useful Plotting Functions

1. ezplot@9funtionnme9Dxmin xmxD figureA 2. fplot@9funtionnme9Dxmin xmxA Example:

ezplot@9sin@xAGx9DERBpi RBpiA

46

5.1.2

Logarithmic Plots

It is possible to plot data on logarithmic scales as well as linear scales. There are four possible combinations of linear and logarithmic scales on the x and y axes, and each combination is produced by a separate function. 1. The plot function plots both x and y data on linear axes. 2. The semilogx function plots x data on logarithmic axes and y data on linear axes. 3. The semilogy function plots x data on linear axes and y data on logarithmic axes. 4. The loglog function plots x and y data on logarithmic axes.
5.1.3 Subplots

subplots. Subplots are created with a suplot command of the form

suplot@mDnDpA

It is possible to place more than one set of axes on a single gure, creating multiple

This command divides the current gure into m n equal-sized regions, arranged in m rows and n columns, and creates a set of axes at position p to receive all current plotting commands. The subplots are numbered from left to right and from top to bottom. For example, the command suplot@PD QD RA would divide the current gure into six regions arranged in two rows and three columns, and create an axis in position 4 (the lower left one) to accept new plot data. Example: Figure 5.2 is produced by the following code:

x a HXpiGQHXPBpiY yI a sin@PBxAY yP a PBos@PBxAY figure@IA suplot@PDPDIA plot@xD yID 9kE9D 9vineidth9D PFHAY xlel@9x9AY ylel@9f@xA9AY suplot@PDPDPA semilogx@xD yID 9E9D 9vineidth9D QFHA xlel@9x9AY ylel@9f@xA9AY suplot@PDPDQA
47

plot@xD yPD 9rEo9D 9vineidth9D PFHAY xlel@9x9AY ylel@9dGdx f@xA9AY xis@H T EQ QAY suplot@PDPDRA plot@xD yPD 9kd9D 9wrkerize9D TFHD 9wrkeridgegolor9D 9r9D 9wrkerpegolor9D 9g9AY xlel@9x9AY ylel@9dGdx f@xA9AY xis@H T EQ QAY

Figure 5.2: A gure containing four subplots.

5.1.4

Creating Multiple Figure Windows

Matlab can create multiple Figure Windows, with dierent data displayed in each window. Window is identied by a gure number, which is a small positive integer. The rst Figure Window is Figure 1, the second is Figure 2, etc. One of the Figure Windows will be the current gure, and all new plotting commands will be displayed in that window. The current gure is selected with the figure function. This function takes the form figure@nA, where n is a gure number. When this command is executed, Figure n becomes the current gure and is used for all plotting commands. The gure is automatically created if it does not already exist. The current gure may also be selected by clicking on it with the mouse.

48

5.1.5

Exporting a Plot as a Graphical Image

The print command can be used to save a plot as a graphical image by specifying appropriate options and a le name.

print Ef`hndleb E`optionsb `filenmeb

where Ef`hndleb species Graphics handle of gure to print, E`optionsb species the format of the output image, and `filenmeb species the name of the output image. For example,

print EfI Edjpeg myplotFjpg

creates a JPEG image of the gure 1 and store it in a le called myplotFjpg. Other options allow image les to be created in other formats. Some of the important image le formats are given in the following table:

print Options to Create Graphics Files

Option Description
dps dps deps deps djpeg dtiff dpng

PostScript for black and white printers PostScript for color printers Encapsulated PostScript Encapsulated Color PostScript JPEG image Compressed TIFF image Portable Network Graphic color image

49

5.2 Three-dimensional Plots

Three-dimensional plots are needed when we want to display plots of functions involving three variables, x, y and z .
5.2.1

plot3

function

This function plots a 3-D curve by connecting (x, y, z ) triplets with line segments (similar to the plot command which connects (x, y ) pairs). General format:

plotQ@xD yD zA}
where x, y, and z are vectors of equal length containing the locations of data points to plot. Figure 5.3 is produced by the following code:

z a HXpiGSHXIHBpiY plotQ@sin@zADos@zADzAY

Figure 5.3: A three-dimensional line plot.

50

5.2.2

The

meshgrid, mesh

and

surf

commands

For plotting a function of two variables z@xDyA we use the mesh and surf commands. Figure 5.4 and Figure 5.5 are produced by the mesh and surf functions in the following code.

xD y a meshgrid@ERXHFPXRAY z a exp@EHFSB@xFP C yFPAAY figure@IA mesh@xD yD zAY figure@PA surf@xD yD zAY

the `grid' of (x, y ) pairs. The function z (x, y ) is then calculated upon this grid of values and plotted using mesh and surf commands.

Note: The meshgrid command is only used to generate the x and y arrays that represent

Figure 5.4: A three-dimensional mesh plot.

Figure 5.5: A three-dimensional surf plot.

51

5.2.3

The Contour functions

Another way to visualize a function of two variables is to use contour plots. MATLAB has can generate contour plots with functions such as: ontour@A, ontourQ@A and ontourf@A. A sample code and its output for the three functions is shown below.

xD yD z a peks@SHAY suplot@PDPDIAY surf@xD yD zAY title@9surf9A suplot@PDPDPAY ontour@xD yD zAY title@9ontour9A suplot@PDPDQAY ontourQ@xD yD zAY title@9ontourQ9A suplot@PDPDRAY ontourf@xD yD zAY title@9ontourf9A

Figure 5.6: MATLAB contour plots.

52

5.2.4

Generating Animations of Plots

Another example of a repetitive process is encountered when one is trying to generate an animation (movie) of plots. Generating a movie can be done rst creating a plot and then over-writing it with another (slightly dierent) plot and doing this repeatedly over and over again.

dely a piGSHY t a HXdelyXPBpiY yI a sin@PBtAY yP a PBos@PBtAY for ii a IXlength@tA hold off plot@t@IXiiAD yI@IXiiAD 9goE9D 9wrkerize9D TFHD 9wrkeridgegolor9D FFF 99D 9wrkerpegolor9D 9g9AY hold onY plot@t@IXiiAD yP@IXiiAD 9rdE9D 9wrkerize9D TFHD 9wrkeridgegolor9D FFF 9r9D 9wrkerpegolor9D 9g9AY xis@H PBpi EP PAY text@SFHDIFVD9t a 9 numPstr@t@iiAAAY puse@delyAY end
The xis command is needed to keep the y-axis constant during the animation. The text command displays the time and the puse command allows the user to see each plot for the specied dely (in seconds) before it is changed. 3-D plots can also be animated. For example:

x y a meshgrid@EpiXpiGIHXpiDEpiXpiGIHXpiAY figure@IAY lfY for t a IXPH 7 voop over time limits for nimtion z a sin@tGPHBPBpiABos@xAFBsin@yAY 7 rete the z vlues surf@xD yD zAY 7 plot the surfe t this instnt xis@Epi pi Epi pi EP PAY 7 set xis limits text@PDEPDPD9ta9 numPstr@tAAY 7 output text puse@HFIAY 7 puse for users to see the imge end

53

Chapter 6 User Dened Functions, Recursion

Why do we need user-dened functions? Well-designed functions enormously reduce the eort required on a large programming project. Their benets include: 1. Independent testing of sub-tasks. 2. Reusable code. 3. Isolation from unintended side eects.

6.1 Introduction to Matlab Functions

The general form of a Matlab function is

funtion [outarg1, outarg2, ...] a fnme@inarg1, inarg2, ... A 7 rI omment line 7 yther omment lines @ixeutle odeA FFF @returnA @endA

54

For example, suppose in a Matlab program, many times you need convert Spherical coordinates, (r, , ), to Cartesian coordinates, (x, y, z ). Your code may look like this:

FFF 7 onverts from @rID thetID phiIA to @xID yID zIA zI a rIBos@phiIAY xI a rIBsin@phiIABos@thetIAY yI a rIBsin@phiIABsin@thetIAY FFF 7 onverts from @rPD thetPD phiPA to @xPD yPD zPA zP a rPBos@phiPAY xP a rPBsin@phiPABos@thetPAY yP a rPBsin@phiPABsin@thetPAY FFF 7 onverts from @rQD thetQD phiQA to @xQD yQD zQA zQ a rQBos@phiQAY xQ a rQBsin@phiQABos@thetQAY yQ a rQBsin@phiQABsin@thetQAY FFF

Alternately if a user dened function, which converts from Spherical coordinates to Cartesian coordinates, had been used, your code may look like this:

FFF 7 onverts from @rID thetID phiIA to @xID yID zIA [xID yID zI] a pherilPgrtesin@rID thetID phiIAY FFF 7 onverts from @rPD thetPD phiPA to @xPD yPD zPA [xPD yPD zP] a pherilPgrtesin@rPD thetPD phiPAY FFF 7 onverts from @rQD thetQD phiQA to @xQD yQD zQA [xQD yQD zQ] a pherilPgrtesin@rQD thetQD phiQAY FFF where function pherilPgrtesin is dened in the Matlab script le pherilPgrtesinFm. funtion [xD yD z] a pherilPgrtesin@rD thetD phiA 7 onverts from @rD thetD phiA to @xD yD zAF z a rBos@phiAY x a rBsin@phiABos@thetAY y a rBsin@phiABsin@thetAY returnY

55

Notes
1. The funtion statement species the name of the function and the input and output argument lists. The input argument list appears in parentheses after the function name, and the output argument list appears in brackets to the left of the equal sign. If there is only one output argument, the brackets can be dropped. 2. Matlab function should be placed in a le with the same name (including capitalization) as the function, and the le extent Fm. For example, if a function is named wyfun, then that function should be placed in a le named wyfunFm. 3. The input argument list is a list of names representing values that will be passed from the caller to the function. These names are called dummy arguments. They are just placeholders for actual values that are passed from the caller when the function is invoked. Similarly, the output argument list contains a list of dummy arguments that are placeholders for the values returned to the caller when a function nishes executing. 4. A function is invoked by naming it in an expression together with a list of actual arguments. A function may be invoked by typing its name directly in the Command Window or by including it in a script le or another function. The name in the calling program must exactly match the function name (including capitalization). When the function is invoked, the value of the rst actual argument is used in place of the rst dummy argument, and so forth for each other actual argument/dummy argument pair. 5. Execution begins at the top of the function and ends when either a return statement, an end statement, or the end of the function is reached. The return statement or the end statement is not actually required in most functions. The only information returned from the function is contained in the output arguments. So, each item in the output argument list must appear on the left side of at least one assignment statement in the function. When the function returns, the values stored in the output argument list are returned to the caller. 6. Matlab function is a special type of M-le that runs in its own independent workspace. For example, the variable dened in the main program, i.e. x, has nothing to do with the variable in a function although they may have the same name x. In other words, the change of the value of x in the function does not change the value of x in the main program or vice versa. 7. The initial comment lines in a function serve a special purpose. The rst comment line after the function statement is called the H1 comment line. It should always contain a one-line summary of the purpose of the function. The special signicance of this line is that it is searched and displayed by the lookfor command. The comment lines from the H1 line until the rst blank line or the rst executable statement are displayed by the help command. They should contain a brief summary of how to use the function.

56

Examples:

Single input and single output function

funtion volume a spherevol@rA 7 his funtion lultes the volume of spheres volume a RGQBpiBrFBrFBrY

Multiple inputs and single output function

funtion volume a ylindervol@rD hA 7 his funtion lultes the volume of ylinders volume a piBrFBrFBhY

Single input and multiple outputs function

funtion [sreD volume] a spheresurvol@rA 7 his funtion lultes the volume nd surfe re of spheres volume a RGQBpiBrFBrFBrY sre a RBpiBrFBrY

Multiple inputs and multiple outputs function

funtion [sreD volume] a ylindersurvol@rD hA 7 his funtion lultes the volume nd surfe re of ylinders sre a PFHBpiBrFB@h C rAY volume a piBrFBrFBhY

Searching
funtion lotions a myserh@rrD numPserhA 7 his funtion serhes for given numer @numPserhA in given rry @rrAF 7 st returns n rry of the indies lotions where the numer is foundF lotions a Y tol a IeEIRY n a length@rrAY for ii a IXn if s@rr@iiAEnumPserhA`tol lotions a lotions iiY end end

57

Sorting
funtion sortedrr a mysort@inprrA 7 his funtion sorts given rry of numers @inprrA using seletion sort n a length@inprrAY for ii a IXnEI 7 inprr 7 puse min a inprr@iiAY lo a iiY for jj a ii C IXn if inprr@jjA`min lo a jjY min a inprr@jjAY end end if lo ~a ii 7 xeed to swp inprr@loA nd inprr@iiA inprr@loA a inprr@iiAY inprr@iiA a minY end end sortedrr a inprrY

58

6.2 Variable Passing in Matlab: The Pass-by-Value Scheme

Matlab programs communicate with their functions using a pass-by-value scheme. When a function call occurs, Matlab makes a copy of the actual arguments and passes them to the function. This copying is highly signicant, because it means that even if the function modies the input arguments, it won't aect the original data in the caller. This feature helps to prevent unintended side eects, in which an error in the function might unintentionally modify variables in the calling program. This behavior is illustrated in the function shown below. This function has two input arguments: and . During its calculation, it modies both input arguments.

funtion out a smple@D A fprintf@sn smpleX a 7fD a 7f 7f \n9D D AY a @IA C PBY a FBY out a C @IAY fprintf@sn smpleX a 7fD a 7f 7f \n9D D AY
A simple test program to call this function is shown below.

a PY a [T R]Y fprintf@fefore smpleX a 7fD a 7f 7f \n9D D AY out a smple@D AY fprintf@efter smpleX a 7fD a 7f 7f \n9D D AY fprintf@efter smpleX out a 7f \n9D outAY
When this program is executed, the results are:

fefore smpleX a PFHHHHHHD a TFHHHHHH RFHHHHHH sn smpleX a PFHHHHHHD a TFHHHHHH RFHHHHHH sn smpleX a IHFHHHHHHD a THFHHHHHH RHFHHHHHH efter smpleX a PFHHHHHHD a TFHHHHHH RFHHHHHH efter smpleX out a UHFHHHHHH
Note that and were both changed inside the function smple, but those changes had eect on the values in the calling program.

no

59

6.3 Optional Arguments

nrgin Returns the actual numbers of input arguments that the function was called with. nrgout Returns the actual numbers of ouput arguments that the function was called with. msg a nrghk@minrgD mxrgD nrgsA 7 nrgs a nrgin y nrgout Generates a standard message if the function was called with incorrect number of input OR output arguments. If there is no error then msga is empty error@msgAD wrning@msgA Displays the error or warning message msg. If msg is empty then no action is taken.

Aside: Global and Persistent variables

Global variables are declared using the keyword glol and can be used within all functions that declare them as global. They are present in a declaring function's workspace but they are not lost when the function exits after completion. Persistent variables are local to some function that declares them with the keyword persistent. The dierence between persistent variables and normal function variables is that when a function is called again and again, persistent variables preserve their value between calls.

6.4 Function of functions

Function functions are useful when you need to specify the name of a MATLAB function or a user dened function as an input argument.

fevl@'functionname', x1, ... xN A Evaluates the specied function functionname with inputs x1, ... evl@'string' A Executes the given 'string' as a MATLAB statement. Examples fzeroD qudD fminnd

xN .

60

6.5 Recursive Functions

An important class of functions are Recursive functions. A function is said to be recursive if it calls itself in its own denition. Recursion is useful for computing the result of a function which can be expressed in terms of an integer @nA number of repetitive operations. In such cases, one can express the function itself as recursive formula. For example, the sum of rst n integers can be written as:

S (n) = 1 + 2 + 3 + ... + (n 1) +n S (n) = S (n 1) +n

The rst equation shows a non -recursive way of calculating the sum of rst @nA integers. This equation can be implemented using the familiar loops. The second equation denes a recursive formula for calculating the sum. The MATLAB code corresponding to the second recursive equation above:

funtion outsum a sumre@nA 7 sumnonre X lultes the sum of the first n integers using reursion if n`I error@9irrorX n must e positive n9AY elseif naaI outsum a IY else outsum a sumre@nEIA C nY end

Important points about Recursion

1. Any recursive function calls itself in its own denition. In the above function sumre@A calls itself with a an input that is one less @nEIA than the given input @nA. Thus, when a user calls this function with (say) sumre@IHA, the function calls itself to nd sumre@WA, which in-turn calls sumre@VA and so on:

sumre@IHA: sumre @WA: . . . sumre @PA: sumre @IA:

outsum outsum . . . outsum outsum

a sumre@WA C IHY a sumre@VA C WY . . . a sumre@IA C PY a IY

In the end, when sumre@PA calls sumre@IA, recursion stops and the value of sumre@IA = 1 is returned. Then sumre@PA adds this value to 2 and returns 3 and so on back upto the level of the original call. 2. Recursion Terminating Condition: Every recursive function must have a terminating condition. In the above example the terminating condition is chosen as naaI. If the terminating condition is missing, then the recursive function would keep calling itself an innite number of times. 61

Examples:
Factorial: n! = 1 2 3 ... (n 1) n

funtion outft a ftre@nA 7 ftreX lultes the ftoril of n integer using reursion if n`I error@9irrorX n must e positive n9AY elseif naaI outft a IY else outft a ftre@nEIA B nY end

Calculation of a Power series: For example, the exponential of a real number x can be obtained using the following innite sum: xn x 2 x3 x4 + + + + + ex = 1 + x + 2! 3! 4! n! A recursive MATLAB user dened function that takes as input the values of x and n and returns two outputs ftn expxn which are n! and ex upto n terms can be implemented as:

funtion ftn expxna reexp@xDnA 7his funtion tkes inputs @xDnA to lulte n3 nd ex with n terms nd outputs if n`H || n ~around@nA error@9irrorX n must e nonEnegtive integer9AY elseif naaH 7terminting ondition ftn a IY expxn a IY else 7reursive ll for @nEIA3 nd ex with nEI terms fnmI expnmI a reexp@xDnEIAY ftn a fnmIBnY expxn a expnmI C xnGftnY 7 get n3 7 get ex with n terms

end

Note that we combined multiple outputs in the above recursive function for better eciency (instead of having dierent recursive calls for summing and factorial).

62

 Fibonacci numbers: 0 1 1 2 3 5 8 13 21 ... F (n) = F (n 1) + F (n 2)

where F (0) = 0 and F (1) = 1.

funtion outfn a fiore@nA 7 fioreX lultes the first n fioni numers if n`I error@9irrorX n must e positive n9AY elseif naaI outfn a HY elseif naaP outfn a H IY else fnmI a fiore@nEIAY outfn a fnmI@nEIA C fnmI@nEPAY outfn a fnmI outfnY end
The following MATLAB code computes and displays the rst n Fibonacci numbers using Loops and Recursion:

ler llY lY n a input@9inter the numer of pioni numers requiredX 9AY fprintf@9pioni numers with voopsX n9AY for ii a IXn if iiaaI fn a HY elseif iiaaP fnmI a HY fn a IY else fnew a fn C fnmIY fnmI a fnY fn a fnewY end fprintf@97d 9D fnAY end fprintf@9n9AY fprintf@9pioni numers with eursionX n9AY fprintf@97d 9Dfiore@nAA fprintf@9n9AY
63

Chapter 7 External File Input/Output

Previously, we have learned some of the Matlab input/output functions, such as loading and saving Matlab data using the lod and sve commands, and writing out formatted data using the fprintf function. In this chapter, we will learn more about Matlab's input/output capabilities.

7.1 The textread() Function

The textred function reads ASCII les that are formatted into columns of data, where each column can be of a dierent type, and stores the contents of each column in a separate output array. The form of the textred function is

D D D FFF

a textred@filenmeD formtD nA

where filenme is the name of the le to open, formt is a string containing a description of the type of data in each column, and n is the number of lines to read. If n is missing, the function reads to the end of the le. The format string contains the same types of format descriptors as function fprintf. Note that the number of output arguments must match the number of columns that you are reading. For example, suppose that le inputFdt contains the following data:

tmes tones yC QFSI PP es lly mith eC QFPV PQ xo

This data could be read into a series of arrays with the following function:

firstDlstDloodDgpDgeDnswer a textred@inputFdt9D7s 7s 7s 7f 7d 7s9A

This function can also skip selected columns by adding an asterisk to the corresponding format descriptor (for example, 7Bs). The following statement reads only the rst name, last name, and gp from the le:

firstDlstDgp a textred@inputFdt9D7s 7s 7Bs 7f 7Bd 7Bs9A

64

Function textred is much more exible than the lod command. The lod command assumes that all of the data in the input le is of a single type  it cannot support dierent types of data in dierent columns. In addition, it stores all of the data into a single array. In contrast, the textred function allows each column to go into a separate variable, which is much more convenient when working with columns of mixed data.

7.2 Introduction to MATLAB File Processing

To use external les within a MATLAB program, we need some way to select the desired le and to read from or write to it. MATLAB has a very exible method to read and write les, whether they are on disk, magnetic tape, or some other device attached to the computer. This mechanism is called the le identier, @fidA. The le identier is a positive integer number that MATLAB assigns to a le when it is opened, and used for all reading, writing, and control operations on that le. File identiers are assigned to disk les or devices using the fopen@A statement, and are detached from them using the flose@A statement. Once an external le is opened using the fopen@A statement and assigned a le id, we can read and write to that le using MATLAB le input and output statements. When we are done with the le, the flose@A statement closes the le and makes the le id invalid. The frewind@A and fseek@A statements may be used to change the current reading or writing position in a le while it is open. Data can be written to and read from les in two possible ways: as formatted text (character) data or as binary data.

Data in formatted les is translated into characters that can be read directly by a user. However, formatted I/O operations are slower and less ecient than binary I/O operations. Binary data consists of the actual bit (0,1) patterns that are used to store the data in computer memory. Reading and writing binary data is very ecient, but a user cannot read the data stored in the le.
We will discuss both types of I/O operations later in this chapter.

7.3 File Opening and Closing

7.3.1 The

fopen

Function

Then fopen function opens a le and returns a le id number that MATLAB has assigned to it. The basic forms of this statement are

fid a fopen@filenmeD permissionA fidD messge a fopen@filenmeD permissionA fidD messge a fopen@filenmeD permissionD formtA
where filenme is a string specifying the name of the le to open, permission is a character string specifying the mode (read, write or append) in which the le is opened, and formt is 65

an optional string specifying the numeric format of the data in the le. The formt option is rarely used. If the open is successful, fid will contain a positive integer after this statement is executed, and messge will be an empty string. If the open fails, fid will contain a EI after this statement is executed, and messge will be a string explaining the error. The possible permission strings are shown in the following table.

fopen@A File Permissions r9 rC9 w9 wC9 9 C9

File Permission Meaning

Open an existing le for reading only (default). Open an existing le for reading and writing. Delete the contents of an existing le (or create a new le) and open it for writing only. Delete the contents of an existing le (or create a new le) and open it for reading and writing. Open an existing le (or create a new le) and open it for writing only, appending to the end of the le. Open an existing le (or create a new le) and open it for reading and writing, appending to the end of the le.

On some platforms such as PCs, it is important to distinguish between text les and binary les. If a le is to be opened in text mode, then a t should be added to the permissions string (for example, rt9 or rtC). If a le is to be opened in binary mode, a may be added to the permissions string (for example, r9), but this is not actually required, since les are opened in binary mode by default. This distinction between text and binary les does not exist on Unix or Linux computers, so the t and is never needed on those systems. Some examples of correct fopen functions are shown below.

Case 1: Opening a Binary File for Input

The function below opens a le named exmpleFdt for binary input only.

fid a fopen@exmpleFdt9D r9A

The permission string is r9, indicating that the le is to be opened for reading only.

Case 2: Opening a File for Text Output

The function below open a le named outdt for text output only.

fid a fopen@outdt9D wt9A or fid a fopen@outdt9D t9A

The wt9 permissions string species that the le is a new text le; if it already exists, the old le will be deleted and a new empty le will be opened for writing. This is the

66

proper form of the fopen function for an output le is we want to replace pre-existing data. The t9 permissions string species that we want to append to an existing text le. If it is already exists, then it will be opened and new data will be appended to the currently existing information. This is the proper form of the fopen function for an output le if we don't want replace pre-existing data.

Case 3: Opening a Binary File for Read/Write Access

The function below opens a le named inexmple for binary input and output.

fid a fopen@inexmple9D rC9A

The function below also opens the le for binary input and output.

fid a fopen@inexmple9D wC9A

The dierence between the rst and the second statements is that the rst statement required the le to exist before it is opened, while the second statement will delete any pre-existing le. Always be careful to specify the proper permissions in fopen statements, depending on whether you are reading from or writing to a le. This practice will help prevent errors such as accidentally overwriting data les that you want to keep. It is also important to check for errors after you attempt to open a le. If the fid is EI, then the le failed to open. You should report this problem to the user, and either select another le or else quit the program.
7.3.2 The

fclose

Function

The flose function closes a le. Its form is

sttus a flose@fidA sttus a flose@ll9A

where fid is a le id and sttus is the result of the operation. If the operation is successful, sttus will be H; if it is unsuccessful, sttus will be EI. The form sttus a flose@ll9A closes all opened les. It returns a status of H if all les close successfully, and EI otherwise.

67

7.4 File Positioning and Status Functions

exist: Check whether an item of a specied kind (variable, inbuilt, le) and specied name exists or not. It returns 2 if a `le' of the given name exists. Its form is:

sttus a exist@name, kind A

Example: stt a exist@9fileioFtxt9D9file9A

feof : Returns logical true(1) or false(0) depending upon whether the end of le for le id fid has been reached or not.

eofstt a feof@fidA
frewind: To reset the le position at the beginning of the le.

frewind@fidA
ftell: Returns the current position within a le using a non-negative integer denoting bytes from the beginning.

position a ftell@fidA
fseek: To relocate the le position by offset number of bytes relative to a position specied by origin which can be @9of9D 9of9D 9eof9A denoting beginning of le, current position in le and end of le.

sttus a fseek@fidD offset, origin A

ferror: Returns the error status number errnum and msg of the most recent le I/O operation associated with the specied le id fid.

msg errnum a ferror@fidA

68

7.5 I/O Functions for Formatted Text Data

7.5.1 The

fprintf

Function

The fprintf function writes formatted data in a user-specied format to a le. Its form is

ount a fprintf@fidD formtD vlID vlPD FFFA fprintf@formtD vlID vlPD FFFA
where fid is the le id of a le to which the data will be written, and formt is the format string controlling the appearance of the data. If fid is missing, the data is written to the standard output device (the Command Window). This is the form of fprintf that we have learned before. The format string species the alignment, signicant digits, eld width, and other aspects of output format. It can contain ordinary alphanumeric characters along with special sequences of characters that specify the exact format in which the output data will be displayed. A single % character always marks the beginning of a format. If an ordinary % is to be printed out, then it must appear in the format string as %%. After the % character, the format can have a ag, a eld-width and precision specier, and a conversion specier. The % character and the conversion specier are always required in any format, while the eld, eld-width, and precision specier are optional. The possible conversion speciers are listed in the following table.

Format Conversion Speciers for fprintf Specier Description

79 7s9 7d9 7u9 7e9 7i9 7f9 7g9 7q9 7o9 7x9 79

Single character String of characters Decimal notation (signed) Decimal notation (unsigned) Exponential notation using a lowercase e Exponential notation using an uppercase i Fixed-point notation The more compact of 7e or 7f Same as 7g, but using an uppercase i Octal notation (unsigned) Hexadecimal notation (using lowercase letters Ef) Hexadecimal notation (using uppercase letters eEp)

If a eld width and precision are specied in a format, then the number before the decimal point is the eld width, which is the number of characters used to display the number. The number after the decimal point is the precision, which is the minimum number of signicant digits to display after the decimal point. Make sure that there is a one-to-one correspondence between the types of the data in an fprintf function and the types of the format conversion speciers in the associated format string; otherwise, your program will produce unexpected results. 69

7.5.2

The

fscanf

Function

The fsnf function reads formatted data in a user-specied format from a le:

rry a fsnf@fidD formtA rryD ount a fsnf@fidD formtD sizeA

where fid is the le id of a le from which the data will be read, formt is the format string controlling how the data is read, and rry is the array that receives the data. The output argument ount returns the number of values read from the le. The optional argument size species the amount of data to be read from the le. There are three versions of this argument:

n  Read exactly n values. After this statement, rry will be a column vector containing n values read from the le. snf  Read until the end of the le. After this statement, rry will be a column vector containing all of the data until the end of the le. mD n  Read exactly m n values, and format the data as an m n array.
The format string species the format of the data to be read. It can contain ordinary characters along with format conversion speciers. The fsnf function compares the data in the le with the format conversion speciers in the format string. As long as the two match, fsnf converts the value and stores it in the output array. The process continues until the end of the le or until the amount of data in size has been read, whichever comes rst. If the data in the le does not match the format conversion speciers, the operation of fsnf stops immediately. The format conversion speciers for fsnf are basically the same as those for fprintf.
7.5.3 The

fgetl

and

fgets

Functions

The fgetl and fgets functions read the next line of input from the le specied by le id fid and return a character array (string), line as output. Their general form is

line a fgetl@fidA line a fgets@fidA

The dierence is that for fgetl@A, the output string line does not contain the end of line character in the end where as for fgets@A, it does. If an end of a le is encountered, the value of line is set to EI.

70

7.6 I/O Functions for Binary Data

7.6.1 The

fwrite

Function

The fwrite function writes binary data in a user-specied format to a le. Its form is

ount a fwrite@fidD rryD preisionA ount a fwrite@fidD rryD preisionD skipA

where fid is the le id of a le opened with the fopen function, rry is the array of values to write out, and ount is the number of values written to the le. Matlab writes out data in olumn order, which means that the entire rst column is written out, followed by the entire second column, etc. For example, if 1 2 rry = 3 4 , 5 6 the data will be written out in the order ID QD SD PD RD T. The optional preision string species the format in which the data will be output. Matlab supports both platform-independent precision strings, which are the same for all computers that Matlab runs on, and platform-dependent precision strings, which vary among dierent types of computers. The possible platform-independent precisions are presented in the following table. All of these precisions work in units of bytes, except for itx9 or uitx9, which work in units of bits.

Selected Matlab Precision Strings Matlab C/Fortran Precision String Equivalent

hr9 shr9 uhr9 intV9 intIT9 intQP9 intTR9 uintV9 uintIT9 uintQP9 uintTR9 flotQP9 flotTR9 itx9 uitx9

hrBI9 signed hr9 unsigned hr9 integerBI9 integerBP9 integerBR9 integerBV9 integerBI9 integerBP9 integerBR9 integerBV9 relBR9 relBV9

8-bit character 8-bit signed character 8-bit unsigned character 8-bit integer 16-bit integer 32-bit integer 64-bit integer 8-bit unsigned integer 16-bit unsigned integer 32-bit unsigned integer 64-bit unsigned integer 32-bit oating point 64-bit oating point N-bit signed integer, 1 <= N <= 64 N-bit unsigned integer, 1 <= N <= 64

Meaning

71

The optional argument skip species the number of bytes to skip in the output le before each write. Note that if preision is a bit format like itx9 or uitx9, skip is specied in bits instead of bytes.
7.6.2 The

fread

Function

The fred function reads binary data in a user-specied format from a le, and returns the data in a (possibly dierent) user-specied format. Its form is

rryD ount a fred@fidD sizeD preisionA rryD ount a fred@fidD sizeD preisionD skipA
where fid is the le id of a le opened with the fopen function, size is the number of values to read, rry is the array to contain the data, and ount is the number of values read from the le. The optional argument size species the amount of data to be read from the le. There are three versions of this argument:

n  Read exactly n values. After this statement, rry will be a column vector containing n values read from the le. snf  Read until the end of the le. After this statement, rry will be a column vector containing all of the data until the end of the le. mD n  Read exactly m n values, and format the data as an m n array.
The preision argument species both the format of the data on the disk and the format of the data array to be returned to the calling program. The general form of the precision string is

diskpreision ab rrypreision9
where diskpreision and rrypreision are both among the precision strings found in the previous table. The rrypreision value can be defaulted. If it is missing, then the data is returned in a doule array. There is also a shortcut form of this expression if the disk precision and the array precision are the same: Bdiskpreision9.

72

Chapter 8 Numerical Methods in MATLAB

8.1 Matrix Algebra
One of the most common linear algebra problems is the solution of a linear set of equations. For example, consider the set of equations 1 2 3 x1 10 4 5 6 x2 = 11 7 8 0 x3 12 or Ax = y The above equations dene the product of the n n matrix A and the n 1 vector x as being equal to the n 1 vector y. The existence of solution to the above equation is a fundamental problem in linear algebra. Moreover, when a solution does exist, there are numerous approaches to nding the solution, such as Gaussian elimination, LU factorization, or direct use of A1 . It is beyond the scope of this text to discuss the many analytical and numerical issues of matrix algebra. We will only demonstrate how Matlab can be used to solve problems like the one above. Ax = y has a unique solution if the rank of A and the rank of the augmented matrix [A y] are both equal to n. Alternatively, one can check the condition number of A. If the condition number is not excessively large, then A has an inverse with good numerical properties, which means we can nd A1 . The rank of a matrix can be calculated by using Matlab function rnk. And the condition number of a matrix can be obtained by using Matlab function ond. For example, rnk@eA, rnke y, and ond@eA give the rank of A, the rank of [A y], and the condition number of A, respectively. There are two ways to nd the solution of Ax = y in Matlab: 1. x a inv@eABy. This method is more straightforward while less favorable. Please note inv@eA computes A1 . 2. x a e\y. The method using matrix left division is preferable. This method utilizes an LU factorization approach, which requires fewer oating-point operations and as a result is signicantly faster. 73

In addition to the solution of linear sets of equations, Matlab oers numerous matrix functions that are useful for solving numerical linear algebra problems. Some of these functions are listed below:

hol@eA  Cholesky factorization. ond@eA  Matrix condition number. det@eA  Determinant. eig@eA  Vector of eigenvalues. D h a eig@eA  Matrix of eigenvectors, and diagonal matrix containing eigenvalues. inv@eA  Matrix inverse. vD a lu@eA  LU decomposition. norm@eD typeA  Matrix and vector norms. rnk@eA  Matrix rank. shur@eA  Schur decomposition. svd@eA  Singular values. D D a svd@eA  Singular value decomposition. tre@eA  Sum of matrix diagonal elements.

8.2 Data Analysis

Matlab provides numerous functions that performs statistical analyses on data sets. Some of these functions are listed below:

mx@xA  If x is a vector then the function returns the largest value in x. If x is a matrix then the function returns a row vector containing the maximum element from each column. D a mx@xA  In this form, if x is a vector, then the function stores the maximum value of x in a scalar , and the index of the maximum value in the scalar . If there are several identical maximum values, the index of the rst one found is returned. If x is a matrix, then the function stores the maximum values of x in a vector and the indices of the maximum values in a vector . mx@xD DIA  The function returns a row vector containing the maximum element from each column.

74

 mx@xD DPA  The function returns a column vector containing the maximum element from each row. Replacing mx with min in the above functions gives the corresponding functions for nding minimums. men@xA  If x is a vector, this function computes the average value of the elements of the vector x. If x is a matrix, this function computes a row vector that contains the average value of each column. men@xDIA  The function returns a row vector containing the average value from each column. men@xDPA  The function returns a column vector containing the average value from each row. medin@xA  If x is a vector, this function computes the median value of the elements of the vector x. If x is a matrix, this function computes a row vector that contains the median value of each column. medin@xDIA  The function returns a row vector containing the median value from each column. medin@xDPA  The function returns a column vector containing the median value from each row. sum@xA  If x is a vector, this function returns the sum of the elements of x. If x is a matrix, this function returns a row vector that contains the sum of each column. sum@xDIA  The function returns a row vector containing the sum of elements from each column. sum@xDPA  The function returns a column vector containing the sum of elements from each row. sort@xA  If x is a vector then this function sorts the values into ascending order. If x is a matrix then this function sorts each column into ascending order. D a sort@xA  The sorted values are returned to the matrix and reordered indices are stored in the matrix . std@xA  If x is a vector, this function computes the standard deviation of the values of x. If x is a matrix, this function computes a row vector containing the standard deviation of each column. vr@xA  If x is a vector, this function computes the variance of the values of x. If x is a matrix, this function computes a row vector containing the variance of each column.

75

8.3 Polynomials
Matlab provides a number of functions for manipulating polynomials. We will introduce a couple of those functions.
8.3.1 Roots

Finding the roots of a polynomial is a problem common to many disciplines. Matlab solves this problem and nds the roots of a polynomial by using function roots. In Matlab a polynomial is represented by a row vector of its coecients in descending order. For example, the polynomial x4 12x3 + 0x2 + 25x + 116 is represented by a vector

p a I EIP H PS IITY
Note that terms with zero coecients must be included. Matlab has no way of knowing which terms are zero unless you specically identify them. Given this form, the roots can be found by

r a roots@pAY
where r is a column vector containing the roots of the polynomial. Matlab adopts the convention that polynomials are row vectors and roots are column vectors.
8.3.2 Curve Fitting

In numerous application areas, one is faced with the task of tting a curve to measured data. Sometimes the chosen curve passes through the data points, but at other times the curve comes close to, but does not necessarily pass through the data points. In the most common situation, the curve is chosen so that the sum of the squared errors at the data points is minimized. This choice results in a least squares curve t. It is straightforward and common to use a polynomial as the basis function for the least squares curve tting. In Matlab, the function polyfit solves the least squares polynomial curve-tting problem. To illustrate the use of this function, let's start with the data

x a H HFI HFP HFQ HFR HFS HFT HFU HFV HFW IFHY y a EHFRS IFWV QFPV TFIT UFHV UFQR UFTT WFST WFRV WFQH IIFPHY
To use polyfit, we must give it the above data and the order or degree of the polynomial we wish to best t the data. If we choose n = 1 as the order, the best straight-line approximation will be found. This is often called linear regression. If we choose n = 2 as the order, a quadratic polynomial will be found. For example, let's choose a quadratic polynomial:

n a PY p a polyfit@xD yD nAY
76

The output of polyfit is a row vector of the polynomial coecients, which are EWFVIRU PHFIQQV EHFHQPU stored in p. This means the solution is y (x) = 9.8147x2 + 20.1338x 0.0327. Figure 8.1, which is created by the following code, shows the original data and the tted curve.

xi a linspe@HD ID IHHAY yi a polyvl@pD xiAY figure@IA plot@xD yD 9Eo9D xiD yiD 9EE9A xlel@9x9AY ylel@9y a f@xA9AY title@9eond yrder gurve pitting9AY
where function polyvl is used to evaluate the polynomial at a set of data points.

Figure 8.1: Original data points and the second order curve tting.

77

8.4 Integration
Matlab provides functions for numerically approximating the integral of a function. These functions are qud, qudl, and dlqud.

qud  Numerically evaluate integral, adaptive Simpson quadrature. qudl  Numerically evaluate integral, adaptive Lobatto quadrature. dlqud  Numerically evaluate double integral. f (x) = ex occurs frequently in physics problems but cannot be integrated analytically. 1 Let's show how to use function qud to compute 0 f (x)dx numerically.
First, we need to dene the function to be integrated. The function must support a vector 2 input argument and return a vector of outputs for a vector inputs. For example, f (x) = ex is dened in myfunFm as
2

funtion y a myfun@xA y a exp@ExFPAY

Then, we may use qud to compute the integration.

z a qud@dmyfunD HD IAY or z a qud@myfun9D HD IAY

where the rst argument is the dened function to be integrated, the second and third arguments are the lower limit and upper limit of integration, respectively.

8.5 Dierential Equations

Matlab has the capability to solve as wide variety of problems involving dierential equations, such as ordinary dierential equations (ODEs), initial value problems (IVPs), boundary value problems (BVPs), and partial dierential equations (PDEs).
8.5.1 IVP Format

Matlab computes the time history of a set of coupled rst-order dierential equations with known initial conditions. These problems are called initial value problems and have the form = f (t, y) y y(t0 ) = y0 which is vector notation for the set of dierential equations

y 1 = f1 (t, y1 , y2 , . . . , yn ) y1 (t0 ) = y10 y 2 = f2 (t, y1 , y2 , . . . , yn ) y2 (t0 ) = y20 . . . . . . y n = fn (t, y1 , y2 , . . . , yn ) yn (t0 ) = yn0

78

where y i = dyi /dt, n is the number of rst-order dierential equations, and yi0 is the initial condition associated with the ith equation. When an initial value problem is not specied as a set of rst-order dierential equations, it must be rewritten as one. For example, consider the classic van der Pol equation

x (1 x2 )x +x=0
where is a parameter greater than zero. If we choose y1 = x and y2 = dx/dt, the van der Pol equation becomes y 1 = y2 2 )y2 y1 y 2 = (1 y1 We will demonstrate Matlab ODE solvers by solving this problem.
8.5.2 ODE Solvers

Matlab oers seven initial value problem solvers. We will talk about two of them.

odeRS  An explicit one-step Runge Kutta medium-order (4th- to 5th-order) solver. Suitable for non-sti problems that require moderate accuracy. This is typically the rst solver to try on a new problem. odeISs  An implicit, multistep numerical dierentiation solver of varying order (1stto 5th-order). Suitable for sti problems that require moderate accuracy. This is typically the solver to try if odeRS fails or is too inecient.
8.5.3 Basic Use

Before a set of dierential equations can be solved, they must be coded in a function M-le. That is, the le must accept a time t and a solution y and return values for the derivatives. For the van del Pol equation, this ODE le can be written as follows.

funtion ydot a vdpol@tD yA 7 his funtion @ydotaf@tDyAA defines the system of yhis you re trying to solveF 7 he input 9y9 n e xxI vetorY hus 9ydot9 must lso e xxI vetor mu a PY 7 some onstnt yIdot a y@PAY yPdot a muB@IEy@IAPABy@PA E y@IAY ydot a yIdotY yPdotY
Note that the input arguments are t and y but that the function does not use t. Note also that the output ydot must be a column vector. Given the above ODE le, this set of ODEs is solved by the following statements.

tspn a H PHY 7 time spn to integrte over yH a PY HY 7 initil onditions @must e olumnA tD y a odeRS@9vdpol9D tspnD yHAY plot@tDy@XDIAD9FE9DtDy@XDPAD9gFE9A legend@9y9D9dyGdt9A
79

where t contains the time points, y@XDIA contains the solution of x@tA, and y@XDPA contains the values of dxGdt. sliding down a rotating support as shown in 8.2.

Example 1: In a dynamic problem, we have the following governing equation for a bowl
2 r 0 r = g sin(0 + 0 t)

where r is the distance of bowl away from the support end. Initial distance L = 2m ; Angular rotating speed of slope 0 = 0.1 rad/s ; Initial angle of the slope 0 = 0.2 rad. g is the acceleration due to gravity. Solve for r(t) and r (t) using Matlab. Verify with the exact solution given by:

r(t) = r (t) =

g cos0 g g sin0 cosh(0 t) sinh(0 t) + 2 sin(0 + 0 t) 2 2 20 20 20 g sin0 g cos0 g L0 sinh(0 t) cosh(0 t) + cos(0 + 0 t) 20 20 20 L

Figure 8.2: Object on rotating slope.

Solution

First we need convert this second order ODE to a set of rst-order dierential equations. Let y1 = r and y2 = dr/dt, we have

y 1 = y2 y 2 = 2 y1 g sin(0 + 0 t)
The code for the derivative function as well as the statement for solving the ODEs are listed below.

80

funtion ydot a owlonslope@tD yA wH a HFIY thetH a HFPY g a WFVY ydot a y@PAY wHPBy@IA E gBsin@thetHCwHBtAY

AND
ler llY lose llY v a PFHY tspn a H IFHY yH a vY HY g a WFVY wH a HFIY thetH a HFPY tD y a odeRS@9owlonslope9D tspnD yHAY ti a linspe@HD IFHD IHHAY ri a @v E gGPGwHPBsin@thetHAABosh@wHBtiA E FFF gGPGwHPBos@thetHABsinh@wHBtiA C gGPGwHPBsin@thetHCwHBtiAY ridot a @vBwH E gGPGwHBsin@thetHAABsinh@wHBtiA E FFF gGPGwHBos@thetHABosh@wHBtiA C gGPGwHBos@thetHCwHBtiAY figure@IA plot@tD y@XDIAD 9FE9D tiD riD 9kE9AY xlel@9t9AY ylel@9distne r9AY legend@9wtl xumeril olution9D9enlytil ixt olution9A figure@PA plot@tD y@XDPAD 9FE9D tiD ridotD 9kE9AY xlel@9t9AY ylel@9owl speed9AY legend@9wtl xumeril olution9D9enlytil ixt olution9A
Figure 8.3 shows the distance r versus time t. In this gure, analytical solution is also given for comparison. Figure 8.4 shows the speed of the bowl r versus time t.

81

Figure 8.3: Distance versus time.

Figure 8.4: Speed versus time.

82

the change in concentration three species in a tank. The reactions A B C occur within the tank. The constants k1 , and k2 describe the reaction rate for A B and B C respectively. The following ODEs are obtained:
dCb dt dCa dt

Example 2: In Chemical Engineering, the following set of dierential equations describes

= k1 Ca = k1 Ca k2 Cb dCc = k2 Cb dt

Where k1 = 1 hr1 and k2 = 2 hr1 and at time t = 0, Ca = 5 mol and Cb = Cc = 0 mol. Solve the system of equations and plot the change in concentration of each species over time. Select a time interval [0 5] for the integration.
Solution:

The derivative function

funtion ydot a onentrtion@tD yA kI a IY kP a PY ydot a EkIBy@IAY kIBy@IAEkPBy@PAY kPBy@PAY

The main program

ler llY lose llY tspn a H SY yH a S H HY tD y a odeRS@9onentrtion9D tspnD yHAY figure@IA plot@tD y@XDIAD 9rE9D tD y@XDPAD 9EE9D tD y@XDQAD 9kB9A xlel@9ime @hrA9AY ylel@9onentrtion of eh speies @molGhrA9AY legend@9g9D9g9D9g9D9votion9D9ist9AY
Figure 8.5 shows the result.

83

Figure 8.5: Concentrations versus time.

8.6 Advanced MATLAB Features

Cell Arrays User dened structures and Classes Visual Programming with GUI Symbolic Math Toolbox Objects and Classes Parallel Programming

84

Chapter 9 Application to Civil Engineering: Structural Dynamics

In this chapter, a structural dynamics problem is presented and solved with Matlab. Suppose the dynamic behavior of a three-oor building, which is shown in Figure 9.1, needs to be determined. The building's displacement is measured at each oor. The columns separating the oors are modelled as springs and all the mass of the oors can be lumped into one value per oor. Additionally, velocity proportional dampers exert a small force on the lumped mass at a oor directly proportional to the velocity of that degree of freedom.

Figure 9.1: Three-oor building modelled as mass spring system.

85

The governing dierential equation along with initial conditions for this system is

(t) + C x (t) + K x(t) = F (t) Mx (0) = v0 x and x(0) = x0

where M is the 3 3 mass matrix, K is the 3 3 stiness matrix, and C is the 3 3 damping matrix. Vectors x and F store the oor displacements and applied loads at each oor, respectively. x0 and v0 give the initial state (displacement and velocity) of the structure at time t = 0. Based on the given information about the mass of each oor and the stiness of each column, The mass matrix, stiness matrix and damping matrix can be constructed as m 0 0 2k k 0 0 ; K = k 2k k ; C = M + K M = 0 m 0 0 m/2 0 k k The above data can be dened in an input le strdyninput.m:

funtion w g u v d uH vH spn delt a strdyninput 7 snitiliztion m a IFHY k a IFHY v a HFSY d a HFSY w a m H HY H m HY H H mGPFHY u a PBk Ek HY Ek PBk EkY H Ek kY lph a HFHY et a HFHY g a lphBw C etBuY 7 snitil onditions uH a HFHSY HFHSY HFHSY vH a HY HY HY 7uH a HY HY HY vH a HFHSY HFHSY HFHSY 7uH a HY HY HY vH a HY HY HY 7 hefine time spn nd time step spn a PSFHY delt a HFIY
The externally applied force can also be dened as a function of time t in strdynforce.m:

funtion fore a strdynfore@tA fore a HY HY HY 7fore a EHFHSBsin@tAY HFIBsin@tAY HFIBos@tAY

Let's consider a free vibration problem, which has F = 0. Initially, a rope pulled the rst oor with a displacement s0 , as shown in Figure 9.2, and suddenly got cut at time t = 0. The (0) = [0 0 0]T and x(0) = [s0 s0 s0 ]T . initial conditions of this problem can be described as x 86

Figure 9.2: Initial conguration of the three-oor building. The following two methods can be used to solve the Structural Dynamics problem:

Method 1: Using ode45()

Convert the given system into standard form:

= f (t, y ) y (t) = f (t, y (t)) y y1 (t) 1 (t) = x (t) and , y1 (t) = x(t) ; y2 (t) = y y2 (t) 1 (t) = y2 (t) y 2 (t) = M 1 [F (t) C y2 (t) K y1 (t)] y
Now we can code this using odeRS@A. Dene the standard form of the equation to be solved: strdynfun.m:

where y (t) =

funtion ydot a strdynfun@tD yA w g u a strdyninputY p a strdynfore@tAY

87

Figure 9.3: Displacements versus time.

n a size@wDIAY yI a y@IXnAY yP a y@nCIXPBnAY yIdot a yPY yPdot a inv@wAB@pEuByIEgByPAY ydot a yIdotY yPdotY
Main program strdyn.m

ler llY lose llY 7 snitiliztion w g u v d uH vH spn a strdyninputY tmtD ymt a odeRS@9strdynfun9D H spnD uHY vHAY plot@tmtDymt@XDIAD9r9D tmtDymt@XDPAD99D tmtDymt@XDQAD9k9AY xlel@9time9AY ylel@9hisplement9AY legend@9tory I9D9tory P9D9tory Q9D9votion9D9outhist9A

88

Method 2: Central dierence method

The central dierence method with a time step t is used for solving the second order dierential governing equation. The calculation steps are listed below:

0. 1. Calculate the vector value of initial acceleration, x 0 = M 1 (F0 C x 0 Kx0 ) x

2. Compute one backward step (to time -1) by using the following equation.

x1

(t)2 0 0 + x = x0 tx 2

3. Compute the eective stiness matrix by using the following equation.

Ke =
4. Loop over time steps i = 1:N

1 1 M+ C 2 (t) 2t

(a) Compute the next state (at ti+1 ) of the system in terms of the two previous states ti and ti1 :
1 xi+1 = Ke Fi K

2 M (t)2

xi

1 1 M C xi1 (t)2 2t

(b) Store the solution and move to the next time step. Main Program strdyncd.m

ler llY lose llY 7 snitiliztion w g u v d uH vH spn delt a strdyninputY 7 hefine the time steps t a EdeltXdeltXspnY nsteps a length@tAY 7 glulte the vetor vlue of initil elertion p a strdynfore@HAY H a inv@wAB@p E gBvH E uBuHAY 7 hefine mtrix storing the solutions 7 he rows of the mtrix re the degrees of freedom
89

7 he olumns of the rry re the stte of the system in time sol a zeros@QD nstepsAY 7 he seond olumn is the position t t a H sol@XDPA a uH@XAY 7 gompute one kwrd step @t a Ehelt tAD nd store x@Ehelt tA in the 7 first olumn of the solution mtrix sol@XDIA a uH@XA E deltBvH@XA C deltPGPFHBH@XAY 7 gompute the effetive stiffness mtrix nd store the inverse ueff a wG@deltPA C gG@PFHBdeltAY ueffinv a inv@ueffAY 7 wrh the eqution forwrdD reting new steps y ssigning the next 7 stte of the system s funtion of the two previous sttes for iiaQXnsteps p a strdynfore@t@iiAAY sol@XDiiA a ueffinvB@p E @uEPFHBwG@deltPAABsol@XDiiEIA FFF E @wG@deltPAEgG@PFHBdeltAABsol@XDiiEPAAY end 7 lot the time history of the displements of eh floor figure@IAY plot@t@PXnstepsAD sol@IDPXnstepsAD 9rE9D 9vineidth9D IFSA hold on plot@t@PXnstepsAD sol@PDPXnstepsAD 9E9D 9vineidth9D IFSA plot@t@PXnstepsAD sol@QDPXnstepsAD 9kE9D 9vineidth9D IFSA xlel@9time9AY ylel@9hisplement9AY legend@9tory I9D9tory P9D9tory Q9D9votion9D9outhist9A puseY 7 enimtion xpI a linspe@HD vD IHAY xpP a linspe@vD PBvD IHAY xpQ a linspe@PBvD QBvD IHAY oefwtI a H H H IY @vAQ @vAP v IY H H I HY QB@vAP PB@vA I HY oefwtP a vQ vP v IY @PBvAQ @PBvAP PBv IY QBvP PBv I HY FFF QB@PBvAP PB@PBvA I HY oefwtQ a @PBvAQ @PBvAP @PBvA IY @QBvAQ @QBvAP @QBvA IY FFF QB@PBvAP PB@PBvA I HY QB@QBvAP PB@QBvA I HY invoefwtI a inv@oefwtIAY invoefwtP a inv@oefwtPAY
90

invoefwtQ a inv@oefwtQAY figure@PAY for iiaPXnsteps dI a sol@IDiiAY dP a sol@PDiiAY dQ a sol@QDiiAY oefI a invoefwtIBHY dIY HY HY oefP a invoefwtPBdIY dPY HY HY oefQ a invoefwtQBdPY dQY HY HY ypI a polyvl@oefID xpIAY ypP a polyvl@oefPD xpPAY ypQ a polyvl@oefQD xpQAY plot@ypID xpID 9E9D 9vineidth9D PAY hold onY plot@ypPD xpPD 9E9D 9vineidth9D PAY plot@ypQD xpQD 9E9D 9vineidth9D PAY plot@ypICdD xpID 9E9D 9vineidth9D PAY plot@ypPCdD xpPD 9E9D 9vineidth9D PAY plot@ypQCdD xpQD 9E9D 9vineidth9D PAY plot@dI dCdID vD vD 9rE9D 9vineidth9D RAY plot@dP dCdPD PBvD PBvD 9rE9D 9vineidth9D RAY plot@dQ dCdQD QBvD QBvD 9rE9D 9vineidth9D RAY xis@EHFPS HFUS HFH IFUSAY text@HFTDIFTD9t a 9 numPstr@t@iiAAAY hold offY puse@HFHIAY

end

7 gompre with wevef odeRS solutionX tmtD ymt a odeRS@9strdynfun9D H spnD uHY vHAY figure@IA hold onY puseY plot@tmtDymt@XDIAD9r9D tmtDymt@XDPAD99D tmtDymt@XDQAD9k9AY

91