Kursus Matlab di Jakarta Selatan KURSUS MATLAB ONLINE Skripsi, Tesis, DISERTASI 081219449060: KURSUS MATLAB ONLINE Skripsi, Tesis, DISERTASI 081219449060 - MULTIPLY AN ARBITARY NUMBER OF ARRAYS AND function tempPlot(temps)

% Author: Spike

% Date: 22/3/1999

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

function tempPlot(temps)

%%%%%%%%%%%%%%%%%%%%

% Generate the hours

%%%%%%%%%%%%%%%%%%%%

hours=1:size(temps,2);

%%%%%%%%%%%%%%%%%%%%%%%%

% Plot & label the graph

%%%%%%%%%%%%%%%%%%%%%%%%

plot(hours,temps);

xlabel('Hour of the Day');

ylabel('Temperature (Celcius)');

title('Daily Temperature Plot');

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

% Highlight the value at mid-day

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

if hours(end)>=12

text(12,temps(12),'*');

end;

24.22 Review

• Motivation for function use

• calling functions

• design principles for functions

• function declaration syntax

• comments & functions

• top-down & bottom-up views of functions

• black-box paradigm

• parameters

• returned values

25 FURTHER MATLAB FUNCTIONS

• Functions are a vital part of all procedural languages

- issues of re-use, cohesion, & coupling

• Much of Matlab’s power focuses on the built-in functions

• Matlab allows users to write and employ their own functions

- encourages modularity & re-use

• Lecture is the 2nd in a series of two concerning functions in Matlab. Topics covered include:

- formal vs. actual parameters & outputs

- workspaces & scope of variables

- stack & run-time support for function calls

- functions with variable numbers of inputs & outputs

- efficiency & encapsulation issues

References: For Engineers (Ch. 1, 5)

Student (Ch. 12, 24)

Mastering (Ch.14, Appendix A)

25.1 Formal vs. Actual Parameters & Outputs

• In order to minimise coupling a function should not know unnecessary details about the caller

- no knowledge of larger task of the caller

- no knowledge of the variables of the caller

• However, communication of values is till necessary

- inputs (arguments) for the function to work upon

- outputs (returned values) for the function to communicate results

• The apparent conflict of goals can be resolved by allowing the function to employ its own name for arguments & returned value independent of whatever they are named by the caller:

- names employed by the function are known as formal parameters & outputs

- names employed (values supplied) by caller are known as actual parameters & outputs

- must be a 1-1 mapping between them

25.2 Parameter Association

• A 1-1 mapping between parameters (& outputs) of the caller (actual) and function (formal) is needed

• Mapping (association) is based simply on order:

- first actual parameters is mapped to 1st formal parameter, 2nd actual parameters is mapped to 2nd formal parameter…etc.

- 1st actual output is mapped to 1st formal output…etc.

• Consider a function declared with formal parameters fp1…fpm and outputs fo1…fon that is called with actual parameters p1…pm

25.3 Parameter Association Example

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

% sumProdDiff - SUM, PRODUCT & DIFFERENCE OF TWO ARRAYS

% [s p d] = sumProdDiff(a,b)

% Return the scalar operator sum (s), product

% (p) and difference (d) of two arrays a and b.

% A simple function constructed to illustrate

% parameter association (formal vs. actual).

% Author: Spike

% Date: 24/3/1999

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

function [s, p, d] = sumProdDiff(a,b)

if size(a)~=size(b)

warning('Arrays are different sizes');

s=[], p=[], d=[];

end;

s=a+b;

p=a.*b;

d=a-b;

>> vec1=[1 2 3 4 5 6];

>> vec2=[7 6 5 4 3 2];

>> [sum product difference]=sumProdDiff(vec1,vec2)

sum =

8 8 8 8 8 8

product =

7 12 15 16 15 12

difference =

-6 -4 -2 0 2 4

Mapping

vec1<->a, vec2<->b

sum<->s, product<->p, difference<->d

Parameter Assoc. Example (Cont.)

>> [add multiply subtract]=sumProdDiff([1 2],[3 4])

add =

4 6

multiply =

3 8

subtract =

-2 -2

Mapping

[1 2]<->a, [3 4]<->b

add<->s, multiply<->p, subtract<->d

>> sumProdDiff([1 2],[3 4])

ans =

4 6

Mapping

[1 2]<->a, [3 4]<->b

ans<->s

>> [sum prod]=sumProdDiff([1 2],[3 4])

sum =

4 6

prod =

3 8

Mapping

[1 2]<->a, [3 4]<->b

sum<->s, prod<->p

25.4 Pass by Value & Pass by Reference

• Matlab functions are provided copies of the arguments being passed:

- this is called pass by value

• The implications are that:

- the function may alter the value of its copy of what was passed to it

- any changes made to arguments do not propagate back to the caller

• The complement of pass by value is that known as pass by reference

- not supported by Matlab

- many (most?) 3rd generation procedural languages support both methods of passing

- function/procedure is provided a pointer (reference) to the actual passed value

- if function/procedure makes change to a parameter then that change propagates back to the caller

25.5 Pass by Value Example

% try2Change - TRY TO CHANGE THE ARGUMENT

% y=try2Change(x)

% A simple fn. that return x+1 but

% also tries to increment x itself

% by 1.

Kami ada di Jakarta Selatan. KAMI MEMBERIKAN KURSUS MATLAB ONLINE - HUBUNGI MASTER ENGINEERING EXPERT (MEE) 081219449060. Kami membuka kursus Matlab untuk pemula dan mahasiswa atau insinyur yang ingin memperdalam Matlab dan menerapkan dalam bidang teknikal, engineering, rekayasa, dsb. Format bimbingannya tugas-tugas yang bisa membantu Skripsi, Tesis, DISERTASI
Bimbingan dilakukan secara online bisa lewat WA atau email
Dijamin Bisa, atau bisa mengulang kembali. Kami juga dapat membantumembuatkan aplikasi atau program matlab/lainnya. Anda akan dilatih oleh Tim Profesional - HUBUNGI MASTER ENGINEERING EXPERT (MEE) 081219449060. Email: kursusmatlab@gmail.com

% Used as an example of the pass by

% value paradigm of Matlab.

% Author: Spike

% Date: 24/3/1999

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

function y = try2Change(x)

x=x+1;

fprintf('Inside Function, x=%g\n',x);

y=x;

>> num=7

num =

>> returned=try2Change(num)

Inside Function, x=8

returned =

>> num

num =

>> x=0

x =

>> try2Change(x)

Inside Function, x=1

ans =

>> x

x =

25.6 Scope Rules & Workspaces

• Each function has its own workspace:

- area for its own variables, parameters, outputs

- functions can only see (access and alter) variables in their own workspace

• In practice this means:

- functions can’t access the variables in other functions

- functions can’t access the values of their caller’s variables (except those explicitly passed to them) nor of the base workspace

• The workspace of a function only exists while the function is running

- before the function executes its variables do not exist

- after the function terminates its variables do not exist

• The scope of a variable is simply that part of a program in which the variable exists (can be seen)

• Script M-files are not functions,  they run in the workspace of their caller

25.7 Scope Example

% scopeExample - EXAMPLE OF VARIABLE SCOPE

% z=scopeExample(x,y)

% What variables can and can't be seen.

% Author: Spike

% Date: 24/3/1999

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

function z = scopeExample(x,y)

z=x+y;

fprintf('In function: x=%g, y=%g, z=%g\n',x,y,z);

>> a=10; b=20; c=30;

>> fprintf('Before Call, a=%g, b=%g, c=%g\n',a,b,c);

Before Call, a=10, b=20, c=30

>> fprintf('Before Call, x=%g, y=%g, z=%g\n',x,y,z);

??? Undefined function or variable 'x'.

In function: x=20, y=30, z=50

>> fprintf('After Return, a=%g, b=%g, c=%g\n',a,b,c);

After Return, a=50, b=20, c=30

>> fprintf('After Return, x=%g, y=%g, z=%g\n',x,y,z);

??? Undefined function or variable 'x'.

>> x=40

x =

>> fprintf('Before Call, a=%g, b=%g, c=%g, x=%g\n',a,b,c,x);

Before Call, a=50, b=20, c=30, x=40

>> x=scopeExample(a,x);

In function: x=50, y=40, z=90

>> fprintf('After Return, a=%g, b=%g, c=%g, x=%g\n',a,b,c,x);

After Return, a=50, b=20, c=30, x=90

25.8 Run-Time Structure & the Stack

• Dynamic allocation (function starts) and de-allocation (function terminates) of workspaces requires a dynamic data structure

- functions can call other functions which call yet others or call themselves etc.

• A stack is employed to support function/procedure calls in programming languages

- like a physical stack of items (LIFO)

- as functions start their workspaces are added (pushed) to the stack (space in memory is allocated to them)

- as functions terminate their workspaces are removed (popped) from the stack (the memory is freed)

25.9 Return, Error & Warning

• By default functions terminate (return) after the last statement in the function M-file

• User’s can alter this behaviour

return Terminate execution of the function at this point and return to caller

if evalFinished(x,y,z) # are we done?

return;

else

[x y z]=update(x,y,z);

end;

error Print an error message and terminate all execution, returning to base workspace

if abs(y)<eps

error(‘Divide by zero error’);

else

z=x/y;

end;

• Functions may also report a warning but continue execution. Differs from simply using using disp() because user can turn on and off with warning on and warning off

if abs(y)<eps

warning(‘Div 0 Error. Z set to 0’);

z=0;

else

z=x/y;

end;

25.10 nargin & Variable Inputs

• Function calls may be made with less arguments than declared in the function

- not a syntax error (in Matlab)

- can be useful if arguments represent options or for which clear defaults exist

• Function needs a mechanism to detect how many arguments it has received

- each function can call the function nargin to determine the number of arguments received

- hopefully do something appropriate if less than desired number received

function result = narginEx(arg1,arg2,arg3)

if nargin==3

fprintf(‘Received all three arguments’);

elseif nargin==2

disp(‘3rd argument missing’);

elseif nargin==1

disp(‘Received only one of three args’);

else

disp(‘No arguments received.’);

end;

25.11 nargout & Variable Outputs

• It is also possible for the caller of a function not to save all the values that the function returns

- by default the additional values are ignored

- if the caller does not save any values then the first returned value is always placed in ans

• However it is possible for a function to detect how many of its outputs are being saved by the caller

- use the function call nargout

- function might modify its behaviour based on number of values caller wants

function [o1,o2] nargoutEx(arg1,arg2,arg3)

res1=arg1.*arg2-arg3;

res2=arg2.*arg3+arg1;

if nargout==2

o1=res1;

o2=res2;

elseif nargout==1

o1=[res1 res2];

else

fprintf(‘%f’,[res1 res2]);

end;

25.12 nargin & nargout Example

A function to measure the performance of a single set or commands or contrast the performance of two sets of commands.

Performance can be measured solely as CPU time, or CPU time plus User time.

For a single command the CPU time and User time should be returned. For two commands the ratio of times should be returned.

Useful for contrasting two alternate implementations of the same problem.

% perfMeasure - MEASURE THE TIME PERFORMANCE OF A FUNCTION

% [cpu user]=perfMeasure(commands)

% Return the cpu and user time taken to

% complete "commands".

% cpu=perMeasure(commands)

% As above but only the cputime

% perMeasure(commands)

% As above but prints a message to the screen

% concerning the CPU time.

% [cpu user]=perMeasure(commands1,commands2)

% Returns the ratio of cpu time and user time

% as taken by commands1 contrasted with

% commands2

perfMeasure (Cont)

% cpu=perMeasure(commands1,commands2)

% As above but only the ratio of CPU times is

% returned.

% perMeasure(commands1,commands2)

% Prints a message about the relative CPU times

% to the display.

% For example:

% shell1Time=perfMeasure('shell1');

% grabs the cpu time for running the M-file

% shell1.

% Written to illustrate nargin and nargout.

% A couple of points:

% * eval is used to run the list of commands

% passed by the caller.

% string to be eval-ed so that the

% overhead of calling eval is not added to

% the times.

% * For some reason Matlab seems to hate

% expressions of the form

% value=eval(string); Hence I've

% been reduced to calling eval and allowing

% the returned values to be stored in the

% default ans. I then access them from

% there. I can see no reason why the above

% syntax is disallowed but it stores the

% value in ans.

% Author: Spike

% Date: 25/3/1999

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

perfMeasure (Cont)

function [cpu, user] = perfMeasure(comm1,comm2)

%%%%%%%%%%%%%%%%%%%%%%%%%%

% No commands to run. Quit.

if nargin==0

error('Must have at least one command list to process');

end;

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

% Know that there is at least one set of

% commands to run. However the timing

% information required depends on the number of

% outputs the user has requested. If two

% outputs then we want both cpu & user.

% Otherwise only cpu.

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

if nargout==2

evalString=['ctime=cputime; tic;' comm1 '; utime=toc; [cputime-ctime utime];'];

eval(evalString);

ctime1=ans(1); % Cpu time for comm1

utime1=ans(2); % User time for comm1

else

evalString=['ctime=cputime;' comm1 ';cputime-ctime;'];

eval(evalString);

ctime1=ans;

end;

perfMeasure (Cont)

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

% If there is a 2nd set of commands to run then

% repeat the above operations.

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

if nargin==2

if nargout==2

evalString=['ctime=cputime; tic;' comm2 '; utime=toc; [cputime-ctime utime];'];

eval(evalString);

ctime2=ans(1);

utime2=ans(2);

else

evalString=['ctime=cputime;' comm2 ';cputime-ctime;'];

eval(evalString);

ctime2=ans;

end;

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

% Now all the work has been done determine the

% form of output required. If no returned value

% then print a message. If only 1 returned

% value then its the cpu time. Otherwise (2)

% then both cpu and user time.

% In all cases it is an absolute measure of

% time if only 1 command was to be run, and a

% ratio if two commands were run.

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

if nargout==0 % Output to Screen

if nargin==1

fprintf('CPU time taken: %6.2f seconds\n',ctime1);

else

fprintf('Ratio of CPU times: %5.2f\n',ctime1/ctime2);

end;

perfMeasure (Cont)

elseif nargout==1 % Only CPU time

if nargin==1

cpu=ctime1;

else

cpu=ctime1/ctime2;

end;

else % CPU & User time

if nargin==1

cpu=ctime1;

user=utime1;

else

cpu=ctime1/ctime2;

user=utime1/utime2;

end;

>> perfMeasure('shell2','shell3');

% <Outputs of shell2 & shell3 deleted>

Ratio of CPU times: 89.00

>> perfMeasure('shell3');

% <Outputs of shell3 deleted>

CPU time taken: 0.02 seconds

>> [cpu user]=perfMeasure('shell2')

% <Outputs of shell2 deleted>

cpu =

1.8100

user =

1.8191

25.13 Global Variables

• It is possible in Matlab to access variables outside the workspace of a function

- it is generally very bad programming practice: anything a function needs should be explicitly passed to it and anything the caller needs should be explicitly returned

- The methods are shown here merely for completeness

• One approach is the global operator

- syntax: global <variable-name>

- allows user access to the variable <variable-name> in the special “global” workspace

- TIC & TOC are implemented this way (see text)

In function1

global specialValue; % specialValue now in

: “global” workspace

In function2

global specialValue % Can now access the

: % variable specialValue

% stored in the global

% workspace % created by fn1

• Other options are evalin and assignin

25.14 Efficiency Issues

• Number of efficiency issues concerning functions of which users should be aware

• 1st time a function is invoked (called) in a session it must be loaded in:

- must be “compiled” into an internal (to Matlab) micro-code

- micro-code version of function is available to Matlab throughout rest of session

- hence first invocation of a function is somewhat slower than all subsequent

• The micro-code (pcode) version of a function can be saved to disk

- use pcode <functionName> command

- creates a file <functionName>.p

- loaded in deference to the .m file

- a means of distributing functions/scripts without letting users see the code

• Arguments to Matlab functions are not copied until the function modifies the argument

- performance implications for large arrays

25.15 varargin & varargout

• It is possible for functions to be written to accept any number of inputs (examples include Matlab’s max()) and produce any number of outputs

• Based on the special argument varargin and special output varargout

- cell arrays (not covered in lectures) that contain the individual elements specified by the caller

% mult - MULTIPLY AN ARBITARY NUMBER OF ARRAYS

% y=mult(a1,a2,...)

% Uses vector/scalar multiplication to

% multiply an arbitrary number of inputs.

Kursus Matlab di Jakarta Selatan KURSUS MATLAB ONLINE Skripsi, Tesis, DISERTASI 081219449060

Selasa, 13 Januari 2015

KURSUS MATLAB ONLINE Skripsi, Tesis, DISERTASI 081219449060 - MULTIPLY AN ARBITARY NUMBER OF ARRAYS AND function tempPlot(temps)

Tidak ada komentar:

Posting Komentar