Matlab


I think you're looking for scatteredInterpolant.

Basic Matlab

X./a         divide every element of X by a (piecewise)
A.*B   element vise multiplication of A and B (both same size)
format long;     format short;
a = pi;  sprintf('%0.3f', a)   => 3.142

size(A, 1)   size(A, 2)
length(A)    size of longest dimension

who / whos          what variables I have in memory right now
load('featuresX.mat')       clear XVariable       clear that specific variable
save hello.mat   v;     save variable v as mat file     
save hello.txt v -ascii;          human readble form

A = [1 2; 3 4; 5 6];
A([1 3], :)    Specific indexing:   rows with index 1 and 3 , all columns      1 2 ;  5 6
A( :, 2)  =  [10; 11; 12]      assign to a range

log(A)  elemen-vise log;          exp(A)   element-vise        abs(A)      -A      A + 1   (element-vise) add one to all            A < 3     element-vise operation (results in a matrix of zeros and ones)     
find(A<3)   returns indices of elements that satisfy that         [r c] = find(A <= 7)    r row, c column
sum(A)       prod(A)  product of elements.       floor/ceil            max (rand(3) rand(3))      elemnt-vise max                      max(A, [], 1)   max of columns                 max(A, [], 2)    max of rows        max(max(A))    OR   max(A(:))   global maximum
A(:) turns A to a column matrix
A = magic(9)   sum(A, 1) column-vise sum  sum(A, 2) row-vise sum        sum(sum(A .* eye(9)))  lement-vise multiplication of diagonals to get the diagon and sum
flipud   flip upside down

Plotting

t = [0:0.01:4];   y1 = sin(2 * pi * t);   plot(t, y1);  hold on;     y2 = cos(2 * pi * t);   plot(t, y2, 'r');   xlabel('Time'); ylabel('value');  legend('sin', 'cos');  title('My Plot');         print -dpng  'myplot.png''           help plot to save other formats
figure(1); plot(t, y1);     figure(2); plot(t, y2);     two figures
subplot(1,2, 1);  plot(t, y1); subplot(1,2, 2);   plot(t, y2);% Divides plot a 1*2 grid and access the 1st subplot
axis(0.5 1 -1 1)    set range of plot
clf; % clear figure

imagesc(magic(9));                   imagesc(magic(15)) ,  colorbar , colormap gray


for/if/while/function
 indices = 1:10
for i = indices
  disp(i)
end

 while i < 5

v(i) = 100;
i = i+1;
end
 if i == 6

end
 if i == 1

elseif i ==2

else

end

addpath('/home/morteza/');    add path fopr matlab to lookup

function [y1, y2] = squareandcube(x)
y1 = x^2
y2 = x^3


[a, b] = squareandcube(5)

vectorization   (much more efficient! than for loops sigmas etc)

Σ_i θ(i) X(i)   ->    X' θ






Matlab Doc

--------------------------
close all; clear; clc

indexes start at 1. Add comments using the percent (%) symbol.
a = 1
a = [1 2 3]    single row vector
a = [1 2 3; 4 5 6; 7 8 10]   matrix, each row separated by semi-colon
Another way to create a matrix is to use a function, such as ones, zeros, or rand. For example, create a 5-by-1 column vector of zeros.
z = zeros(5, 1)      5 rows, 1 column
Apply a single arithmetic function on all elements of a matrix
a + 10      will add 10 to all elements of a
sin(a)
To transpose (use apastrophy character)  a'
To inverse   inv(a)          a * inv(a)   is identity matrix
To display more decimal digits
format long
format short
To perform element-wise multiplication rather than matrix multiplication, use the .* operator:
p = a.*a     will multiple element by its associative element. You can also use: a.^3   multiplication, division, and power can operate element-wise in this format.
Concatenation
 In fact, you made your first array by concatenating its individual elements. The pair of square brackets [] is the concatenation operator.
[a a]  will concatenate a to itself in a row. is equal to [a, a]
[a ; a]  will concatenate a to itself in a column fashion (one on top of another)
Complex numbers: To represent the imaginary part of complex numbers, use either i or j.
c = [3+4i, 4+3j, -i, 10j]
Array Indexing
4-by-4 magic square A = magic(4)                             random generation:  rand(3,5,2);
provide row and column number:     A(4, 2)      = 14           Less common, but sometimes useful, is to use a single subscript that traverses down each column in order:A(8)=14
Index out of bound: if the array is on right hand side of equal, you get error test = A(4,5) . if array is on left hand side, . The size of the array increases to accommodate the newcomers. A(4,5) =17
Range
To refer to multiple elements of an array, use the colon operator, which allows you to specify a range of the form start:end. For example, list the elements in the first three rows and the second column of A:
A(1:3, 2)
The colon alone, without start or end values, specifies all of the elements in that dimension. For example, select all the columns in the third row of A:
A(3, :)
The colon operator also allows you to create an equally spaced vector of values using the more general form start:step:end.B = 0:10:100         0 10 20 ....
Workspace variables
save myfile.mat
load myfile.mat
Strings
'string'      'You''re right'           concatenate strings just as you concatenate numbers to make a matrix               ['hello world', ' - ' , 'you''re right' ]   hello world - you're right
To convert numeric values to strings, use functions, such as num2str or int2str.
functions
When there are multiple output arguments, enclose them in square brackets:
[maxA,location] = max(A);
To display a text message
disp('message for you')        clc  to clear screen
Plot
x = 0 : pi/100 : 2*pi
y = sin(x)
plot(x, y)
xlabel('x)
ylabel('sin(x)')
title('Plot of the Sine Function')
By adding a third input argument to the plot function, you can plot the same variables using a red dashed line.  plot(x,y,'r--')   such as a +, o, or *. For example, 'g:*' requests a dotted green line with * markers.
Notice that the titles and labels that you defined for the first plot are no longer in the current figure window. By default, MATLAB® clears the figure each time you call a plotting function, resetting the axes and other elements to prepare the new
plot.
x = 0:pi/100:2*pi;
y = sin(x);
plot(x,y)                        plot(X(:, 2), y, '*')

hold on        # add the next plots to current plot.        figure; make a new window for the plot.

y2 = cos(x);
plot(x,y2,'r:')
legend('sin','cos')
3D Plots
Three-dimensional plots typically display a surface defined by a function in two variables, z = f (x,y).
[X,Y] = meshgrid(-2:.2:2);         create the base x, y
Z = X .* exp(-X.^2 - Y.^2);        calculate  the function
surf(X,Y,Z)                              to draw a surface plot
Sub-plots
You can display multiple plots in different subregions of the same window using the subplot function.For example, create four plots in a 2-by-2 grid within a figure window.
t = 0:pi/10:2*pi;
[X,Y,Z] = cylinder(4*cos(t));
subplot(2,2,1); mesh(X); title('X');
subplot(2,2,2); mesh(Y); title('Y');
subplot(2,2,3); mesh(Z); title('Z');
subplot(2,2,4); mesh(X,Y,Z); title('X,Y,Z');
Scripts
edit portland                     creates a script file named portland.m
n = 50;
r = rand(n,1);
plot(r)

m = mean(r);
hold on
plot([0,n],[m,m])      a plot with two points (0,m) and (n,m) ::::: [0, n] is the x points
hold off
title('Mean of Random Uniform Data')
To run the script, type its name at the command line:
portland
For loop, IF
nsamples = 5;
npoints = 50;

for k = 1:nsamples
    currentData = rand(npoints,1);
    sampleMean(k) = mean(currentData);
end
if overallMean < .49
   disp('Mean is less than expected')
elseif overallMean > .51
   disp('Mean is greater than expected')
else
   disp('Mean is within the expected range')
end

ftp://ftp.mathworks.com/pub/customer_pickup/UT/


dataMatrix(matrix, rowname, columnname) -> bundle all data together, can be used to show heat map


order of applying filters matters. they depend on percentile, but it seems like there is no standard saying you should apply this first and then the next filter on the result of first filter ....


on the editor click evaluate cell and advance



%% Heirarchical Clustering of profiles

%% makes cells


publish icon beside the print icon, runs a file cell by cell and makes a report containing outputs and figures


clc


to make gui type > guide

matlab graph theory


multialign : multiple sequence alignmrnt

bioindexedfile : for large data files, index based on a column

large file : memory map file, memory map object.

featuresmap

fastQ format



tic toc - exec time

vectorization-Techniques for Improving Performance

matlab is column major

blas lapak

spline curve fitting

profiler to give a report of execution time

textscan is faster than fscanf




additional process:
parfor-distributed
task job

optimset('useparallel','always');
Optimization toolbox


matlabpool open //assign number of matlab workers (duplicates of matlab thread





message passing to use recursive parfor


data parallel
code parallel

rand(4, codistributor)













Subpages (1): libsvm
Comments