How to filter noisy function ?

I have a gaussian function which outputs a 1-by-1001 matrix. It also has a noise. I want it to work 1000 times and to choose matrices with relatively less noise. Then, I want to create another matrix composed of the relatively less noisy matrices. How can I achieve this ?

    function [gf] = two_stars_random(C1,x01,s1,C2,x02,s2,N)

randCoeff=0.5;
a=(1-randCoeff)+2*randCoeff*rand(1);
C1t = a*C1;

a=(1-randCoeff)+2*randCoeff*rand(1);
x01t = a*x01;
x02t = x02+(x01t-x01);

a=(1-randCoeff)+2*randCoeff*rand(1);
s1t=a*s1;
a=(1-randCoeff)+2*randCoeff*rand(1);
C2t = a*C2;
a=(1-randCoeff)+2*randCoeff*rand(1);
s2t=a*s2;

N=1000;
range_begin = 0;
range_end   = 1;
x=range_begin:(1.0/N):range_end; 
g1=C1t*exp(-(x-x01t).^2/(2*s1t^2));
g2=C2t*exp(-(x-x02t).^2/(2*s2t^2));
b=0.2*rand(1)*(C1t+C2t)*rand(1,N+1);
gf=g1+g2+b;

end

[Jan]

It depends on what "relatively less noise" means and what kind of "compostion" you mean. Does the shown code help to understand what you want to achieve?

[Yasemin Derme]

There is a variable b in the function which adds noise to the function. I want the include outputs of the function in which the b is less than 0.1. Then I want to create another matrix composed of these outputs.

[Image Analyst]

"create another matrix composed of the relatively less noisy matrices" How? Concatenate them? Add them? How are we supposed to know what "create" means to you? Also, I don't see a loop over 1000 iterations where you create various arrays and inspect and keep track of the actual noise in each one, so how can you select the N least noisy arrays?

[Yasemin Derme]

I want to concatenate the outputs. The function that I wrote will be used inside another function which will repeat the mentioned process 1000 times.

[Image Analyst]

I'll try to look at it later today. In the meantime, please add comments saying what each chunk of code is doing, and add the program where you call this function 1000 times and concatenate the results, if you have it.

[Yasemin Derme]

Okay, thanks a lot for your help.

function [gf] = two_stars(C1,x01,s1,C2,x02,s2,N)

C1=27.8498 ; % either expands (C1>1) or shrinks (0<C1<1) the curve vertically, reflects with respect to x axis when C1<0

C2=54.6882; % either expands (C2>1) or shrinks (0<C2<1) the curve vertically, reflects with respect to x axis when C1<0

x01=95.7507; % either shift right(x01>0) or left(x01<0)

x02=96.4889; % either shift right(x02>0) or left(x02<0)

s1=15.7613; %either expand (if s1 increases) or shrink (if s1 decreases) the curve horizontally

s2=97.0593; %either expand (if s2 increases) or shrink (if s2 decreases) the curve horizontally

N=10000;

range_begin = -50;%start from x1 point

range_end = 50;%ends in x2 point

x=range_begin:(1.0/N):range_end;%determines the rate with which the function will take into account the points

g1=C1*exp(-(x-x01).^2/(2*s1^2));%gaussian function

g2=C2*exp(-(x-x02).^2/(2*s2^2));%gaussian function

gf=g1+g2;%the sum of two gaussian functions

plot(g1+g2+ 0.5*rand(size(x)))

[Yasemin Derme]

This is the function that I've been working on:

t=1001;

A = zeros(t-1,t);

for i =1:t-1

    stack(i,:) = two_stars_random(C1,x01,s1,C2,x02,s2,N);

    if b<0.1

       A = stack(i,t);

    elseif b>0.1

       A = zeros(1,t);

    end
    [r,~]=size(A);
    filtered(r,t) = A;
end