I generate some points with latin hypercube sampling and also by for loop I tried to separate out(i,j) to 4 kinds for plotting .but it only plots like below picture.

rng default
LHsampling=lhsdesign(100,5,"criterion","correlation","smooth","off"); %criterion','correlation'  give "equal distribution".


for i=1:size(LHsampling,1)
   for j=1:size(LHsampling,2)
       
    out(i,j)=(LHsampling(i,j)*2+12*LHsampling(i,j)^2)^3;

        if out(i,j)>5
        scatter(size(LHsampling,1),out(i,j),'s','r');
        hold on;
        elseif out(i,j)>50
        scatter(size(LHsampling,1),out(i,j),'o','b');   
        hold on
        elseif out(i,j)>100
        scatter(size(LHsampling,1),out(i,j),'m','*');
        hold on
        else
        scatter(size(LHsampling,1),out(i,j),'x','g ');    
    
        end
    
   end

[sogol bandekian]

@Star Strider please look at this

[Torsten]

Although I'm not Star Strider I can tell you that "out" has only 1 column, thus out(:,2) does not exist.

Answer 1

[Walter Roberson]

        if out(i,j)>5
        scatter(size(LHsampling,1),out(i,j),'s','r');
        hold on;
        elseif out(i,j)>50

Under what circumstances can out>5 fail but then out>50 succeeds?

Why do you use the number of rows (size) as the x coordinates?

You scatter() once for each point. That is very inefficient.

I recommend that you calculate out once before the loop using vectorized calculations. Then use logical indexing to select all of points in a color range and scatter() all of them at the same time.

[sogol bandekian]

rng default
LHsampling=lhsdesign(100,5,"criterion","correlation","smooth","off"); %criterion','correlation'  give "equal distribution".


for i=1:size(LHsampling,1)
    for j=1:size(LHsampling,2)
        
        out=(LHsampling.*2+12.*LHsampling.^2).^3;
    end 
end
out(i,j)=(LHsampling(i,j)*2+12*LHsampling(i,j)^2)^3;

if out(i,j)>2
    scatter(size(LHsampling),out(i,j),'s','r');
    hold on;
elseif out(i,j)>20
    scatter(size(LHsampling),out(i,j),'o','b');   
    hold on
elseif out(i,j)>100
    scatter(size(LHsampling),out(i,j),'m','*');
    hold on
else
    scatter(size(LHsampling),out(i,j),'x','g ');    
    
end

you mean like this?

but this does not generate more than two points

[Walter Roberson]

I had to guess what you wanted to do.

rng default
LHsampling=lhsdesign(100,5,"criterion","correlation","smooth","off"); %criterion','correlation'  give "equal distribution".
out=(LHsampling.*2+12.*LHsampling.^2).^3;

mask = out <= 2;
[r, c] = find(mask);
scatter(r, out(mask), 'gx');
hold on

mask = out > 2 & out <= 20;
[r, c] = find(mask);
scatter(r, out(mask), 'rs');

mask = out > 20 & out <= 100;
[r, c] = find(mask);
scatter(r, out(mask), 'ob');

mask = out > 100;
[r, c] = find(mask);
scatter(r, out(mask), 'm*');

set(gca, 'yscale', 'log')
xlabel('row');
ylabel('out')

[sogol bandekian]

Thanks alot,you are great

[sogol bandekian]

why use mask instead of if?