how to reduse the size of an array for meshgrid

I have dataset of x,y,z

x=[1:13670];
y=[1:13670];

when i meshgrid x and y,

[X,Y]=meshgrid(x,y);

I recieved the following error:

Error using repmat

Requested 135670x135670 (137.1GB) array exceeds maximum array size preference. Creation of

arrays greater than this limit may take a long time and cause MATLAB to become

unresponsive. See array size limit or preference panel for more information.

how can i solve this problem?

Do I need to do some iteration? if so, how?

what will Z look like? will it be,

Z=z(X,Y);

[Adam Danz]

Perhaps you could subsample the data. This example below uses every 100 samples.

x=[1:100:13670];
y=[1:100:13670];
[X,Y]=meshgrid(x,y);

[Rik]

So you are planning to have an array Z that is 135670 by 135670? Then you will probably need a big upgrade for your computer.

What is your end goal? Such an enormous array is hard to work with, so solving the problem in smaller chunks is probably much more efficient and will also likely require more feasible amounts of memory.

[Walter Roberson]

sometimes instead of meshgrid you can use bsxfun() or implicit expansion (R2016b or later) but those are only going to help reduce the number of temporary variables, whereas you have problems with even one copy of the array

In some cases, tall() arrays can help, especially in connection with implicit expansion. I am not convinced at the moment that your situation is suitable for tall() however.

Are you doing a finite element mesh simulation with that many elements per side?

[Kafayat Olayinka]

My end goal is to be able to create a contourf plot of

contourf(X,Y,Z)

It looks like in other for me to make this plot, I have to meshgrid x,y.

So that z can be dependant of x and y.

[Kafayat Olayinka]

the number 135670 is the an index lenght of X.

In my real problem, X represents time while Y represents height, and Z represents temperature.

x,y,z all have the same sizes. however, in other for me to create a contourf plot of temperrature, i need to meshgrid(x,y) so my problem now is how i can minimize the size array for meshgrid

[Adam Danz]

What about trying scatter(x,y,36,z,'filled')?

[Walter Roberson]

How large is your display? Few displays are more than 4096 wide so you would need 38+ of them on a side to display the contour plot at one pixel each.

It is not always necessary to use meshgrid for creating array of results; bsxfun or implicit expansion can substitute (but they do not have magic to store 137 gigabytes in 16 gigabytes of ram)

[Kafayat Olayinka]

my result when i plotted scatter(x,y,36,z,'filled')

However, I dont want the empty spaces to show. what should i do?

[Adam Danz]

I wonder if this would work for you

imagesc(x,y,z)

[Walter Roberson]

You have vectors of length 135670 that have integer x values 1 to 12, and corresponding y values over 200, and corresponding z values?

You should be using scatteredInterpolant()

[Kafayat Olayinka]

imagesc(x,y,z) did not work for me. I might have to meshgrid(x,y) first.

Answer 1

[Walter Roberson]

N = 100;
F = scatteredInterpolant(x, y, z);
xvals = min(x):max(x);
yvals = linspace(min(y), max(y), N);
[X,Y] = meshgrid(xvals, yvals);
Z = F(X, Y);
contour(X, Y, Z)

[Kafayat Olayinka]

i used contourf(X,Y,Z) instead and it worked!! thanks alot