Asked by Maurice
on 18 Apr 2011

hi there

I have a model that handles particles and lets them react. in the end, I have the coordinates in [x,y] where the particles center lies, it's radius and for instance its temperature.

I would now like to plot that particle at [x,y], as a dot representing its size (e.g. with MarkerSize?) and color by temperature.

Is there anything around that can actually do this? The problem is also that the particle size should be realistically displayed.

thank you for your help! maurice

*No products are associated with this question.*

Answer by Patrick Kalita
on 18 Apr 2011

You should be able to do this with the scatter command. The first two inputs are the `x` and `y` locations. The third input is the size of the marker to be drawn. And the fourth controls the color. The only tricky bit would be to scale the temperature data to a range of appropriate marker sizes.

Here's a very simply example that just scales the `T` data so that the third input argument is in the range [0 100]:

x = rand(1, 10); y = rand(1, 10); T = x + y; scatter(x,y,T./max(T) * 100, T,'filled')

Answer by Teja Muppirala
on 18 Apr 2011

I think the downside with using SCATTER is that the marker doesn't change size with zooming. Using PATCH might be another option:

N = 100; %Number of points

r = 0.2*rand(N,1); %Radii T = rand(N,1); %Temperatures p = randn(N,2); %Positions

colordef(figure,'black');

q = linspace(0,2*pi,32)'; C = [cos(q) sin(q)]; hold all; for n = 1:numel(r) patch(r(n)*C(:,1)+p(n,1), r(n)*C(:,2)+p(n,2),T(n),'edgecolor','none'); end axis equal; set(gca,'Clim',[0 1]); alpha 0.9 colorbar

Answer by Artur Palha
on 29 Jan 2013

Edited by Artur Palha
on 29 Jan 2013

A small change to the option Teja Muppirala gave can give a considerable speed up, especially when plotting many data points. The alternative is to plot all the patches at once in the following way:

N = 1000; %Number of points r = 0.2*rand(N,1); %Radii T = rand(N,1); %Temperatures p = randn(N,2); %Positions figure(1) clf(1) colordef(1,'black'); q = linspace(0,2*pi,32)'; C = [cos(q) sin(q)];

hold all;

% Teja Muppirala approach tic for n = 1:numel(r) patch(r(n)*C(:,1)+p(n,1), r(n)*C(:,2)+p(n,2),T(n),'edgecolor','none'); end toc axis equal; set(gca,'Clim',[0 1]); alpha 0.9 colorbar

hold off

figure(2) clf(2) colordef(2,'black');

hold all;

% faster approach

tic patch(bsxfun(@plus,bsxfun(@times,r',repmat(C(:,1),[1 N])),p(:,1)'), bsxfun(@plus,bsxfun(@times,r',repmat(C(:,2),[1 N])),p(:,2)'),T','edgecolor','none'); toc axis equal; set(gca,'Clim',[0 1]); alpha 0.9 colorbar

hold off

I made this change because I need to plot thousands of data points. for 1e5 data points I get a speedup of about 30x.

Opportunities for recent engineering grads.

## 0 Comments