I have a 2D equation, for example y = sin(x + t). For each unique value of t, I would like to plot a 1D realization of y. For example, if x = 0:0.1:2*pi and t = 1:10, for each value of t I would like to plot y = sin(x + t) for x = 0:0.1:2*pi. Basically, I would like to have lines along one direction for each value of t.
Is there a way I can do this in MATLAB?
Would something like a waterfall plot be beneficial for your case? Given a vector of x coordinates, for each unique value of y (in your case t), it would plot a one-dimensional realization of that curve. First, you would generate a 2D grid of coordinates X, Y where each row of X and Y together would be a vector of x coordinates for one realization of y and you'd plot all of these together in one plot.
Something like this:
[t,x] = meshgrid(0:0.1:2*pi, 1:10);
waterfall(t, x, sin(x + t));
view(-50, 50); % Adjust for a better view
xlabel('x'); ylabel('t'); zlabel('y'); % Add axis labels
We get this plot:
If you don't desire the "vertical" baselines that you see in the plot, then you can get away with using surf by specifying some additional properties to it:
[t,x] = meshgrid(0:0.1:2*pi, 1:10);
surf(t, x, sin(x + t), 'FaceColor', 'white', 'EdgeColor', 'interp', 'MeshStyle', 'row');
view(-50, 50);
xlabel('x'); ylabel('t'); zlabel('y');
The FaceColor and EdgeColor attributes are there to mimic what you see in the waterfall plot. Each visualization has a white face and the amplitude colours are interpolated. What is important is the MeshStyle attribute where you want to display the edges of the plot row wise. The default way for mesh is to show both rows and columns, so you'll visualize your plot in a grid like pattern, which is not what you want. Setting MeshStyle to row will simulate the waterfall plot but without the vertical baselines that you see in that plot.
You'll get:
Related
I have a set of n=8000 cartesian coordinates X,Y and Z as vectors and also a vector V of same size which I want to use as values to create a heatmap on a sphere.
I saw this link (visualization of scattered data over a sphere surface MATLAB), but I don't understand how I convert this set of data into a meshgrid for plotting using surf.
Almost every example I saw uses meshgrids.
Right now, I am doing by plotting a sphere and then use scatter3 to plot my points as big balls and try to smooth them later. I looks like this:
I would like to get the figure as the plotting of the example in that link, where he uses:
k = 5;
n = 2^k-1;
[x,y,z] = sphere(n);
c = hadamard(2^k);
surf(x,y,z,c);
colormap([1 1 0; 0 1 1])
axis equal
EDIT:
(Sorry for taking so long to reply, the corona crises kept away from work)
What I am actually doing is:
for i=1:numel(pop0n)
ori(i,:)=ori(i,:)/norm(ori(i,:));
end
x = ori(:,1);
y = ori(:,2);
z = ori(:,3);
%// plot
m=100;
[aa,bb,cc] = sphere(m);
surf(aa,bb,cc,ones(m+1,m+1)*min(pop0n))
hold on
colormap jet;
scatter3(x,y,z,400,pop0n/norm(pop0n),'filled');
colorbar
shading interp
The array 'ori' is 8000x3, and contains the x, y and z coordinates of the points I want to plot and pop0n is a 8000 sized vector with the intensities of each coordinate.
My main question is how do I transform my x, y, z and pop0n, that are vectors, into 2D arrays (meshgrid) to use surf?
Because I cannot simply do surf(x,y,z,pop0n) if they are vectors.
Thanks in advance
As David suggested, griddata does the job.
What I did was:
for i=1:numel(pop0n)
ori(i,:)=ori(i,:)/norm(ori(i,:));
end
x = ori(:,1);
y = ori(:,2);
z = ori(:,3);
%// plot
m=100;
[aa,bb,cc] = sphere(m);
v = griddata(x,y,z,pop0n,aa,bb,cc,'nearest');
surf(aa,bb,cc,v)
colormap jet;
colorbar
shading interp
I wanted to generate a plot (X vs Y), and Z values depend on the Y. The example is shown in the figure below. The matrix size of X is same with Z but not Y. I can plot Z against X, but I wanted to combine all the plot into a single plot and become Y against X. I can plot multiple plots into a single plot but the plot is overlapping each other.
My question is there any method I can merge multiple plots into a single plot without overlapping each plot as the difference between each plot is very small (e.g Z1=1,2,3,4,5 and Z2=1.0001,2.0002,3.0001,4.0002,5.0001). So, I wanted to set each Z plot at different Y axis. (e.g Z1 at Y=0, Z2 at Y=2 ...)
Does anyone have any suggestions or idea?
Thank You
I'll clarify the ideas I wrote in a comment.
First, let's get some data:
x = 470:0.1:484;
z1 = cos(x)/2;
z2 = sin(x)/3;
z3 = cos(x+0.2)/2.3;
I'll plot just three data sets, all of this is trivial to extend to any number of data sets.
Idea 1: multiple axes
The idea here is simply to use subplot to create a small-multiple type plot:
ytick = [-0.5,0.0,0.5];
ylim = [-0.9,0.9]);
figure
h1 = subplot(3,1,1);
plot(x,z1);
set(h1,'ylim',ylim,'ytick',ytick);
title('z1')
h2 = subplot(3,1,2);
plot(x,z2);
set(h2,'ylim',ylim,'ytick',ytick);
title('z2')
h3 = subplot(3,1,3);
plot(x,z3);
set(h3,'ylim',ylim,'ytick',ytick);
title('z3')
Note that it is possible to, e.g., remove the tick labels from the top two plot, leaving only labels on the bottom one. You can then also move the axes so that they are closer together (which might be necessary if there are lots of these lines in the same plot):
set(h1,'xticklabel',[],'box','off')
set(h2,'xticklabel',[],'box','off')
set(h3,'box','off')
set(h1,'position',[0.13,0.71,0.8,0.24])
set(h2,'position',[0.13,0.41,0.8,0.24])
set(h3,'position',[0.13,0.11,0.8,0.24])
axes(h1)
title('')
ylabel('z1')
axes(h2)
title('')
ylabel('z2')
axes(h3)
title('')
ylabel('z3')
Idea 2: same axes, plot with offset
This is the simpler approach, as you're dealing only with a single axis. #Zizy Archer already showed how easy it is to shift data if they're all in a single 2D matrix Z. Here I'll just plot z1, z2+2, and z3+4. Adjust the offsets to your liking. Next, I set the 'ytick' property to create the illusion of separate graphs, and set the 'yticklabel' property so that the numbers along the y-axis match the actual data plotted. The end result is similar to the multiple axes plots above, but they're all in a single axes:
figure
plot(x,z1);
hold on
plot(x,z2+2);
plot(x,z3+4);
ytick = [-0.5,0.0,0.5];
set(gca,'ytick',[ytick,ytick+2,ytick+4]);
set(gca,'yticklabel',[ytick,ytick,ytick]);
text(484.5,0,'z1')
text(484.5,2,'z2')
text(484.5,4,'z3')
The simplest would be to shift Z data. But note that Z2 would look like to be oscillating around 1 - so this is a neat visual representation, but might mislead.
% Simple version - shift Z curves by 0, 1, ... (as recommended by #Cris Luengo)
shiftMat = repmat(0 : size(Z, 2)-1, size(Z,1), 1);
Z = Z + shiftMat;
%Min shift up to have non-overlapping - curves touching
for i = 2 : size(Z, 2)
Zdif = (Z(:, i-1) - Z(:, i));
Z(:, i) = Z(:, i) + max(Zdif); % + 0.01 to separate them a little bit.
end
%Bigger shift up, to have all points of Z(2) equal or above all points of z1.
for i = 2 : numZ
Zdif = max(Z(:, i-1))-min(Z(:, i));
Z(:, i) = Z(:, i) + Zdif;
end
Another possibility is to have multiple Y axis and each Z curve plotted against its own Y axis. This is likely fancier and shouldn't mislead, but it is way more work, even after you grab the function, as you still need to position all those axes. MATLAB by default lets you use only 2 axes, so grab a function from fileexchange to add more: https://www.mathworks.com/matlabcentral/fileexchange/9016-addaxis
I would like to achieve a 3D surface plot. My outputs are as follows.
For a particular value of z, I get y for each value of x (x ranges like 0:0.1:1.4).
Then I vary z and, for the same range of x, I get y values.
The result can be visualised as 2D plots at discrete z values, consisting of the range of x and its corresponding y. Here is my original plot:
I would like to create a 3D surface plot instead, like a blanket wrapped over the above 2D plots.
Here's an example for the two types of plots:
figure
hold on
grid on
view(30,40)
x = 0:.01:4;
z = .3:.3:3;
y = NaN(numel(x), numel(z));
for k = 1:numel(z)
y(:,k) = abs((4-x).*sin(x/(1+z(k)))); % compute a vector as function output
% for input vector x, for each z. Store as a column in matrix y
plot3(x,repmat(z(k),size(x)),y(:,k)) % plot in 3D space
end
figure
surf(x,z,y.','edgecolor','none') % surface plot
view(30,40)
grid on
This question already has answers here:
Change color of 2D plot line depending on 3rd value
(5 answers)
Closed 5 years ago.
How do you create a color gradient in Matlab such that you plot a 2D line plot of y=y(x), and you color it using another variable that also depends on x such that z=z(x). A scatter or point plot is also fine by me.
I would also like to have a colormap legend kind of thing showing the color gradient and it's actual representation of z. This stuff is quite common in visualisation tools such as VisIt and ParaView but I could not yet FIGURE it out in Matlab.
If a scatter plot is fine, you can use the 4th input to scatter:
x = -10:0.01:10;
y = sinc(x);
z = sin(x);
scatter(x,y,[],z,'fill')
where z is the color.
The only way I know of to do this is with a little trick using surf:
% Create some sample data:
x = cumsum(rand(1,20)); % X data
y = cumsum(rand(1,20)); % Y data
z = 1:20; % "Color" data
% Plot data:
surf([x(:) x(:)], [y(:) y(:)], [z(:) z(:)], ... % Reshape and replicate data
'FaceColor', 'none', ... % Don't bother filling faces with color
'EdgeColor', 'interp', ... % Use interpolated color for edges
'LineWidth', 2); % Make a thicker line
view(2); % Default 2-D view
colorbar; % Add a colorbar
And the plot:
To manipulate the color of the line continuously, you'll want to use surface.
While at first look, this function looks most useful for plotting 3d surfaces, it provides more flexibility for line coloring than the basic plot function. We can use the edges of the mesh to plot our line, and take advantage of the vertex colors, C, to render interpolated color along the edge.
You can check out the full list of rendering properties, but the ones you are most likely to want are
'FaceColor', 'none', do not draw the faces
'EdgeColor', 'interp', interpolate between vertices
Here's an example adapted from MATLAB Answers post
x = 0:.05:2*pi;
y = sin(x);
z = zeros(size(x)); % We don't need a z-coordinate since we are plotting a 2d function
C = cos(x); % This is the color, vary with x in this case.
surface([x;x],[y;y],[z;z],[C;C],...
'FaceColor','none',...
'EdgeColor','interp');
Generate a colormap, e.g. jet(10). The example would generate a 10*3 matrix.
Use interp1 to interpolate between the values in the RGB space using your data by setting the first color as the lowest value and the last color as the highest value. This would generate a n*3 matrix where n is the number of data points.
Use scatter with the optional parameter c to plot the points with the interpolated colors.
activate colorbar to show the colorbar.
Using a dataset of the format [x,y,z,value], I can create a 3D scatter plot as shown in the images, where the color of each point at location (x,y,z) is based on value. Note that all of the images are of the same plot, just from different views. It is intended to be the surface of an octant of a sphere.
Is there a way I can do an interpolation of the colors in 3D such that we see a solid surface instead of individual points? I am looking for something along the lines of imagesc, but in 3D. I've tried a wide variety of functions, including scatteredInterpolant, patch, mesh, and surf, but those do not seem to work in 3 dimensions the way I'd like.
Based on your example data (with multiple shells), I put this program together. It uses trisurf so it doesn't need meshed data.
function PlotColoredSpheres(x,y,z)
tol = 0.01;
r = sqrt(x.^2 + y.^2 + z.^2);
data = num2cell(sortrows([r, x, y, z], [1,2,3,4]),1);
[R, X, Y, Z] = data{:};
figure(1)
RUnique = uniquetol(r, tol);
for Ru = RUnique(:)'
j = abs(R - Ru) < tol;
tri = delaunay(X(j),Y(j));
trisurf(tri, X(j), Y(j), Z(j));
lighting phong
shading interp
hold on
end
hold off
end