FFNDGRID Fast 'n' Furious N-D data gridding.
CALL: [fgrid, xvec] = ffndgrid(x,f, delta, limits, aver );
fgrid = Matrix of gridded data.
xvec = cellarray of gridvectors
xl1:dx1:xu1 or linspace(xl1,xu1,Nx1) depending on delta.
x = [x1 x2,...xD] coordinate matrix for unevenly spaced data, f.
size NxD.
f = f(x), vector of function values length N.
delta = [dx1+i*pad, dx2 ,...,dxD] or [-Nx1+i*pad -Nx2,...,NxD], where
dx1 to dxD and Nx1 to NxD defines the stepsize of grids and
number of bins, respectively, in the D dimesional space. Empty
gridpoints are padded with IMAG(delta(1))=pad. (default -75)
limits = [xl1 xu1 xl2 ...xuN fl fu n0], contain the limits in the
D-dimensional x1-x2...xN-f-plane of where to grid. The last
parameter, n0, removes outliers from the data set by ignoring
grid points with n0 or less observations. When n0 is negative
it is treated as the percentage of the total number of data points.
(default [min(x1),max(x1),min(x2),.... max(xN),min(f),max(f),0])
aver = 0 If each value of fgrid is the sum of all points falling
within each cell.
1 If each value of fgrid is the average of all points falling
within each cell. (default)
FFNDGRID grids unevenly spaced data in vector f into a matrix fgrid.
NOTE: - The vector limits can be padded with NaNs if only
certain limits are desired, e g if xl1 and fl are wanted:
ffndgrid(x, f, [],[.5 nan nan nan 45])
- If no output arguments are given, FFGRID will plot the gridded
function with the prescribed axes using PCOLOR.
Examples:
N = 500;D=2; sz = [N ,D ];
x = randn(sz); z = ones(sz(1),1);
[nc, xv] = ffndgrid(x,z,-15,[],0); % Histogram
pcolor(xv{:},nc) %
[XV,YV]=meshgrid(xv{:});
text(XV(:),YV(:),int2str(nc(:)))
dx = [diff(xv{1}(1:2)) diff(xv{2}(1:2))];
contourf(xv{:}, nc/(N*prod(dx))) % 2-D probability density plot.
colorbar
colormap jet
See also: griddata