Faster interpolation of 2-D spatial data at different time steps without looping through the time steps?

Question

Darcy Cordell 2023-1-16

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1895090-faster-interpolation-of-2-d-spatial-data-at-different-time-steps-without-looping-through-the-time-st

评论： Darcy Cordell 2023-1-16

I have a relatively small number of scattered data point locations (~30) which I want to interpolate onto a uniform spatial grid (e.g. 100 x 100). But the data set is also temporally varying, so I need to interpolate each spatial grid thousands (or tens of thousands) of times. Note, that I do not need to interpolate in time, I only need to interpolate spatially at each time step.

I am currently doing this with a for loop, but I am wondering if there is a way to do it faster in one step. See simple example below. Any help is appreciated!

%Create random data points:
xLoc = rand(30,1); 
yLoc = rand(30,1);
%Create random data at each point varying temporally with 10000 points:
nT = 10000;
Data = randn(30,nT);
%uniform grid to interpolate onto:
xInt = 0:0.01:1;
yInt = 0:0.01:1;
[X,Y] = meshgrid(xInt,yInt);
%Create interpolant:
F = scatteredInterpolant(xLoc,yLoc,Data(:,1));
%Now loop over each time stamp. <<<--- This is the part I want to speed up
%This example current takes about 1.25 min on my machine, but I presume could
%be faster if I avoided a loop.
tic
V = nan(length(X(:)),nT);
for i = 1:size(Data,2)
    
    F.Values = Data(:,i); 
    V(:,i) = F(X(:),Y(:));
end
toc

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Matt J 2023-1-16

在 MATLAB Online 中打开

Using func2mat from this FEX download,

https://www.mathworks.com/matlabcentral/fileexchange/44669-convert-linear-functions-to-matrix-form

%Create random data points:
xLoc = rand(30,1); 
yLoc = rand(30,1);
%Create random data at each point varying temporally with 10000 points:
nT = 10000;
Data = randn(30,nT);
%uniform grid to interpolate onto:
xInt = 0:0.01:1;
yInt = 0:0.01:1;
[X,Y] = meshgrid(xInt,yInt);
%Create interpolant:
F = scatteredInterpolant(xLoc,yLoc,Data(:,1));
tic
A=func2mat(@(D) func(D,F,X,Y), Data(:,1) );
V=A*Data;
toc
Elapsed time is 0.558389 seconds.
function V=func(D,F,X,Y)
 F.Values=D;
 V=F(X(:),Y(:));
 
end