Surface plots inside loop

Hi,

I am working on a code that approximates a general bivariate function using a piecewise linear bivariate function. I am able to retrieve the data and plot it, my code works as follows:

It plots one subdomain at a time, as a surface plot. The plot is put inside a loop and so during each, iteration a new subdomain is plotted.

The issue here is that the final plot is displayed with discontinuities as shown in the figure attached.

Kindly help me remove those discontinuities, so as to make the final plot solid or suggest any alternate idea that would work.

[darova]

Attach your script please

Answer 1

[darova]

Just collected all data and used griddata()

[Mahesh M S]

The plots look exactly as I want it to be. Can you please explain the procedure?

[darova]

SOrry, attached wrong script. Try now

[Mahesh M S]

Thank you for the code. It works fine for the above case.

But, the dummy code I provided was a bit specific. I tried to run the code with necessary modifications on a more general case and I am stuck with an error. Kindly help me rectify it.

The code is given below:

xdata = xlsread('debug2.xlsx','B1:AO7');
ydata = xlsread('debug2.xlsx','B9:AO15');

%The whole domain is a square with x_min = 0.2, y_min = 0.2, x_max = pi/2 &
%y_max = pi/2
%The whole domain is divided into 7 parts (sample code) and F{i}
%corresponds to the function values of each subdomain
%Here, F{1} is the subdomain bounded by X{1} and Y{1}
%      F{2} is the subdomain bounded by X{2} and Y{2}
%      F{3} is the subdomain bounded by X{3} and Y{3}
%      F{4} is the subdomain bounded by X{4} and Y{4}
%      F{5} is the subdomain bounded by X{5} and Y{5}
%      F{6} is the subdomain bounded by X{6} and Y{6}
%      F{7} is the subdomain bounded by X{7} and Y{7}

X{1} = xdata(1,1:20); X{2} = xdata(2,1:20); X{3} = xdata(3,1:20); X{4} = xdata(4,:);
X{5} = xdata(5,:); X{6} = xdata(6,:); X{7} = xdata(7,:);
Y{1} = ydata(1,1:20); Y{2} = ydata(2,1:20); Y{3} = ydata(3,1:20); Y{4} = ydata(4,:);
Y{5} = ydata(5,:); Y{6} = ydata(6,:); Y{7} = ydata(7,:);
F{1} = xlsread('debug2.xlsx','B18:U37');
F{2} = xlsread('debug2.xlsx','B39:U58');
F{3} = xlsread('debug2.xlsx','B60:U79');
F{4} = xlsread('debug2.xlsx','B81:AO120');
F{5} = xlsread('debug2.xlsx','B122:AO161');
F{6} = xlsread('debug2.xlsx','B163:AO202');
F{7} = xlsread('debug2.xlsx','B204:AO243');
l = 7;

%The requirement is to plot the total domain in a single plot

% Note: The size of X{i} and Y{i} can vary and also value of l can vary.

X = [];
Y = [];
Z = [];
	% make all data in one array
for i = 1:7
    [X0, Y0] = meshgrid(xdata(i,:),ydata(i,:));
    X = [X; X0(:)];
    Y = [Y; Y0(:)];
    Z = [Z; F{i}(:)];
end
    % create new mesh
x1 = linspace(min(xdata(:)), max(xdata(:)),40);
y1 = linspace(min(ydata(:)), max(ydata(:)),40);
[X1, Y1] = meshgrid(x1,y1);
    % create set of data
Z1 = griddata(X,Y,Z,X1,Y1);
h = surf(X1,Y1,Z1);
set(h,'LineStyle','none');

I am getting the following error:

[Mahesh M S]

Sorry. Modified the code myself. Thank you for your help.

Answer 2

[KSSV]

YOu can follow a samll demo code given here:

[X,Y,Z] = peaks(100) ; 
surf(X,Y,Z)
%% Make surface discontinuous for demo 
idx = meshgrid(10:10:100) ; 
for i = 1:idx
    Z(idx(:,i),:) = NaN ; 
    Z(:,idx(:,i)) = NaN ; 
end
surf(X,Y,Z)
%% Have a continuous (x,y,z) data 
idx = ~isnan(Z) ; 
c = [X(idx) Y(idx) Z(idx)] ; 
%% Make c continuous 
x = c(:,1) ; y = c(:,2) ; z = c(:,3) ; 
xi = linspace(min(x),max(x),100) ; 
yi = linspace(min(y),max(y),100) ; 
[X,Y] = meshgrid(xi,yi) ; 
Z = NaN(size(X)) ; 
% get nearest neighbors of new X, Y from (x,y) 
idx = knnsearch([X(:) Y(:)],[x y]) ; 
Z(idx) = z ; 
%% fill gaps by interpolation 
F = scatteredInterpolant(x,y,z) ; 
idx = isnan(Z) ; 
Z(idx) = F(X(idx),Y(idx)) ; 
figure
surf(X,Y,Z)

YOu can also use griddata, fillgaps, fillmissing.

[Mahesh M S]

Thanks for the code. But, my code doesnt have gaps as such (as depicted by the top view of the figure shown below) [The figure has been updated. sorry for the error].

Its just discontinuous.

Aso, I can only retrieve data of one subdomain at a time (loop) and thus interpolation is also difficult. A possible way is to extract all data separately from each subdomain and plot later but it takes a lot of time which I am not happy with.

Kinldy help me out with this.

[darova]

i can extract it for you

[Mahesh M S]

I can provide a dummy code.

Answer 3

[Mahesh M S]

xdata = xlsread('debug.xlsx','B1:U4');
ydata = xlsread('debug.xlsx','B6:U9');

%The whole domain is a square with x_min = 0.2, y_min = 0.2, x_max = pi/2 & y_max = pi/2
%The whole domain is divided into 4 parts (sample code) and F{i}
%corresponds to the function values of each subdomain
%Here, F{1} is the subdomain bounded by X{1} and Y{1}
%      F{2} is the subdomain bounded by X{2} and Y{2}
%      F{3} is the subdomain bounded by X{3} and Y{3}
%      F{4} is the subdomain bounded by X{4} and Y{4}

X{1} = xdata(1,:); X{2} = xdata(2,:); X{3} = xdata(3,:); X{4} = xdata(4,:);
Y{1} = ydata(1,:); Y{2} = ydata(2,:); Y{3} = ydata(3,:); Y{4} = ydata(4,:);
F{1} = xlsread('debug.xlsx','B11:U30');
F{2} = xlsread('debug.xlsx','B32:U51');
F{3} = xlsread('debug.xlsx','B53:U72');
F{4} = xlsread('debug.xlsx','B74:U93');
l = 4;

%The requirement is to plot the total domain in a single plot

% Note: The size of X{i} and Y{i} can vary and also value of l can vary.

The reference data has been updated in excel sheet attached herewith.

Answer 4

[Bruno Luong]

Assuming your xdata and ydata are sorted. My code doest not make new resampling of x and y, or interpolation, I just stitch them together

% Read your data
xdata = xlsread('debug.xlsx','B1:U4');
ydata = xlsread('debug.xlsx','B6:U9');

X{1} = xdata(1,:); X{2} = xdata(2,:); X{3} = xdata(3,:); X{4} = xdata(4,:);
Y{1} = ydata(1,:); Y{2} = ydata(2,:); Y{3} = ydata(3,:); Y{4} = ydata(4,:);
F{1} = xlsread('debug.xlsx','B11:U30');
F{2} = xlsread('debug.xlsx','B32:U51');
F{3} = xlsread('debug.xlsx','B53:U72');
F{4} = xlsread('debug.xlsx','B74:U93');

% Stitching the sub-rectangule data
Xu = unique([X{:}]);
Yu = unique([Y{:}]);
Fu = nan(length(Yu),length(Yu));
for k=1:length(F)
    [~,ix] = ismember(X{k},Xu);
    [~,iy] = ismember(Y{k},Yu);
    Fu(iy,ix) = F{k};
end

figure; 
surf(Xu,Yu,Fu)

[Mahesh M S]

The code does work for this specific dummy data I have given, but it has some minor errors when we go to a more general level.

I can provide a more general code for your reference.

[Bruno Luong]

It's up to you. I can only work on what is available since you haven't described precisely how the data looks like.

And please DO NOT attach Excel file this is proprietary format.

[Mahesh M S]

Sorry for the excel file. How am I supposed to provide the data?

With the data only, the problem is clearly defined.

Also I will provide the general set of data.

[Bruno Luong]

MATFILE, we are all working with MATLAB here.

[Mahesh M S]

The required X, Y and F data ar attached herewith for your reference.

[Bruno Luong]

load('datareqd.mat')
% Stitching the sub-rectangule data
Xu = unique([X{:}]);
Yu = unique([Y{:}]);
Fu = nan(length(Yu),length(Yu));
for k=1:length(F)
    [~,ix] = ismember(X{k},Xu);
    [~,iy] = ismember(Y{k},Yu);
    Fu(iy,ix) = F{k};
end

Fu = fillmissing(Fu,'linear');

close all
surf(Xu,Yu,Fu)