Save values in a 2d matrix corresponding to coordinates

I have a set of values that correspond to certain coordinates on a grid. How can I create a matrix that stores these values in the corresponding location, with blank values for all other points on the matrix?

I have three sets of data x,y and values

simplified

x = [1, 5, 6, 7] 
y = [1, 4, 6, 4] 
values = [5, 3, 4, 6]

where value of 5 is located at point (1,1)

I have earlier asked this question and gotten the

accumarray ([x(:),y(:)],values(:))

which works fine with the simplified values. The problem is that my x and y coordinates are decimals, and I wish to keep them this way. Is there a way around this problem in accumarray or some other example of how this can be done?

Answer 1

[Stephen23]

You need a way to convert decimal values to a list of indices. One easy way to do this is to use unique, or even better to use uniquetol (more recent versions only):

tol = 0.1; % pick appropriate tolerance
values = [5, 3, 4, 6];
x = [1.1, 5.2, 6.3, 7.4];
y = [1.1, 4.2, 6.3, 4.2];
[xv,~,xp] = unique(round(x/tol));
[yv,~,yp] = unique(round(y/tol));
accumarray([xp(:),yp(:)],values(:))

displays this in the command window:

ans =
     5     0     0
     0     3     0
     0     0     4
     0     6     0
>> xv*tol % rows correspond to these values:
xv =
    1.1000    5.2000    6.3000    7.4000
>> yv*tol % columns correspond to these values:
yv =
    1.1000    4.2000    6.3000

[Elise]

Thanks!

I may specify my problem a bit more; I wish to make my the final answer in a finer grid, say 0.25x0.25.

[Stephen23]

Can you give some sample values please. It is not clear what you require, as matrices do not have non-integer indices *. If you have specific requirements, then these need to be clearly explained. It seems that you want the final matrix to be a "grid" that includes columns and rows for values that do not occur in the data. If this is the case, then maybe you want something like this:

tol = 0.5; % pick appropriate tolerance
values = [5, 3, 4, 6];
x = [1, 2.5, 2.5, 3];
y = [1, 1.5, 2.5, 1];
xp = round(x/tol);
yp = round(y/tol);
accumarray([xp(:)-min(xp)+1,yp(:)-min(yp)+1], values(:))

Which displays this in the command window:

ans =
   5     0     0     0
   0     0     0     0
   0     0     0     0
   0     3     0     4
   6     0     0     0
>>  tol*(min(xp):max(xp)) % row values
ans =
  1.0000    1.5000    2.0000    2.5000    3.0000
>> tol*(min(yp):max(yp)) % column values
ans =
  1.0000    1.5000    2.0000    2.5000

Of course it is easy to adjust the grid, but you have to define it first!

* not in standard mathematics anyway

[Elise]

Thank you! This was great help!