Inverse problem: 2D convolution

I have this exercise. I need to find a matrix G based on d = Gm with a point spread function and an image. The task says I can find the kth column of g. But not sure whether to set up a for loop function?

My attempt is here, but here is G = D, which I don't think it should be:

%Loading data

Data = load('wavedata.txt'); %observed data
P = load('pointspread.txt'); %point spreading

%Creating a random subset of the data, 50x50 = 2500 elements

New_data = Data(randi(numel(Data),50,50));

%Make a convolution
%Image M 

M = zeros(50);

one = [1];

for k = 1 : length(one)
    index = randi(numel(M),1);
    M(index) = one(k);
end

%2D convolution

D = conv(M,P,'same');

g = D(:);

G = reshape(g,50,50); %inverse of g to the G matrix

The exercise description is uploaded as a png here: