Fill NaN matrix with vector of unequal sizes

Hi all,

I have the following code:

temp = {'AccNorm','GyrNorm'};
NSensors = 5;
for a = 1:length(temp) 
    for k = 1:NSensors
        CMC_RT(k).RF.(temp{a}) = nan(10,2);
    end
end

In which I just preallocate CMC_T with NaN values, because the 2 vectors I want to put in the matrix have different lengths.

However when I try to fill them up:

for a = 1:length(temp) 
    for k = 1:NSensors
        CMC_RT(k).RF.GyrNorm = CMC_RTT.RF.(temp{a})(k,1:10)';
        CMC_RT(k).RF.GyrNorm(:,2) = CMC_RTT.RF.(temp{a})(k,11:19)';
    end
end

Matlab gives me the error: Unable to perform assignment because the size of the left side is 10-by-1 and the size of the right side is 9-by-1.

Which obviously makes sense, but I would think matlab would just fill the NaN matrix with the numbers and leave the NaN values for the ones where there is no substitute value. But this was apparently whishful thinking...

Any help on a idea for a solution would be much appreciated.

Thank you!

Answer 1

[Stephen23]

CMC_RT(k).RF.GyrNorm(1:10,1) = CMC_RTT.RF.(temp{a})(k,1:10);
CMC_RT(k).RF.GyrNorm(1:9,2)  = CMC_RTT.RF.(temp{a})(k,11:19);

The complex conjugates are not required.

[Inti Vanmechelen]

Very very very stupid of me to not think of doing that myself.

Thank you.

[CheshireM]

@Stephen@Inti Vanmechelen

Do you know how to get it automatically though?what if I have thousand vectors of different sizes, how should I fill out my NaN array?

[Stephen23]

@CheshireM: store the vectors in a cell array, preallocate the output array, then use a loop.

[CheshireM]

@Stephen do you mean do it by element?

Like if I have a cell array t, and prealocated array T with following parameters

[s,d] = cellfun(@size,time)
s =

1150        1150        1150        1000         853
d =

1     1     1     1     1

out = max([s,d])
out =

        1150
T=NaN(length(t),out);

I should do something like this?

for i = 1:length(t)
    for ii = 1:out
        if ~isempty(t{1,i}(ii))
            T(i,ii) = t{1,i}(ii);
        end    
    end
end

[CheshireM]

@Stephen This code just gives me an error

Index exceeds the number of array elements (1000)

[Stephen23]

"I should do something like this?"

One loop should be sufficient:

n = cellfun(@numel,t);
m = nan(numel(t),max(n));
for k = 1:numel(t)
    m(k,1:n(k)) = t{k};
end

This code assumes that cell array t and its content are vectors.

Alternatively you could download this: https://www.mathworks.com/matlabcentral/fileexchange/22909-padcat