Question

I have a function that has 11 input parameters.

MyFunction(40, 40, 1, 1, 1, 5, 0, 1, 0, 1500, 'MyFile');

The input parameter 'MyFile' when passed through the MyFunction saves a text file using the save command that is 6 columns by the 10th input parameter of rows (e.g. 1500). I usually then load this files back into MATLAB when I am ready to analyze different runs.

I’d like to run MyFunction m times and ultimately have the 'MyFile' be a measure of central tendency (e.g. mean or median) of those m runs.

m=10
for i = 1:m;
    MyFunction(40, 40, 1, 1, 1, 5, 0, 1, 0, 1500, 'MyFile');
end;

I could use the for-loop to generate a new 'MyFile' name for each iteration (e.g. MyFile1, MyFile2,…,MyFileM) with something like MyFile = sprintf('MyFile%m'); and then load all of the MyFiles back into MATLAB and then take their average and save it as a UltimateMyFile, but this seems cumbersome. Is their a better method to average these output files more directly? Should I store the files as an object, use dlmwrite, or -append?

Thanks.

Answer 1

since you are trying to find median, you need access to all the data.

you can define a 3 dimension array say

data = zeros(1500,6,m);

and then at each step of for loop update it:

data(:,:,i) = MyFunction(40, 40, 1, 1, 1, 5, 0, 1, 0, 1500);

of course you will need to redefine your function to get the right output.

However if you need to access the data at some other time, then you are better of writing it to a file and reading it from there.

in case you are only interested in the average, you can keep a running total as each case is analyzed and then then just divide it by number of cases (m).