I have a few suggestions that should help you improve and optimize your code.   I am concerned that the matrix Njxx, which you say should hold the final results, is being set to a zero matrix within the third nested loop. By continually overwriting Njxx, are you not throwing away most of the results, and only collecting information from the very last iteration of the inner most loop, when age_x, age_j and t are all equal to &MaxAge?   The ELSE block near the end of the code is attempting to 'grow' the matrix Njxx by appending a row at the bottom for every iteration. This is very inefficient as it creates a new matrix in a different area of memory every time. Because of the unnecessary copying syntax like this should be avoided. Ideally you want to create Njxx at its final size at the start of the program (full of missing values) and then write each row of results using something like: Njxx[ i, 1:ncol(DiagOffsetSum) ] = DiagOffsetSum; where 'i' is the next available row to hold results.   I think the calculation in the innermost loop can be improved as there is no need to create the maxtrix DiagOffset with all the different lags. Something simpler based on the CUSUM function like DiagOffsetSum = sum( Djxx_diag ) - (0 || t( cusum( Djxx_diag ))); will suffice and give the sums that you want. Note that there will be a redundant zero at the end of the vector DiagOffsetSum, but it is easy to leave this out when you copy the results to Njxx. ... View more

You can use essentially the same syntax to construct Q like this: lab = choose(k<k0,'put','mix'); lab = choose(k>k0,'call',lab); Q = choose(k<k0, p, (p+c)/2); Q = choose(k>k0, c, Q); print k0 k lab p c Q; ... View more

This error message arises when you try to take the square root of negative numbers, and it clear that this has happened 1437 times.  To see where the problem is occuring, I suggest you split the one line of code into 3 or 4 separate lines and print the values at each step for a small data set. The problem may be the data itself, or it may be that your large compound expression is not evaluated in the way that you expect because of operator precedence, so try adding more parentheses to make the evaluation order unambiguous. ... View more

PRINCOMP leaves the input data set umodified, and adds the principal component scores to the OUT= data set.  So you should be using syntax something like this: proc princomp data=myData out=PCOUT <other options>; var ... run; proc corr data=PCOUT <options>; var ... with ... run; ... View more

I think the problem is a simple typo.  The line : L1=length(vis); uses the number '1' in the variable name instead of the letter 'I'.  Then the error that you have shown occurs at the line, wsi=substr(w(|position,1|),1,LI); as the 3rd agrument to the substr() function does not exist. ... View more

I think the issue with PRINCOMP is that the correct syntax is TYPE=CORR with two 'R's. The error message from IML is informative since sigma is not square, actually it is a row vector.  To declare a matrix you need to add commas at the end of the rows like this: sigma={ 1.00000 0.89933 0.18110 -0.17027, 0.89933 1.00000 0.07733 -0.31809, 0.18110 0.07733 1.00000 0.88061, -0.17027 -0.31809 0.88061 1.00000}; ... View more