Puzzler: Anyone can solve this parallel processing problem

I have 800 sensors and have sampled them 1000 times, so I have a 800 x 1000 matrix (A) of (random) numbers between 0 and 1. If I calculate the correlation between sensors I'd have a 800 x 800 matrix (C). Then I have a target correlation matrix CT. To simulate the target condition I need to find one or more instances of "AT" or the sensor readings that would have the target correlation values. I don't have an analytical solution for this so I am brute forcing! for each of the values in A I start moving it by small steps so that the correlation matrix would move toward the target and I repeat this until I get to the target. This works fine and I am happy with the results. But it is very very processor intensive. I have access to a 1500 core cluster but I see no way to efficiently parallelize this procedure. Any ideas?