No. Each array element output would be an "objective", but particleswarm() is strictly single objective.
You would have to return some "measure of goodness" instead.
The optimizer would tell you what parameter pair is optimal, which you could then use to calculate the matrix.