I need to find a way to compute the matrix inverse of some kind of distributed data type in Spark. The data is purely numerical and any way to perform this operation in Rowmatri