Fortran Convert Integer To Double Precision