how to solve direct variation problems