3 ms·Recurrences converging to the wrong limit in finite precision, and some new e [pdf]2 points by fanf2 2y ago