I have been studying the SMA* algorithm and I am having trouble understanding the backup operation. Specifically, I don't understand why the f value of a child node should be the maximum of its own f value and its parent's f value, as described in various pseudocode implementations such as the one found on Wikipedia (https://en.wikipedia.org/wiki/SMA*).
Could someone please explain the reasoning behind this? Also, when exactly is this information used?
Thank you.
