For two mixed points with difference and gradient difference , the input difference is and the output difference is . Their squared-norm difference is . This exact identity is a reusable way to prove averagedness without assuming the two block maps are individually nonexpansive.
Articles by others on the same topic
There are currently no matching articles.