Comparison with APG-TF?

[cmcaine]

Hi,

Just wondering if you have an opinion on the Alternating Proximal Gradient algorithm proposed by Xu and Yin [paper]. It's a fast BCD algorithm but not an ANLS algorithm.

I can't find much critical discussion of it, just the maths, so I thought I'd ask an expert :)