Big things today, with plenaries given by David Donoho and Martin Vetterli.
Donoho answered all the questions I have regarding the variability of recovery algorithms on distributions underlying sparse vectors.
I just need a few years to understand them.
I also need to look more closely at approximate message passing.
And Vetterli gave a great talk, discussing the tendency in algorithm development to jump to a solution before solving the outstanding problem, e.g., sampling the real continuous world on a discretized grid.
Now I need to eat dinner, and run some experiments.