How to formulate block sparsity?

Here’s a useful exercise: suppose you were given a numeric vector r, and you needed some code to compute that norm on that numeric vector. How would you do it? It’s quite likely that is exactly how you would do it in CVX.