We give versions of Shannon's coding theorem where the decoder runs in constant time:

- Let $D=(D_1,D_2,\ldots,D_n)$ be a product distribution where the $D_i$ have constant support and have dyadic probability masses (i.e., of the form $a/2^b$ where $a,b$ are integers). Then $D$ can be sampled in constant time in ... more >>>