Zenna Tavares

Esp for deep models

Parallel Replica exchange

1

Make replica exchange parallel, async

Change temp instead of omega in replica exchange

1

Will allow for different inference algorithms with different omega types

Implement and use (non cassette based) withtemp

2

Instead oi `withkernel`

Control scaling using moving averages

Currently we can only modify Omega, which is rich enough for most (all?) purposes. However it may also be useful to be able to make proposals on the outputs of...

The following code from Omega Models is slow for two reasons that need to be fixed. 1. We are using samplemean instead of the symbolic mean but in this case...

The WolvesVsRabbit demo has a weird issue related to memoization. It only occurs in the first execution. I should isolate it here and fix

Implement rcd of one argument. Semantically it should be `rcd(x) = rcd(x, parents(x))`

Why is this slow?

11

```julia julia> n = normal(0,1) julia> @time mean(n, 10000000) 7.962374 seconds (281.66 M allocations: 9.711 GiB, 9.67% gc time) -0.00016734630189156838 julia> @time mean(randn(10000000)) 0.091020 seconds (7 allocations: 76.294 MiB, 3.16%...