Zenna Tavares
Zenna Tavares
To implement reversible jump mcmc, we need the following things - Given a probabilistic program, that may or may not be conditioned, we need to be able to compute the...
Should be compatible with argmax zenna/Omega.jl#140
Implement Argmax (or argmin, or both) - [ ] Decide on whether this should work with random variables or onl variables - [ ] Integrate with other Julia solvers
Given some real valued random variable, we want to be able to compute the gradients. Gradients are in the package `OmegaGrad`, which is not within `OmegaCore`, so you'll need to...
Most existing Julia samplers that use `rand(rng)` For instance ```julia function X(rng) a = rand(rng) b = rand(rng) a + b end ``` This should be equivalent to something like...
lift should support keyword arguments ``` f(x; n) = [sqrt(x) for i = 1:n] lift(f)(normal(0, 1); n = 30) ```
``` function zygotetest() x = normal(0, 1, (10,)) y = sum(x) lineargradient(y, rand(defΩ()), ZygoteGrad) end ``` Breaks because of mutation in memoization.
`` julia> x = normal(0, 1) 18:Normal(0, 1)::Float64 julia> rand(Ω, x, 10) ERROR: MethodError: no method matching !(::Float64) Closest candidates are: !(::Any, ::RandVar) at /home/zenna/repos/Omega.jl/src/lift/lift.jl:10 !(::Any, ::RandVar, ::RandVar) at /home/zenna/repos/Omega.jl/src/lift/lift.jl:10...
Can we get an analytical expectation in Black Scholes model? https://github.com/zenna/OmegaModels.jl/blob/master/models/BlackScholes/src/BlackScholes.jl Value of `lastsim` is uncertain due to both `σ` and internal normals. For instance, if we were to simulate...