-
```julia
julia> AdvancedHMC.DenseEuclideanMetric(Matrix{Float64}(I, 3, 3))
DiagEuclideanMetric([1.0, 1.0, 1.0])
```
-
I am trying to reproduce the prophet model using Turing, but I am receiving an error when sampling with NUTS. I am fairly new to Julia so it could definitely be a user error. Currently, NUTS seems to …
-
migrated from https://github.com/TuringLang/Turing.jl/issues/268
yebai updated
5 years ago
-
The following 3 types in `src/metric.jl`:
- https://github.com/TuringLang/AdvancedHMC.jl/blob/a0af7ccc668a2a1e537dc5acdbe8c7bd22f11a86/src/metric.jl#L16-L24
- https://github.com/TuringLang/Advan…
yebai updated
5 years ago
-
Discussions in https://github.com/TuringLang/AdvancedHMC.jl/pull/5#issuecomment-470335981
>@mohamed82008 I reverted some functions about pre-allocating memory so that only intermediate variables ar…
yebai updated
5 years ago
-
- [x] metrics
- abstraction related to NUTS
- [x] tree sampler
- [x] termination critera
- [x] full binary tree
-
Hi,
I'd like to re-run the benchmarks on my local computer (to test the change on https://github.com/TuringLang/AdvancedHMC.jl/pull/79). How do I suppose to run the benchmarks in this repo?
Tha…
-
Maybe we should make the `sample` function not return those samples by default (and support an optional keyword for it)?
-
First of all, best wishes for safety and wellbeing.
```julia> using Turing
[ Info: Precompiling Turing [fce5fe82-541a-59a6-adf8-730c64b5f9a0]
WARNING: could not import MCMCChains.AbstractChains …
-
Approximation error of leapfrog integration (i.e. accumulated Hamiltonian energy error) can sometimes explode, for example when the curvature of the current region is very high. This type of approxima…
yebai updated
5 years ago