Minimal refactor of #439 for merging RHMC implementation into main #484
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
Comment on lines
+17
to
+19
| - `premetric`: a function which, for a given posterior position `pos`, computes either | ||
| a) a symmetric, **positive definite** matrix acting as the position dependent Riemannian metric (if `metric_map = IdentityMap()`), or | ||
| b) a symmetric, **not necessarily positive definite** matrix acting as the position dependent Riemannian metric after being passed through the `metric_map` argument, which will have to ensure that its return value *is* positive definite (like `metric_map = SoftAbsMap(alpha)`), |
Contributor
There was a problem hiding this comment.
[JuliaFormatter] reported by reviewdog 🐶
Suggested change
| - `premetric`: a function which, for a given posterior position `pos`, computes either | |
| a) a symmetric, **positive definite** matrix acting as the position dependent Riemannian metric (if `metric_map = IdentityMap()`), or | |
| b) a symmetric, **not necessarily positive definite** matrix acting as the position dependent Riemannian metric after being passed through the `metric_map` argument, which will have to ensure that its return value *is* positive definite (like `metric_map = SoftAbsMap(alpha)`), | |
| - `premetric`: a function which, for a given posterior position `pos`, computes either | |
| a) a symmetric, **positive definite** matrix acting as the position dependent Riemannian metric (if `metric_map = IdentityMap()`), or | |
| b) a symmetric, **not necessarily positive definite** matrix acting as the position dependent Riemannian metric after being passed through the `metric_map` argument, which will have to ensure that its return value *is* positive definite (like `metric_map = SoftAbsMap(alpha)`), |
|
|
||
| ϵ = fwd ? step_size(lf) : -step_size(lf) | ||
| ϵ = ϵ' | ||
|
|
Contributor
There was a problem hiding this comment.
[JuliaFormatter] reported by reviewdog 🐶
Suggested change
| h::Hamiltonian{<:DenseRiemannianMetric{T,<:IdentityMap},<:GaussianKinetic}, | ||
| θ::AbstractVector{T}, | ||
| r::AbstractVector{T}; | ||
| cache=nothing |
Contributor
There was a problem hiding this comment.
[JuliaFormatter] reported by reviewdog 🐶
Suggested change
| cache=nothing | |
| cache=nothing, |
| r::AbstractVector{T}; | ||
| cache=nothing | ||
| ) where {T} | ||
| cache = @something cache begin |
Contributor
There was a problem hiding this comment.
[JuliaFormatter] reported by reviewdog 🐶
Suggested change
| cache = @something cache begin | |
| cache = @something cache begin |
Comment on lines
+47
to
+48
| rv1 = map(eachindex(log_density_gradient)) do i | ||
| -log_density_gradient[i] + .5 * tr_product(inv_metric, metric_sensitivities[:, :, i]) |
Contributor
There was a problem hiding this comment.
[JuliaFormatter] reported by reviewdog 🐶
Suggested change
| rv1 = map(eachindex(log_density_gradient)) do i | |
| -log_density_gradient[i] + .5 * tr_product(inv_metric, metric_sensitivities[:, :, i]) | |
| rv1 = map(eachindex(log_density_gradient)) do i | |
| -log_density_gradient[i] + | |
| 0.5 * tr_product(inv_metric, metric_sensitivities[:, :, i]) |
Comment on lines
+57
to
+60
| cache.rv1 .- Base.broadcasted(eachindex(cache.rv1)) do i | ||
| .5 * tr_product(cache.metric_sensitivities[:, :, i], inv_metric_r) | ||
| end | ||
| #! Eq (18) of Girolami & Calderhead (2011) | ||
| (; value, gradient) = ∂H∂θ(h, θ_full, r_half) | ||
| r_full = r_half - ϵ / 2 * gradient | ||
| # println("r_full: ", r_full) | ||
| # Tempering | ||
| #r = temper(lf, r, (i=i, is_half=false), n_steps) | ||
| # Create a new phase point by caching the logdensity and gradient | ||
| z = phasepoint(h, θ_full, r_full; ℓπ=DualValue(value, gradient)) | ||
| # Update result | ||
| if FullTraj | ||
| res[i] = z | ||
| ), cache |
Contributor
There was a problem hiding this comment.
[JuliaFormatter] reported by reviewdog 🐶
Suggested change
| cache.rv1 .- Base.broadcasted(eachindex(cache.rv1)) do i | |
| .5 * tr_product(cache.metric_sensitivities[:, :, i], inv_metric_r) | |
| end | |
| #! Eq (18) of Girolami & Calderhead (2011) | |
| (; value, gradient) = ∂H∂θ(h, θ_full, r_half) | |
| r_full = r_half - ϵ / 2 * gradient | |
| # println("r_full: ", r_full) | |
| # Tempering | |
| #r = temper(lf, r, (i=i, is_half=false), n_steps) | |
| # Create a new phase point by caching the logdensity and gradient | |
| z = phasepoint(h, θ_full, r_full; ℓπ=DualValue(value, gradient)) | |
| # Update result | |
| if FullTraj | |
| res[i] = z | |
| ), cache | |
| cache.rv1 .- Base.broadcasted(eachindex(cache.rv1)) do i | |
| 0.5 * tr_product(cache.metric_sensitivities[:, :, i], inv_metric_r) | |
| end, | |
| ), | |
| cache |
| r::AbstractVector{T}; | ||
| cache=nothing, | ||
| ) where {T} | ||
| cache = @something cache begin |
Contributor
There was a problem hiding this comment.
[JuliaFormatter] reported by reviewdog 🐶
Suggested change
| cache = @something cache begin | |
| cache = @something cache begin |
Comment on lines
+96
to
+97
| rv1 = map(eachindex(log_density_gradient)) do i | ||
| -log_density_gradient[i] + .5 * tr_product(tmpm, premetric_sensitivities[:, :, i]) |
Contributor
There was a problem hiding this comment.
[JuliaFormatter] reported by reviewdog 🐶
Suggested change
| rv1 = map(eachindex(log_density_gradient)) do i | |
| -log_density_gradient[i] + .5 * tr_product(tmpm, premetric_sensitivities[:, :, i]) | |
| rv1 = map(eachindex(log_density_gradient)) do i | |
| -log_density_gradient[i] + 0.5 * tr_product(tmpm, premetric_sensitivities[:, :, i]) |
| rv1 = map(eachindex(log_density_gradient)) do i | ||
| -log_density_gradient[i] + .5 * tr_product(tmpm, premetric_sensitivities[:, :, i]) | ||
| end | ||
| (;log_density, Q, softabsλ, tmpv, tmpm, rv1) |
Contributor
There was a problem hiding this comment.
[JuliaFormatter] reported by reviewdog 🐶
Suggested change
| (;log_density, Q, softabsλ, tmpv, tmpm, rv1) | |
| (; log_density, Q, softabsλ, tmpv, tmpm, rv1) |
Comment on lines
+106
to
+109
| cache.rv1 .- Base.broadcasted(eachindex(cache.rv1)) do i | ||
| .5 * tr_product(cache.tmpm, cache.premetric_sensitivities[:, :, i]) | ||
| end | ||
| ), cache |
Contributor
There was a problem hiding this comment.
[JuliaFormatter] reported by reviewdog 🐶
Suggested change
| cache.rv1 .- Base.broadcasted(eachindex(cache.rv1)) do i | |
| .5 * tr_product(cache.tmpm, cache.premetric_sensitivities[:, :, i]) | |
| end | |
| ), cache | |
| cache.rv1 .- Base.broadcasted(eachindex(cache.rv1)) do i | |
| 0.5 * tr_product(cache.tmpm, cache.premetric_sensitivities[:, :, i]) | |
| end, | |
| ), | |
| cache |
Contributor
|
AdvancedHMC.jl documentation for PR #484 is available at: |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Thinks were to messy for my tastes to merge as is or with small changes. Supersedes #439.