Drop support for pre-1.10

.github/workflows/CI.yml

@@ -19,8 +19,8 @@ jobs:
       fail-fast: false
       matrix:
         version:
-          - '1.7'
           - '1.10'
+          - '1'
         os:
           - ubuntu-latest
           - macOS-latest

Project.toml

@@ -18,7 +18,6 @@ Optimisers = "3bd65402-5787-11e9-1adc-39752487f4e2"
 ProgressMeter = "92933f4c-e287-5a05-a399-4b506db050ca"
 Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
 Requires = "ae029012-a4dd-5104-9daa-d747884805df"
-SimpleUnPack = "ce78b400-467f-4804-87d8-8f486da07d0a"
 StatsBase = "2913bbd2-ae8a-5f71-8c99-4fb6c76f3a91"
 
 [weakdeps]
@@ -54,20 +53,15 @@ ProgressMeter = "1.6"
 Random = "1"
 Requires = "1.0"
 ReverseDiff = "1"
-SimpleUnPack = "1.1.0"
 StatsBase = "0.32, 0.33, 0.34"
 Zygote = "0.6"
 julia = "1.7"
 
 [extras]
 Bijectors = "76274a88-744f-5084-9051-94815aaf08c4"
 Enzyme = "7da242da-08ed-463a-9acd-ee780be4f1d9"
-ForwardDiff = "f6369f11-7733-5829-9624-2563aa707210"
-Mooncake = "da2b9cff-9c12-43a0-ae48-6db2b0edb7d6"
 Pkg = "44cfe95a-1eb2-52ea-b672-e2afdf69b78f"
-ReverseDiff = "37e2e3b7-166d-5795-8a7a-e32c996b4267"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
-Zygote = "e88e6eb3-aa80-5325-afca-941959d7151f"
 
 [targets]
 test = ["Pkg", "Test"]

README.md

@@ -26,7 +26,6 @@ a `LogDensityProblem` can be implemented as
 
 ```julia
 using LogDensityProblems
-using SimpleUnPack
 
 struct NormalLogNormal{MX,SX,MY,SY}
     μ_x::MX

docs/Project.toml

@@ -26,7 +26,10 @@ LogDensityProblems = "2.1.1"
 Optimisers = "0.3"
 Plots = "1"
 QuasiMonteCarlo = "0.3"
+<<<<<<< HEAD
+=======
 ReverseDiff = "1"
 SimpleUnPack = "1"
+>>>>>>> 0383ce6422892c623ac9aacc525158a73e522cab
 StatsFuns = "1"
 julia = "1.10"

docs/src/examples.md

@@ -15,7 +15,6 @@ Using the `LogDensityProblems` interface, we the model can be defined as follows
 
 ```@example elboexample
 using LogDensityProblems
-using SimpleUnPack
 
 struct NormalLogNormal{MX,SX,MY,SY}
     μ_x::MX
@@ -25,7 +24,7 @@ struct NormalLogNormal{MX,SX,MY,SY}
 end
 
 function LogDensityProblems.logdensity(model::NormalLogNormal, θ)
-    @unpack μ_x, σ_x, μ_y, Σ_y = model
+    (; μ_x, σ_x, μ_y, Σ_y) = model
     return logpdf(LogNormal(μ_x, σ_x), θ[1]) + logpdf(MvNormal(μ_y, Σ_y), θ[2:end])
 end
 
@@ -59,7 +58,7 @@ Thus, we will use [Bijectors](https://github.com/TuringLang/Bijectors.jl) to mat
 using Bijectors
 
 function Bijectors.bijector(model::NormalLogNormal)
-    @unpack μ_x, σ_x, μ_y, Σ_y = model
+    (; μ_x, σ_x, μ_y, Σ_y) = model
     return Bijectors.Stacked(
         Bijectors.bijector.([LogNormal(μ_x, σ_x), MvNormal(μ_y, Σ_y)]),
         [1:1, 2:(1 + length(μ_y))],

src/AdvancedVI.jl

@@ -1,7 +1,6 @@
 
 module AdvancedVI
 
-using SimpleUnPack: @unpack, @pack!
 using Accessors
 
 using Random

src/families/location_scale.jl

@@ -56,7 +56,7 @@ function (re::RestructureMeanField)(flat::AbstractVector)
 end
 
 function Optimisers.destructure(q::MvLocationScale{<:Diagonal,D,L,E}) where {D,L,E}
-    @unpack location, scale, dist = q
+    (; location, scale, dist) = q
     flat = vcat(location, diag(scale))
     return flat, RestructureMeanField(q)
 end
@@ -69,27 +69,27 @@ Base.size(q::MvLocationScale) = size(q.location)
 Base.eltype(::Type{<:MvLocationScale{S,D,L,E}}) where {S,D,L,E} = eltype(D)
 
 function StatsBase.entropy(q::MvLocationScale)
-    @unpack location, scale, dist = q
+    (; location, scale, dist) = q
     n_dims = length(location)
     # `convert` is necessary because `entropy` is not type stable upstream
     return n_dims * convert(eltype(location), entropy(dist)) + logdet(scale)
 end
 
 function Distributions.logpdf(q::MvLocationScale, z::AbstractVector{<:Real})
-    @unpack location, scale, dist = q
+    (; location, scale, dist) = q
     return sum(Base.Fix1(logpdf, dist), scale \ (z - location)) - logdet(scale)
 end
 
 function Distributions.rand(q::MvLocationScale)
-    @unpack location, scale, dist = q
+    (; location, scale, dist) = q
     n_dims = length(location)
     return scale * rand(dist, n_dims) + location
 end
 
 function Distributions.rand(
     rng::AbstractRNG, q::MvLocationScale{S,D,L}, num_samples::Int
 ) where {S,D,L}
-    @unpack location, scale, dist = q
+    (; location, scale, dist) = q
     n_dims = length(location)
     return scale * rand(rng, dist, n_dims, num_samples) .+ location
 end
@@ -98,7 +98,7 @@ end
 function Distributions.rand(
     rng::AbstractRNG, q::MvLocationScale{<:Diagonal,D,L}, num_samples::Int
 ) where {L,D}
-    @unpack location, scale, dist = q
+    (; location, scale, dist) = q
     n_dims = length(location)
     scale_diag = diag(scale)
     return scale_diag .* rand(rng, dist, n_dims, num_samples) .+ location
@@ -107,14 +107,14 @@ end
 function Distributions._rand!(
     rng::AbstractRNG, q::MvLocationScale, x::AbstractVecOrMat{<:Real}
 )
-    @unpack location, scale, dist = q
+    (; location, scale, dist) = q
     rand!(rng, dist, x)
     x[:] = scale * x
     return x .+= location
 end
 
 function Distributions.mean(q::MvLocationScale)
-    @unpack location, scale = q
+    (; location, scale) = q
     return location + scale * Fill(mean(q.dist), length(location))
 end
 

src/families/location_scale_low_rank.jl

@@ -52,7 +52,7 @@ Base.size(q::MvLocationScaleLowRank) = size(q.location)
 Base.eltype(::Type{<:MvLocationScaleLowRank{L,SD,SF,D,E}}) where {L,SD,SF,D,E} = eltype(L)
 
 function StatsBase.entropy(q::MvLocationScaleLowRank)
-    @unpack location, scale_diag, scale_factors, dist = q
+    (; location, scale_diag, scale_factors, dist) = q
     n_dims = length(location)
     scale_diag2 = scale_diag .* scale_diag
     UtDinvU = Hermitian(scale_factors' * (scale_factors ./ scale_diag2))
@@ -63,7 +63,7 @@ end
 function Distributions.logpdf(
     q::MvLocationScaleLowRank, z::AbstractVector{<:Real}; non_differntiable::Bool=false
 )
-    @unpack location, scale_diag, scale_factors, dist = q
+    (; location, scale_diag, scale_factors, dist) = q
     μ_base = mean(dist)
     n_dims = length(location)
 
@@ -86,7 +86,7 @@ function Distributions.logpdf(
 end
 
 function Distributions.rand(q::MvLocationScaleLowRank)
-    @unpack location, scale_diag, scale_factors, dist = q
+    (; location, scale_diag, scale_factors, dist) = q
     n_dims = length(location)
     n_factors = size(scale_factors, 2)
     u_diag = rand(dist, n_dims)
@@ -97,7 +97,7 @@ end
 function Distributions.rand(
     rng::AbstractRNG, q::MvLocationScaleLowRank{S,D,L}, num_samples::Int
 ) where {S,D,L}
-    @unpack location, scale_diag, scale_factors, dist = q
+    (; location, scale_diag, scale_factors, dist) = q
     n_dims = length(location)
     n_factors = size(scale_factors, 2)
     u_diag = rand(rng, dist, n_dims, num_samples)
@@ -108,7 +108,7 @@ end
 function Distributions._rand!(
     rng::AbstractRNG, q::MvLocationScaleLowRank, x::AbstractVecOrMat{<:Real}
 )
-    @unpack location, scale_diag, scale_factors, dist = q
+    (; location, scale_diag, scale_factors, dist) = q
 
     rand!(rng, dist, x)
     x[:] = scale_diag .* x
@@ -120,22 +120,22 @@ function Distributions._rand!(
 end
 
 function Distributions.mean(q::MvLocationScaleLowRank)
-    @unpack location, scale_diag, scale_factors = q
+    (; location, scale_diag, scale_factors) = q
     μ = mean(q.dist)
     return location +
            scale_diag .* Fill(μ, length(scale_diag)) +
            scale_factors * Fill(μ, size(scale_factors, 2))
 end
 
 function Distributions.var(q::MvLocationScaleLowRank)
-    @unpack scale_diag, scale_factors = q
+    (; scale_diag, scale_factors) = q
     σ2 = var(q.dist)
     return σ2 *
            (scale_diag .* scale_diag + sum(scale_factors .* scale_factors; dims=2)[:, 1])
 end
 
 function Distributions.cov(q::MvLocationScaleLowRank)
-    @unpack scale_diag, scale_factors = q
+    (; scale_diag, scale_factors) = q
     σ2 = var(q.dist)
     return σ2 * (Diagonal(scale_diag .* scale_diag) + scale_factors * scale_factors')
 end

src/objectives/elbo/repgradelbo.jl

@@ -86,7 +86,7 @@ function estimate_objective(obj::RepGradELBO, q, prob; n_samples::Int=obj.n_samp
 end
 
 function estimate_repgradelbo_ad_forward(params′, aux)
-    @unpack rng, obj, problem, adtype, restructure, q_stop = aux
+    (; rng, obj, problem, adtype, restructure, q_stop) = aux
     q = restructure_ad_forward(adtype, restructure, params′)
     samples, entropy = reparam_with_entropy(rng, q, q_stop, obj.n_samples, obj.entropy)
     energy = estimate_energy_with_samples(problem, samples)

src/objectives/elbo/scoregradelbo.jl

@@ -94,7 +94,7 @@ function estimate_objective(obj::ScoreGradELBO, q, prob; n_samples::Int=obj.n_sa
 end
 
 function estimate_scoregradelbo_ad_forward(params′, aux)
-    @unpack rng, obj, problem, adtype, restructure, q_stop = aux
+    (; rng, obj, problem, adtype, restructure, q_stop) = aux
     baseline = compute_control_variate_baseline(
         obj.baseline_history, obj.baseline_window_size
     )

test/Project.toml

@@ -5,17 +5,18 @@ DiffResults = "163ba53b-c6d8-5494-b064-1a9d43ac40c5"
 DifferentiationInterface = "a0c0ee7d-e4b9-4e03-894e-1c5f64a51d63"
 Distributions = "31c24e10-a181-5473-b8eb-7969acd0382f"
 DistributionsAD = "ced4e74d-a319-5a8a-b0ac-84af2272839c"
+Enzyme = "7da242da-08ed-463a-9acd-ee780be4f1d9"
 FillArrays = "1a297f60-69ca-5386-bcde-b61e274b549b"
 ForwardDiff = "f6369f11-7733-5829-9624-2563aa707210"
 Functors = "d9f16b24-f501-4c13-a1f2-28368ffc5196"
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
 LogDensityProblems = "6fdf6af0-433a-55f7-b3ed-c6c6e0b8df7c"
+Mooncake = "da2b9cff-9c12-43a0-ae48-6db2b0edb7d6"
 Optimisers = "3bd65402-5787-11e9-1adc-39752487f4e2"
 PDMats = "90014a1f-27ba-587c-ab20-58faa44d9150"
 Pkg = "44cfe95a-1eb2-52ea-b672-e2afdf69b78f"
 Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
 ReverseDiff = "37e2e3b7-166d-5795-8a7a-e32c996b4267"
-SimpleUnPack = "ce78b400-467f-4804-87d8-8f486da07d0a"
 StableRNGs = "860ef19b-820b-49d6-a774-d7a799459cd3"
 Statistics = "10745b16-79ce-11e8-11f9-7d13ad32a3b2"
 StatsBase = "2913bbd2-ae8a-5f71-8c99-4fb6c76f3a91"
@@ -30,16 +31,18 @@ DiffResults = "1"
 DifferentiationInterface = "0.6"
 Distributions = "0.25.111"
 DistributionsAD = "0.6.45"
+Enzyme = "0.13"
 FillArrays = "1.6.1"
 ForwardDiff = "0.10.36"
 Functors = "0.4.5"
 LinearAlgebra = "1"
 LogDensityProblems = "2.1.1"
+Mooncake = "0.4"
 Optimisers = "0.2.16, 0.3"
 PDMats = "0.11.7"
+Pkg = "1"
 Random = "1"
 ReverseDiff = "1.15.1"
-SimpleUnPack = "1.1.0"
 StableRNGs = "1.0.0"
 Statistics = "1"
 StatsBase = "0.34"

test/inference/repgradelbo_distributionsad.jl

@@ -3,18 +3,10 @@ AD_distributionsad = Dict(
     :ForwarDiff => AutoForwardDiff(),
     #:ReverseDiff => AutoReverseDiff(), # DistributionsAD doesn't support ReverseDiff at the moment
     :Zygote => AutoZygote(),
+    :Mooncake => AutoMooncake(; config=Mooncake.Config()),
+    :Enzyme => AutoEnzyme(),
 )
 
-if @isdefined(Mooncake)
-    AD_distributionsad[:Mooncake] = AutoMooncake(; config=Mooncake.Config())
-end
-
-if @isdefined(Enzyme)
-    AD_distributionsad[:Enzyme] = AutoEnzyme(;
-        mode=set_runtime_activity(ReverseWithPrimal), function_annotation=Const
-    )
-end
-
 @testset "inference RepGradELBO DistributionsAD" begin
     @testset "$(modelname) $(objname) $(realtype) $(adbackname)" for realtype in
                                                                      [Float64, Float32],

test/inference/repgradelbo_locationscale.jl

@@ -3,18 +3,10 @@ AD_locationscale = Dict(
     :ForwarDiff => AutoForwardDiff(),
     :ReverseDiff => AutoReverseDiff(),
     :Zygote => AutoZygote(),
+    :Mooncake => AutoMooncake(; config=Mooncake.Config()),
+    :Enzyme => AutoEnzyme(),
 )
 
-if @isdefined(Mooncake)
-    AD_locationscale[:Mooncake] = AutoMooncake(; config=Mooncake.Config())
-end
-
-if @isdefined(Enzyme)
-    AD_locationscale[:Enzyme] = AutoEnzyme(;
-        mode=set_runtime_activity(ReverseWithPrimal), function_annotation=Const
-    )
-end
-
 @testset "inference ScoreGradELBO VILocationScale" begin
     @testset "$(modelname) $(objname) $(realtype) $(adbackname)" for realtype in
                                                                      [Float64, Float32],
@@ -32,7 +24,7 @@ end
         rng = StableRNG(seed)
 
         modelstats = modelconstr(rng, realtype)
-        @unpack model, μ_true, L_true, n_dims, strong_convexity, is_meanfield = modelstats
+        (; model, μ_true, L_true, n_dims, strong_convexity, is_meanfield) = modelstats
 
         T = 1000
         η = 1e-3

test/inference/repgradelbo_locationscale_bijectors.jl

@@ -3,18 +3,10 @@ AD_locationscale_bijectors = Dict(
     :ForwarDiff => AutoForwardDiff(),
     :ReverseDiff => AutoReverseDiff(),
     :Zygote => AutoZygote(),
+    :Mooncake => AutoMooncake(; config=Mooncake.Config()),
+    :Enzyme => AutoEnzyme(),
 )
 
-if @isdefined(Mooncake)
-    AD_locationscale_bijectors[:Mooncake] = AutoMooncake(; config=Mooncake.Config())
-end
-
-if @isdefined(Enzyme)
-    AD_locationscale_bijectors[:Enzyme] = AutoEnzyme(;
-        mode=set_runtime_activity(ReverseWithPrimal), function_annotation=Const
-    )
-end
-
 @testset "inference RepGradELBO VILocationScale Bijectors" begin
     @testset "$(modelname) $(objname) $(realtype) $(adbackname)" for realtype in
                                                                      [Float64, Float32],
@@ -32,7 +24,7 @@ end
         rng = StableRNG(seed)
 
         modelstats = modelconstr(rng, realtype)
-        @unpack model, μ_true, L_true, n_dims, strong_convexity, is_meanfield = modelstats
+        (; model, μ_true, L_true, n_dims, strong_convexity, is_meanfield) = modelstats
 
         T = 1000
         η = 1e-3

test/inference/scoregradelbo_distributionsad.jl

@@ -3,16 +3,10 @@ AD_scoregradelbo_distributionsad = Dict(
     :ForwarDiff => AutoForwardDiff(),
     #:ReverseDiff => AutoReverseDiff(), # DistributionsAD doesn't support ReverseDiff at the moment
     :Zygote => AutoZygote(),
+    #:Mooncake => AutoMooncake(; config=Mooncake.Config()),
+    :Enzyme => AutoEnzyme(),
 )
 
-if @isdefined(Mooncake)
-    AD_scoregradelbo_distributionsad[:Mooncake] = AutoMooncake(; config=Mooncake.Config())
-end
-
-#if @isdefined(Enzyme)
-#    AD_scoregradelbo_distributionsad[:Enzyme] = AutoEnzyme()
-#end
-
 @testset "inference ScoreGradELBO DistributionsAD" begin
     @testset "$(modelname) $(objname) $(realtype) $(adbackname)" for realtype in
                                                                      [Float64, Float32],
@@ -29,7 +23,7 @@ end
         rng = StableRNG(seed)
 
         modelstats = modelconstr(rng, realtype)
-        @unpack model, μ_true, L_true, n_dims, strong_convexity, is_meanfield = modelstats
+        (; model, μ_true, L_true, n_dims, strong_convexity, is_meanfield) = modelstats
 
         T = 1000
         η = 1e-5