Update dependencies, removal of redundant dependencies. Removal of re…

…dundant code.

.travis.yml

@@ -1,13 +1,12 @@
 language: julia
 julia:
-  - 1.2
-  - 1.3
   - 1.4
+  - 1.5
 
 jobs:
   include:
     - stage: "Documentation"
-      julia: 1.3
+      julia: 1.5
       os: linux
       script:
         - julia --project=docs/ -e 'using Pkg; Pkg.develop(PackageSpec(path=pwd()));

Project.toml

@@ -1,10 +1,9 @@
 name = "DPMMSubClusters"
 uuid = "2841fd70-8698-11e9-176d-6dfa142d2ee7"
 authors = ["Or Dinari <[email protected]>"]
-version = "0.1.8"
+version = "0.1.9"
 
 [deps]
-CatViews = "81a5f4ea-a946-549a-aa7e-2a7f63a27d31"
 Clustering = "aaaa29a8-35af-508c-8bc3-b662a17a0fe5"
 Distributed = "8ba89e20-285c-5b6f-9357-94700520ee1b"
 DistributedArrays = "aaf54ef3-cdf8-58ed-94cc-d582ad619b94"
@@ -17,15 +16,14 @@ SpecialFunctions = "276daf66-3868-5448-9aa4-cd146d93841b"
 StatsBase = "2913bbd2-ae8a-5f71-8c99-4fb6c76f3a91"
 
 [compat]
-CatViews = "1"
-Clustering = "0.13.3"
-DistributedArrays = "0, 1"
-Distributions = "0, 1"
-JLD2 = "0, 1"
-NPZ = "0, 1"
-SpecialFunctions = "0.8, 0.9, 0.10, 1"
-StatsBase = "0,1"
-julia = "1"
+Clustering = "^0.13.3,0.14"
+DistributedArrays = "0.1,0.2,0.3,0.4,0.5,0.6,0.7,0.8"
+Distributions = "0.16 - 0.25"
+JLD2 = "0.1, 0.2,0.3,0.4"
+NPZ = "0.1, 0.2,0.3,0.4,0.5,0.6"
+SpecialFunctions = "0.8, 0.9, 0.10, 1.0"
+StatsBase = "0.32,0.33,0.34,0.35"
+julia = "1.0,1.1,1.2,1.3,1.4,1.5,1.6"
 
 [extras]
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"

README.md

@@ -6,7 +6,7 @@
 # DPMMSubClusters.jl
 This is the code repository for the *Julia* package (with an optional [Python wrapper](https://github.com/BGU-CS-VIL/dpmmpython)) that corresponds to our paper, [Distributed MCMC Inference in Dirichlet Process Mixture Models Using Julia](https://www.cs.bgu.ac.il/~dinari/papers/dpmm_hpml2019.pdf), which was presented at CCGrid2019 High Performance Computing Maching Learning workshop (HPML).
 
-Note that due to improvements in the code we have made since the time of the pulication of the paper, this package is now faster than what we reported there. 
+Note that due to improvements in the code we have made since the time of the pulication of the paper, this package is now faster than what we reported there.
 
 <br>
 <p align="center">
@@ -19,7 +19,6 @@ Note that due to improvements in the code we have made since the time of the pul
 ## Requirements
 This package was developed and tested on *Julia 1.0.3*, prior versions will not work.
 The following dependencies are required:
-- CatViews
 - Distributed
 - DistributedArrays
 - Distributions

src/DPMMSubClusters.jl

@@ -8,7 +8,6 @@ using DistributedArrays
 using StatsBase
 using Distributions
 using SpecialFunctions
-using CatViews
 using LinearAlgebra
 using JLD2
 using Clustering

src/local_clusters_actions.jl

@@ -67,14 +67,6 @@ function create_subclusters_labels!(labels::AbstractArray{Int64,1},
     if size(labels,1) == 0
         return
     end
-    # clusts_dist = [cluster_params.l_dist, cluster_params.r_dist]
-    # log_weights = log.(cluster_params.lr_weights)
-    # @inbounds for i=1:size(labels,1)
-    #     x = @view points[:,i]
-    #     probs = RestrictedClusterProbs(log_weights,clusts_dist,x)
-    #     labels[i] = sample(1:2, ProbabilityWeights(probs))
-    #     # println(labels[i])
-    # end
     parr = zeros(Float32,length(labels), 2)
     log_likelihood!((@view parr[:,1]),points,cluster_params.l_dist)
     log_likelihood!((@view parr[:,2]),points,cluster_params.r_dist)
@@ -98,19 +90,6 @@ function sample_labels!(labels::AbstractArray{Int64,1},
 end
 
 
-function RestrictedClusterProbs(logπs::AbstractVector{V}, clusters,
-                                    x::AbstractVector) where V<:Real
-    p = Array{V,1}(undef,length(clusters))
-    max = typemin(V)
-    for (j,c) in enumerate(values(clusters))
-        @inbounds s = p[j] = logπs[j] + logαpdf(c,x)
-        max = s>max ? s : max
-    end
-    pc = exp.(p .- max)
-    return pc ./ sum(pc)
-end
-
-
 function sample_labels_worker!(labels::AbstractArray{Int64,1},
         points::AbstractArray{Float32,2},
         final::Bool,
@@ -122,19 +101,8 @@ function sample_labels_worker!(labels::AbstractArray{Int64,1},
     parr = zeros(Float32,length(indices), length(clusters_vector))
     tic = time()
     for (k,cluster) in enumerate(clusters_vector)
-        # if no_more_splits == false && k==1
-        #      parr[:,k] .= -Inf
-        #      continue
-        # end
         log_likelihood!(reshape((@view parr[:,k]),:,1), pts,cluster.cluster_dist)
     end
-    # println("Time: "* string(time()-tic) * "   size:" *string(size(pts)))
-    # parr = zeros(Float32,length(indices), length(clusters_vector))
-    # newx = copy(localpart(points)')
-    # @time log_likelihood!(parr, localpart(points),[c.cluster_dist for c in clusters_vector],log.(clusters_weights))
-    #
-    #
-    #
     for (k,v) in enumerate(clusters_weights)
         parr[:,k] .+= log(v)
     end
@@ -144,18 +112,6 @@ function sample_labels_worker!(labels::AbstractArray{Int64,1},
     else
         sample_log_cat_array!(lbls,parr)
     end
-
-    # clust_dists = [c.cluster_dist for c in clusters_vector]
-    #
-    # @inbounds for i=1:size(lbls,1)
-    #     x = @view pts[:,i]
-    #     probs = RestrictedClusterProbs(log_weights,clust_dists,x)
-    #     # println(probs)
-    #     lbls[i] = sample(1:length(clusters_vector), ProbabilityWeights(probs))
-    #     # println(lbls[i])
-    # end
-
-
 end