Use faster findmin

The fast version of findmin can also be used with the N2Plain
algorithm, giving a good speedup at higher densities

src/JetReconstruction.jl

@@ -34,7 +34,7 @@ include("HepMC3.jl")
 include("PlainAlgo.jl")
 export plain_jet_reconstruct
 
-## Tiled algorithms
+## N2Tiled algorithm
 # Common pieces
 include("TiledAlgoUtils.jl")
 

src/PlainAlgo.jl

@@ -1,3 +1,5 @@
+using LoopVectorization
+
 Base.@propagate_inbounds function dist(i, j, rapidity_array, phi_array)
     drapidity = rapidity_array[i] - rapidity_array[j]
     dphi = abs(phi_array[i] - phi_array[j])
@@ -104,7 +106,6 @@ function plain_jet_reconstruct(particles::Vector{T}; p = -1, R = 1.0, recombine
 end
 
 
-
 function plain_jet_reconstruct(;objects_array::Vector{J}, kt2_array::Vector{F}, 
         phi_array::Vector{F}, rapidity_array::Vector{F}, p = -1, R = 1.0, recombine = +, ptmin = 0.0) where {J, F<:AbstractFloat}
     # Bounds
@@ -137,12 +138,7 @@ function plain_jet_reconstruct(;objects_array::Vector{J}, kt2_array::Vector{F},
     iteration::Int = 1
     while N != 0
         # findmin
-        i = 1
-        dij_min = nndij[1]
-        @inbounds @simd for k in 2:N
-            cond = nndij[k] < dij_min
-            dij_min, i = ifelse(cond, (nndij[k], k), (dij_min, i))
-        end
+        dij_min, i = fast_findmin(nndij, N)
 
         j::Int = nn[i]
 

src/TiledAlgoLL.jl

@@ -259,20 +259,6 @@ function find_tile_neighbours!(tile_union, jetA, jetB, oldB, tiling)
     n_near_tiles
 end
 
-"""Find the lowest value in the array, returning the value and the index"""
-find_lowest(dij, n) = begin
-    best = 1
-    @inbounds dij_min = dij[1]
-    @turbo for here in 2:n
-        newmin = dij[here] < dij_min
-        best = newmin ? here : best
-        dij_min = newmin ? dij[here] : dij_min
-    end
-    # @assert argmin(dij[1:n]) == best
-    dij_min, best
-end
-
-
 
 """Return all inclusive jets of a ClusterSequence with pt > ptmin"""
 function inclusive_jets(clusterseq::ClusterSequence, ptmin = 0.0)
@@ -375,7 +361,7 @@ function tiled_jet_reconstruct(particles::Vector{PseudoJet}; p = -1, R = 1.0, re
         # compact NNs and dij arrays
         ilast = N - (iteration - 1)
         # Search for the lowest value of min_dij_ijet
-        dij_min, ibest = find_lowest(dij, ilast)
+        dij_min, ibest = fast_findmin(dij, ilast)
         @inbounds jetA = NNs[ibest]
         jetB = jetA.NN
 

src/Utils.jl

@@ -127,3 +127,17 @@ function final_jets(jets::Vector{LorentzVectorCyl}, ptmin::AbstractFloat)
 	end
 	final_jets
 end
+
+## Faster findmin function
+"""Find the lowest value in the array, returning the value and the index"""
+fast_findmin(dij, n) = begin
+    # findmin(@inbounds @view dij[1:n])
+    best = 1
+    @inbounds dij_min = dij[1]
+    @turbo for here in 2:n
+        newmin = dij[here] < dij_min
+        best = newmin ? here : best
+        dij_min = newmin ? dij[here] : dij_min
+    end
+    dij_min, best
+end