Test on more recent Julia versions

.github/workflows/CI.yml

@@ -18,8 +18,8 @@ jobs:
       fail-fast: false
       matrix:
         version:
-          - '1.8'
-          #- 'nightly'
+          - '1'
+          - '1.10'
         os:
           - ubuntu-latest
           - macOS-latest
@@ -71,4 +71,4 @@ jobs:
             using Documenter: DocMeta, doctest
             using MultilayerGraphs
             DocMeta.setdocmeta!(MultilayerGraphs, :DocTestSetup, :(using MultilayerGraphs); recursive=true)
-            doctest(MultilayerGraphs)'
+            doctest(MultilayerGraphs)'

Project.toml

@@ -29,12 +29,12 @@ Distributions = "0.25"
 Graphs = "1"
 IterTools = "1.4"
 MetaGraphs = "0.7, 0.8"
-OMEinsum = "0.7, 0.8"
+OMEinsum = "0.7, 0.8, 0.9"
 PrettyTables = "2.2, 3"
 SimpleTraits = "0.9"
 SimpleValueGraphs = "0.4"
 SimpleWeightedGraphs = "1.2"
 Statistics = "1"
-TensorOperations = "3.2, 4"
+TensorOperations = "4"
 WhereTraits = "1"
 julia = "1.8"

src/subgraphs/layer.jl

@@ -189,15 +189,15 @@ Return a random `Layer`.
 """
 function Layer(
     name::Symbol,
-    vertices::Vector{Union{V,N}},
+    vertices::Union{Vector{MultilayerVertex{nothing}},Vector{Node}},
     ne::Int64,
     null_graph::G,
     weighttype::Type{U};
     default_vertex_metadata::Function=mv -> NamedTuple(),
     default_edge_weight::Function=(src, dst) -> nothing,
     default_edge_metadata::Function=(src, dst) -> NamedTuple(),
     allow_self_loops::Bool=false,
-) where {T<:Integer,U<:Real,G<:AbstractGraph{T},V<:MultilayerVertex{nothing},N<:Node}
+) where {T<:Integer,U<:Real,G<:AbstractGraph{T}}
     _nv = length(vertices)
     @assert(
         length(unique(vertices)) == _nv, "The argument `vertices` must be a unique list"
@@ -211,7 +211,7 @@ function Layer(
         "The number of required edges, $ne, is greater than the number of edges the provided graph supports i.e. $maxe"
     )
 
-    vertex_type = @isdefined(V) ? MultilayerVertex : Node
+    vertex_type = eltype(vertices)
 
     edge_list = NTuple{2,MultilayerVertex{nothing}}[]  #MultilayerEdge
     fadjlist = Dict{vertex_type,Vector{vertex_type}}()
@@ -250,7 +250,7 @@ function Layer(
         end
 
         # Add the edge to the edge list. Convert it to a MultilayerVertex if a list of Nodes was given as `vertices`
-        if @isdefined(V)
+        if vertex_type == MultilayerVertex{nothing}
             push!(edge_list, (rand_vertex_1, rand_vertex_2))
         else
             push!(edge_list, (MV(rand_vertex_1), MV(rand_vertex_2)))

src/tensorsfactorizations.jl

@@ -140,7 +140,7 @@ function to_matrix(A, a, b; return_tensor_shape=false)
 
     # Permute the indices of A to the right order
     perm = vcat(a, b)
-    A = tensorcopy(A, collect(1:ndims(A)), perm)
+    A = tensorcopy(collect(1:ndims(A)), A, perm)
     # The lists shp_a and shp_b list the dimensions of the bonds in a and b
     shp = size(A)
     shp_a = shp[1:length(a)]

test/Project.toml

@@ -10,3 +10,6 @@ SimpleValueGraphs = "b43c691f-cac2-5415-8122-396fe16a49fc"
 SimpleWeightedGraphs = "47aef6b3-ad0c-573a-a1e2-d07658019622"
 StatsBase = "2913bbd2-ae8a-5f71-8c99-4fb6c76f3a91"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
+
+[compat]
+Agents = "6"

test/agents_jl_integration.jl

@@ -11,23 +11,17 @@ graph = MultilayerGraph(all_layers_u, all_interlayers_u)
     new_value::Float64
 end
 # Instantiate the agent
-function EigenAgent(id, pos; initial_opinion::Float64)
-    return EigenAgent(id, pos, initial_opinion, initial_opinion, -1.0)
-end
-# Define the agent-based model
-EigenABM = ABM(EigenAgent, GraphSpace(graph))
-# Add agents to the ABM so that agent i is located in vertex i
-for (i, mv) in enumerate(mv_vertices(graph))
-    initial_opinion = 1.0
-    add_agent!(i, EigenABM; initial_opinion=initial_opinion)
-end
+# function EigenAgent(; id, pos, initial_opinion::Float64)
+#     return EigenAgent(id, pos, initial_opinion, initial_opinion, -1.0)
+# end
 # Define the individual-level dynamics (micro-dynamics)
 function agent_step!(agent::EigenAgent, model)
     agent.previous_value = agent.old_value
-    return agent.new_value = sum([
+    agent.new_value = sum([
         outneighbor_agent.old_value for
         outneighbor_agent in nearby_agents(agent, model; neighbor_type=:all)
     ])
+    return agent
 end
 # Define the system-level dynamics (macro-dynamics)
 function model_step!(model)
@@ -37,6 +31,20 @@ function model_step!(model)
         agent.old_value = agent.new_value
     end
 end
+# Define the agent-based model
+EigenABM = ABM(EigenAgent, GraphSpace(graph); agent_step!, model_step!)
+# Add agents to the ABM so that agent i is located in vertex i
+for (i, mv) in enumerate(mv_vertices(graph))
+    initial_opinion = 1.0
+    add_agent!(
+        i,
+        EigenAgent,
+        EigenABM;
+        previous_value=initial_opinion,
+        old_value=initial_opinion,
+        new_value=-1.0,
+    )
+end
 # Set the rule to stop the model simulation  
 function terminate(model, s)
     # println(s, typeof(s))
@@ -49,14 +57,12 @@ function terminate(model, s)
     end
 end
 # Simulate the model 
-agent_data, _ = run!(
-    EigenABM, agent_step!, model_step!, terminate; adata=[:new_value], when=terminate
-)
+agent_data, _ = run!(EigenABM, terminate; adata=[:new_value], when=terminate)
 # Compute the eigenvector centrality of the surrounding multilayer graph 
 eig_centr_swm, err_swm = eigenvector_centrality(
-    EigenABM.space.graph; weighted=false, tol=1e-18, norm="null"
+    abmspace(EigenABM).graph; weighted=false, tol=1e-18, norm="null"
 )
 
-for (val_1, val_2) in zip(eig_centr_swm, agent_data.new_value)
-    @test val_1 ≈ val_2
+for a in allagents(EigenABM)
+    @test a.new_value ≈ eig_centr_swm[a.id]
 end