Update

PlatEMO/Algorithms/Multi-objective optimization/BL-SAEA/BLSAEA.m

@@ -200,7 +200,7 @@ function main(Algorithm,Problem)
                 Archive_xu=[Archive_xu;Offspring(Num',:)]; Archive_xl=[Archive_xl;ND_off_llPop(1:non_n,:)]; Archive_F=[Archive_F;ND_off_ul_obj(1:non_n,:)]; Archive_f=[Archive_f;ND_off_llFunctionValue(1:non_n,:)];
 
                 %Update archive
-                A=roundn(Archive_xu, -9);
+                A=round(Archive_xu, -9);
                 [~,ia,~] = unique(A,'rows');
 
                 Archive_xu = Archive_xu(ia,:);

PlatEMO/Algorithms/Multi-objective optimization/CMOES/CMOES.m

@@ -0,0 +1,107 @@
+classdef CMOES < ALGORITHM
+% <multi> <real/integer/label/binary/permutation> <constrained>
+% Constrained multi-objective optimization based on even search
+% type --- 1 --- Type of operator (1. GA 2. DE)
+
+%------------------------------- Reference --------------------------------
+% F. Ming, W. Gong, and Y. Jin, Even search in a promising region for
+% constrained multi-objective optimization, IEEE/CAA Journal of Automatica
+% Sinica, 2024, 11(2): 474-486.
+%------------------------------- Copyright --------------------------------
+% Copyright (c) 2024 BIMK Group. You are free to use the PlatEMO for
+% research purposes. All publications which use this platform or any code
+% in the platform should acknowledge the use of "PlatEMO" and reference "Ye
+% Tian, Ran Cheng, Xingyi Zhang, and Yaochu Jin, PlatEMO: A MATLAB platform
+% for evolutionary multi-objective optimization [educational forum], IEEE
+% Computational Intelligence Magazine, 2017, 12(4): 73-87".
+%--------------------------------------------------------------------------
+
+% This function is written by Fei Ming (email: 20151000334@cug.edu.cn)
+
+    methods
+        function main(Algorithm,Problem)
+            %% Parameter setting
+            type = Algorithm.ParameterSet(1);
+
+            %% Generate random population
+            Population1 = Problem.Initialization();
+            Population2 = Problem.Initialization();
+            [~,FrontNo,CrowdDis] = EnvironmentalSelection_NSGA2(Population1,Problem.N);
+            Fitness2    = CalFitness(Population2.objs);
+            Population3 = Population1;
+            stage_changed = 0;
+            Population = [Population1,Population2];
+            CV = sum(max(0,Population.cons),2);
+            max_cv = max(CV);
+
+            %% Optimization
+            while Algorithm.NotTerminated(Population1)
+                gen        = ceil(Problem.FE/Problem.N);
+                CV1 = sum(max(0,Population1.cons),2);
+                num_fea = sum(CV1==0);
+                if num_fea <= 0 || gen <= 0.2 * ceil(Problem.maxFE/Problem.N)
+                    MatingPool = TournamentSelection(2,2*Problem.N,FrontNo,-CrowdDis);
+                    if type == 1
+                        Offspring  = OperatorGA(Problem,Population1(MatingPool));
+                    else
+                        Offspring  = OperatorDE(Problem,Population1,Population1(MatingPool(1:end/2)),Population1(MatingPool(end/2+1:end)));
+                    end
+                    MatingPool2 = TournamentSelection(2,2*Problem.N,Fitness2);
+                    if type == 1
+                        Offspring2  = OperatorGAhalf(Problem,Population2(MatingPool2));
+                    else
+                        Offspring2  = OperatorDE(Problem,Population2,Population2(MatingPool2(1:end/2)),Population2(MatingPool2(end/2+1:end)));
+                    end
+                    [Population1,FrontNo,CrowdDis] = EnvironmentalSelection_NSGA2([Population1,Offspring,Offspring2],Problem.N);
+                    [Population2,Fitness2] = EnvironmentalSelection([Population2,Offspring,Offspring2],Problem.N,false);
+                    Population3 = Population2;
+                    Population = [Population1,Population2];
+                    CV = sum(max(0,Population.cons),2);
+                    max_cv = max(max_cv,max(CV));
+                else
+                    tau = gen/ceil(Problem.maxFE/Problem.N);
+                    Population = [Population1,Population2,Population3];
+                    CV = sum(max(0,Population.cons),2);
+                    max_cv = max(CV);
+                    if stage_changed == 0
+                        [~,Fitness2,~,Fitness3] = EvenSearch(Population,Population1(CV1==0),Problem.N,tau,max_cv);
+                        [~,Fitness1] = EnvironmentalSelection(Population1,Problem.N,true);
+                        stage_changed = 1;
+                    end
+
+                    if ~isempty(Population2)
+                        MatingPool2 = TournamentSelection(2,2*length(Population2),Fitness2);
+                        if type == 1
+                            Offspring2  = OperatorGAhalf(Problem,Population2(MatingPool2));
+                        else
+                            Offspring2  = OperatorDE(Problem,Population2,Population2(MatingPool2(1:end/2)),Population2(MatingPool2(end/2+1:end)));
+                        end
+                    else
+                        Offspring2 = [];
+                    end
+                    if ~isempty(Population3)
+                        MatingPool3 = TournamentSelection(2,2*length(Population3),Fitness3);
+                        if type == 1
+                            Offspring3  = OperatorGAhalf(Problem,Population3(MatingPool3));
+                        else
+                            Offspring3  = OperatorDE(Problem,Population3,Population3(MatingPool3(1:end/2)),Population3(MatingPool3(end/2+1:end)));
+                        end
+                    else
+                        Offspring3 = [];
+                    end
+                    MatingPool1 = TournamentSelection(2,2*Problem.N,Fitness1);
+                    if type == 1
+                        Offspring1  = OperatorGAhalf(Problem,Population1(MatingPool1));
+                    else
+                        Offspring1  = OperatorDE(Problem,Population1,Population1(MatingPool1(1:end/2)),Population1(MatingPool1(end/2+1:end)));
+                    end
+
+                    Offspring = [Offspring1,Offspring2,Offspring3];
+
+                    [Population1,Fitness1] = EnvironmentalSelection([Population,Offspring],Problem.N,true);
+                    [Population2,Fitness2,Population3,Fitness3] = EvenSearch([Population,Offspring],Population1(CV1==0),Problem.N,tau,max_cv);
+                end
+            end
+        end
+    end
+end

PlatEMO/Algorithms/Multi-objective optimization/TriP/archive.m → PlatEMO/Algorithms/Multi-objective optimization/CMOES/EnvironmentalSelection.m

@@ -1,5 +1,5 @@
-function Population = archive(Population,N)
-% Select feasible and non-dominated solutions by using SPEA2-CDP
+function [Population,Fitness] = EnvironmentalSelection(Population,N,isOrigin)
+% The environmental selection of SPEA2
 
 %------------------------------- Copyright --------------------------------
 % Copyright (c) 2024 BIMK Group. You are free to use the PlatEMO for
@@ -10,19 +10,29 @@
 % Computational Intelligence Magazine, 2017, 12(4): 73-87".
 %--------------------------------------------------------------------------
 
-    %% Select feasible solutions
-    fIndex     = all(Population.cons <= 0,2);
-    Population = Population(fIndex);
-    if isempty(Population)
-        return
-    elseif size(Population,2) > N
+    %% Calculate the fitness of each solution
+    if isOrigin
         Fitness = CalFitness(Population.objs,Population.cons);
-        Next = Fitness < 1;
+    else
+        Fitness = CalFitness(Population.objs);
+    end
+
+    %% Environmental selection
+    Next = Fitness < 1;
+    if sum(Next) < N
+        [~,Rank] = sort(Fitness);
+        Next(Rank(1:N)) = true;
+    elseif sum(Next) > N
         Del  = Truncation(Population(Next).objs,sum(Next)-N);
         Temp = find(Next);
         Next(Temp(Del)) = false;
-        Population = Population(Next);
     end
+    % Population for next generation
+    Population = Population(Next);
+    Fitness    = Fitness(Next);
+    % Sort the population
+    [Fitness,rank] = sort(Fitness);
+    Population = Population(rank);
 end
 
 function Del = Truncation(PopObj,K)

PlatEMO/Algorithms/Multi-objective optimization/CMOES/EnvironmentalSelection_NSGA2.m

@@ -0,0 +1,29 @@
+function [Population,FrontNo,CrowdDis] = EnvironmentalSelection_NSGA2(Population,N)
+% The environmental selection of NSGA-II
+
+%------------------------------- Copyright --------------------------------
+% Copyright (c) 2024 BIMK Group. You are free to use the PlatEMO for
+% research purposes. All publications which use this platform or any code
+% in the platform should acknowledge the use of "PlatEMO" and reference "Ye
+% Tian, Ran Cheng, Xingyi Zhang, and Yaochu Jin, PlatEMO: A MATLAB platform
+% for evolutionary multi-objective optimization [educational forum], IEEE
+% Computational Intelligence Magazine, 2017, 12(4): 73-87".
+%--------------------------------------------------------------------------
+
+    %% Non-dominated sorting
+    [FrontNo,MaxFNo] = NDSort(Population.objs,Population.cons,N);
+    Next = FrontNo < MaxFNo;
+
+    %% Calculate the crowding distance of each solution
+    CrowdDis = CrowdingDistance(Population.objs,FrontNo);
+
+    %% Select the solutions in the last front based on their crowding distances
+    Last     = find(FrontNo==MaxFNo);
+    [~,Rank] = sort(CrowdDis(Last),'descend');
+    Next(Last(Rank(1:N-sum(Next)))) = true;
+
+    %% Population for next generation
+    Population = Population(Next);
+    FrontNo    = FrontNo(Next);
+    CrowdDis   = CrowdDis(Next);
+end

PlatEMO/Algorithms/Multi-objective optimization/CMOES/EvenSearch.m

@@ -0,0 +1,61 @@
+function [Population2,Fitness,Population3,CV] = EvenSearch(Population,non_dom,N,tau,max_cv)
+% Even search for helper population
+
+%------------------------------- Copyright --------------------------------
+% Copyright (c) 2024 BIMK Group. You are free to use the PlatEMO for
+% research purposes. All publications which use this platform or any code
+% in the platform should acknowledge the use of "PlatEMO" and reference "Ye
+% Tian, Ran Cheng, Xingyi Zhang, and Yaochu Jin, PlatEMO: A MATLAB platform
+% for evolutionary multi-objective optimization [educational forum], IEEE
+% Computational Intelligence Magazine, 2017, 12(4): 73-87".
+%--------------------------------------------------------------------------
+
+% This function is written by Fei Ming (email: 20151000334@cug.edu.cn)
+
+    %% Determine if in promising region (1 is in)
+    lables = GetLable(Population.objs,non_dom.objs);
+    Next1 = lables == 1;
+    First_Population = Population(Next1);
+
+    % first choose non-dominated solutions near UPF
+    Fitness = CalFitness(First_Population.objs);
+    Next2 = Fitness < 1;
+    if sum(Next2) > N
+        Del  = Truncation(First_Population(Next2).objs,sum(Next2)-N);
+        Temp = find(Next2);
+        Next2(Temp(Del)) = false;
+    end
+    % store non-dominated solutions in former N
+    Population2 = First_Population(Next2);
+    Fitness = Fitness(Next2);
+
+    % then choose CV better solutions near or in feasible regions
+    % first limit the position in the objective space
+    Second_Population = First_Population;
+    CV = sum(max(0,Second_Population.cons),2);
+    epsilon = (max_cv)*(1-tau)^2;
+    Next3 = CV <= epsilon;
+    if sum(Next3) > N
+        Del  = Truncation(Second_Population(Next3).objs,sum(Next3)-N);
+        Temp = find(Next3);
+        Next3(Temp(Del)) = false;
+    end
+    % store more feasible solutions in former N
+    Population3 = Second_Population(Next3);
+    CV = CV(Next3);
+end
+
+function Del = Truncation(PopObj,K)
+% Select part of the solutions by truncation
+
+    %% Truncation
+    Distance = pdist2(PopObj,PopObj);
+    Distance(logical(eye(length(Distance)))) = inf;
+    Del = false(1,size(PopObj,1));
+    while sum(Del) < K
+        Remain   = find(~Del);
+        Temp     = sort(Distance(Remain,Remain),2);
+        [~,Rank] = sortrows(Temp);
+        Del(Remain(Rank(1))) = true;
+    end
+end

PlatEMO/Algorithms/Multi-objective optimization/CMOES/GetLable.m

@@ -0,0 +1,30 @@
+function lable = GetLable(Solution,non_dom)
+% Get the label of each solution, 1 represents is
+
+%------------------------------- Copyright --------------------------------
+% Copyright (c) 2024 BIMK Group. You are free to use the PlatEMO for
+% research purposes. All publications which use this platform or any code
+% in the platform should acknowledge the use of "PlatEMO" and reference "Ye
+% Tian, Ran Cheng, Xingyi Zhang, and Yaochu Jin, PlatEMO: A MATLAB platform
+% for evolutionary multi-objective optimization [educational forum], IEEE
+% Computational Intelligence Magazine, 2017, 12(4): 73-87".
+%--------------------------------------------------------------------------
+
+% This function is written by Fei Ming (email: 20151000334@cug.edu.cn)
+
+    N = size(Solution,1);
+    M = size(non_dom,1);
+    Lables = zeros(N,M);
+
+    %% Detect the dominance relation between each solution in Data and fns
+    for i = 1 : N
+        for j = 1 : M
+            k = any(Solution(i,:)<non_dom(j,:)) - any(Solution(i,:)>non_dom(j,:));
+            if k == 1 || k == 0
+                Lables(i,j) = 1;
+            end
+        end
+    end
+    lable = zeros(1,N);
+    lable(sum(Lables,2)==M) = 1;
+end

PlatEMO/Algorithms/Multi-objective optimization/CMaDPPs/CMaDPPs.m

@@ -0,0 +1,106 @@
+classdef CMaDPPs < ALGORITHM
+% <multi/many> <real/integer/label/binary/permutation> <constrained/none>
+% Constrained many-objective optimization with determinantal point processes
+
+%------------------------------- Reference --------------------------------
+% F. Ming, W. Gong, S. Li, L. Wang, and Z. Liao, Handling constrained
+% many-objective optimization problems via determinantal point processes,
+% Information Sciences, 2023, 643: 119260.
+%------------------------------- Copyright --------------------------------
+% Copyright (c) 2024 BIMK Group. You are free to use the PlatEMO for
+% research purposes. All publications which use this platform or any code
+% in the platform should acknowledge the use of "PlatEMO" and reference "Ye
+% Tian, Ran Cheng, Xingyi Zhang, and Yaochu Jin, PlatEMO: A MATLAB platform
+% for evolutionary multi-objective optimization [educational forum], IEEE
+% Computational Intelligence Magazine, 2017, 12(4): 73-87".
+%--------------------------------------------------------------------------
+
+% This function is written by Fei Ming (email: 20151000334@cug.edu.cn)
+
+    methods
+        function main(Algorithm,Problem)
+
+            %% Parameter setting
+            theta = Algorithm.ParameterSet(0);
+
+            %% Generate random population
+            Population = Problem.Initialization();
+            CSA = Population;
+            CV = sum(max(0,Population.cons),2);
+            [z,znad] = deal(min(Population.objs),max(Population.objs));
+
+            % parameters for epsilon
+            CVmax = max(CV);
+            G = ceil(Problem.maxFE/Problem.N);
+            Tc = 0.8 * G;
+            cp = 2;
+            alpha = 0.95;
+            tao = 0.05;
+            epsilon = inf;
+
+            % parameters for change
+            last_gen         = 20;
+            change_threshold = 1e-1;
+            max_change       = 1;
+            ideal_points     = zeros(ceil(Problem.maxFE/Problem.N),Problem.M);
+            nadir_points     = zeros(ceil(Problem.maxFE/Problem.N),Problem.M);
+            [Population,Archive] = EnvironmentalSelection(Population,[],[],Problem.N,z,znad,theta,CSA.objs,epsilon);       
+
+            %% Optimization
+            while Algorithm.NotTerminated(Population)
+                gen        = ceil(Problem.FE/Problem.N);
+                CV = sum(max(0,Population.cons),2);
+                CV_max = max(CV);
+                CVmax = max([CV_max,CVmax]);
+                rf         = sum(CV <= 1e-6) / Problem.N;
+                ideal_points(gen,:) = z;
+                nadir_points(gen,:) = znad;
+
+                % The maximumrate of change of ideal and nadir points rk is calculated.
+                if gen >= last_gen
+                    max_change = calc_maxchange(ideal_points,nadir_points,gen,last_gen);
+                end
+
+                % update epsilon before selection
+                if max_change > change_threshold && gen < 0.4 * G
+                    epsilon = inf;
+                else
+                    epsilon =  update_epsilon(tao,epsilon,CVmax,rf,alpha,gen,Tc,cp,Problem.M);
+                end
+
+                ParentC = MatingSelection([Population,Archive],CSA,Problem.N,z,znad);
+                Offspring = OperatorGA(Problem,ParentC);
+
+                z = min(z,min(Offspring.objs,[],1));
+                CSA = UpdateCSA(CSA,Offspring,Problem.N,epsilon);
+
+                [Population,Archive] = EnvironmentalSelection(Population,Offspring,Archive,Problem.N,z,znad,theta,CSA.objs,epsilon);
+                znad = max(znad,max(Population.objs,[],1));
+                if Problem.FE >= Problem.maxFE
+                    Population = Archive;
+                end
+            end
+        end
+    end
+end
+
+% Calculate the Maximum Rate of Change
+function max_change = calc_maxchange(ideal_points,nadir_points,gen,last_gen)
+    delta_value = 1e-6 * ones(1,size(ideal_points,2));
+    rz = abs((ideal_points(gen,:) - ideal_points(gen - last_gen + 1,:)) ./ max(ideal_points(gen - last_gen + 1,:),delta_value));
+    nrz = abs((nadir_points(gen,:) - nadir_points(gen - last_gen + 1,:)) ./ max(nadir_points(gen - last_gen + 1,:),delta_value));
+    max_change = max([rz, nrz]);
+end
+
+% Update the value of epsilon
+function epsilon = update_epsilon(tao,epsilon_k,epsilon_0,rf,alpha,gen,Tc,cp,M)
+    if gen > Tc
+        epsilon = 0;
+    else
+        if rf < alpha
+            epsilon = (1 - tao)^cp * epsilon_k;
+        else
+            epsilon = cp ^ M * epsilon_0 * ((1 - (gen / Tc)) ^ cp);
+        end
+    end
+end