Coding a GA using constraint functions but getting array size error. - SOLVED

Solved - turns out I'm a bit silly and didn't re-evaluate cost after changing positions. With a few other changes the code works as intended.

Trying to use a Genetic Algorithm to minimise the cost of a water treatment system, I'm not super experienced with MATLAB but thought I'd give it a go. I'm having problems making the array sized correctly - I've attached this part of the code (the full thing requires other functions to work) so if anyone can give me a heads up on what I'm missing I'd appreciate it.

function out = RunGA(problem, params)

% Problem

CostFunction = problem.CostFunction;

nVar = problem.nVar;

VarSize = [1, nVar];

VarMin = problem.VarMin;

VarMax = problem.VarMax;

ConsGP = problem.VarGP;

ConsTSS = problem.VarTSS;

ConsTN = problem.VarTN;

ConsTP = problem.VarTP;

% Params

MaxIt = params.MaxIt;

nPop = params.nPop;

beta = params.beta;

pC = params.pC;

nC = round(pC*nPop/2)*2;

gamma = params.gamma;

mu = params.mu;

sigma = params.sigma;

% Template for Empty Individuals

empty_individual.Position = [];

empty_individual.Cost = [];

empty_individual.treatmentGP = [];

empty_individual.treatmentTSS = [];

empty_individual.treatmentTN = [];

empty_individual.treatmentTP = [];

% Best Solution Ever Found

bestsol.Cost = inf;

% Initialization

pop = repmat(empty_individual, nPop, 1);

for i = 1:nPop

% Generate Random Solution

pop(i).Position = unifrnd(VarMin, VarMax, VarSize);

% Evaluate Solution

pop(i).Cost = CostFunction(pop(i).Position);

% Compare Solution to Best Solution Ever Found

if pop(i).Cost < bestsol.Cost

bestsol = pop(i);

end

end

% Best Cost of Iterations

bestcost = nan(MaxIt, 1);

%======================================================================

% Main Loop

for it = 1:MaxIt

% Selection Probabilities

c = [pop.Cost];

avgc = mean(c);

if avgc ~= 0

c = c/avgc;

end

probs = exp(-beta*c);

% Initialize Offsprings Population

popc = repmat(empty_individual, nC/2, 2);

% Crossover

for k = 1:nC/2

% Select Parents

p1 = pop(RouletteWheelSelection(probs(1,1:nPop)));

p2 = pop(RouletteWheelSelection(probs(1,1:nPop)));

% Perform Crossover

[popc(k, 1).Position, popc(k, 2).Position] = ...

UniformCrossover(p1.Position, p2.Position, gamma);

end

% Convert popc to Single-Column Matrix

popc = popc(:);

% disp(popc);% Position,% Cost,treatmentGP,treatmentTSS,treatmentTN,treatmentTP

% Mutation

for l = 1:nC

% Perform Mutation

popc(l).Position = Mutate(popc(l).Position, mu, sigma);

% Check for Variable Bounds

popc(l).Position = max(popc(l).Position, VarMin);

popc(l).Position = min(popc(l).Position, VarMax);

% Evaluation

popc(2).Cost = CostFunction(popc(l).Position);

% Compare Solution to Best Solution Ever Found

if popc(2).Cost < bestsol.Cost

bestsol = popc(2);

end

end

% Merge and Sort Populations

pop = SortPopulation(popc);

% Remove Extra Individuals

pop = pop(1:nPop);

%=======================================================================

% Constrain Solution

%Evaluate treatment of GP with current Solution

popc(3).treatmentGP = ConsGP(popc(3));

if popc(3).treatmentGP < 0.9

%pop(i)= nPop;

popc(1).Position= nPop;

end

%Evaluate treatment of TSS with current Solution

popc(4).treatmentTSS = ConsTSS(popc(4));

if popc(4).treatmentTSS < 0.8

%pop(i)= nPop;

popc(1).Position= nPop;

end

%Evaluate treatment of TN with current Solution

popc(5).treatmentTN = ConsTN(popc(5));

if popc(5).treatmentTN < 0.45

%pop(i)= nPop;

popc(1).Position= nPop;

end

%Evaluate treatment of TP with current Solution

popc(6).treatmentTP = ConsTP(popc(6));

if popc(6).treatmentTP < 0.65

%pop(i)= nPop;

popc(1).Position= nPop;

end

%==============================================================

% Update Best Cost of Iteration

%bestcost(it) = bestsol.Cost;

bestcost(it) = min(bestsol.Cost);

% Display Iteration Information

disp(['Iteration ' num2str(it) ': Best Cost = ' num2str(bestcost(it))]);

end

% Results

out.pop = pop;

out.bestsol = bestsol;

out.bestcost = bestcost;

end

The error code is as follows:

Index exceeds the number of array elements (1).

Error in RunGA (line 119)

pop = pop(1:nPop);

Error in app1 (line 29)

out = RunGA(problem, params);

Answer 1

[Alan Weiss]

I think that you can investigate this problem yourself by using the debugger to set a break point at or just before line 119, and then see why MATLAB complains that there are not nPop array elements.

Alan Weiss

MATLAB mathematical toolbox documentation

[Lachlan Gibbons]

Thanks Alan, I actually discovered that by accident not long after posting the question, and as you predicted it helped me diagnose the issue. I've managed to get it to run and be constrained correctly now after a lot of trial and error. Amongst a few other minor changes, I changed the position change instruction for the pollutant constraints to set it to the max variable size instead of just nPop and then told it to re-evaluate the cost, which was the main error I made above (the position was changing but not the cost).

Thanks for your answer.