[BUG] mhaco asserts if initial population does not verify constraints

[jschueller]

if the initial population does not verify constraints (through unconstrain) mhaco asserts:

IndexError: 
function: assert_minimisation
where: /tmp/pagmo2-2.18.0/src/utils/hv_algos/hv_algorithm.cpp, 256
what: Reference point is invalid: another point seems to be outside the reference point boundary, or be equal to it:

kind of similar to #493

[jschueller]

@Sceki do you mind looking at this one too ? its a bit similar, not on the bound constraints, but on the general eq/ineq constraints

[Sceki]

@jschueller can you please post a minimal code to reproduce this error?

[jschueller]

here:

#include <pagmo/algorithms/maco.hpp>
#include <pagmo/population.hpp>
#include <pagmo/problems/rosenbrock.hpp>
#include <pagmo/problems/unconstrain.hpp>

using namespace pagmo;

struct problem1 {
    vector_double fitness(const vector_double &x) const
    {
        const double g = 1.0 + 9.0 * (x[0] + x[1]);
        return {x[0], g * (1.0 - std::sqrt(x[0] / g)), x[1]-x[0]}; // zdt1 + constraint x0>x1
    }
    std::pair<vector_double, vector_double> get_bounds() const
    {
        return {{0.0, 0.0}, {1.0, 1.0}};
    }
    vector_double::size_type get_nobj() const
    {
        return 2;
    }
    
    vector_double::size_type get_nic() const
    {
        return 1;
    }
};

int main()
{
  try {
    problem prob{problem1{}};

    prob = unconstrain(prob);
    
    population pop1{prob, 0, 0};
    int size = 80;
    std::uniform_real_distribution<double> unif(0.0, 1.0);
    std::default_random_engine re;
    for (int i=0; i<size; ++i)
    {
      vector_double x(2);
      x[0] = unif(re);
      do {
        x[1] = unif(re);
      } while (false && x[0] <= x[1]);// ok if replace false by true
      
      vector_double y = prob.fitness(x);
      pop1.push_back(x);
    }

    maco user_algo1{10u};
    pop1 = user_algo1.evolve(pop1);

	}
	catch(const std::exception & exc) {
		std::cout << "ex="<<exc.what() <<std::endl;
		return 2;
	}
	return 0;
}

[jschueller]

@Sceki were you able to reproduce using my code ?