Ceres 求解器:如何定义 bounds/constraints?

Ceres Solver: how to define bounds/constraints?

Ceres 求解器到处都说它可以

[...] solve robustified bounds constrained non-linear least squares problems

并且它支持参数块的上限和下限约束(例如在 http://ceres-solver.org/modeling_faqs.html 中它声明 Ceres Solver only supports upper and lower bounds constraints on the parameter blocks),但不知何故我在文档中找不到任何地方我可以设置这些上限和下限。

那么,如何在 ceres 求解器中设置参数块的上限和下限?

具体来说,我该如何在 AutoDiffCostFunction 中做到这一点?如果我使用 if 语句来 return 一个非常大的残差越界,那么这个函数是不可微分的。

例如,这是ceres Hello World:

struct CostFunctor {
   template <typename T>
   bool operator()(const T* const x, T* residual) const {
     residual[0] = 10.0 - x[0];
     return true;
   }
};

int main(int argc, char** argv) {
  google::InitGoogleLogging(argv[0]);

  // The variable to solve for with its initial value.
  double initial_x = 5.0;
  double x = initial_x;

  // Build the problem.
  Problem problem;

  // Set up the only cost function (also known as residual). This uses
  // auto-differentiation to obtain the derivative (jacobian).
  CostFunction* cost_function =
      new AutoDiffCostFunction<CostFunctor, 1, 1>(new CostFunctor);
  problem.AddResidualBlock(cost_function, nullptr, &x);

  // Run the solver!
  Solver::Options options;
  options.linear_solver_type = ceres::DENSE_QR;
  options.minimizer_progress_to_stdout = true;
  Solver::Summary summary;
  Solve(options, &problem, &summary);

  std::cout << summary.BriefReport() << "\n";
  std::cout << "x : " << initial_x
            << " -> " << x << "\n";
  return 0;
}

在此示例中,我如何将 2.0 的下限和 20.0 的上限强加给参数 x?

您可以使用此处定义的方法 setParameterLowerBoundsetParameterUpperBoundhttp://ceres-solver.org/nnls_modeling.html?highlight=setparameterlowerbound#_CPPv4N5ceres7Problem22SetParameterLowerBoundEPdid

在你的情况下,我猜是这样的:

problem.SetParameterLowerBound(&x, 0, lower_bound);
problem.SetParameterUpperBound(&x, 0, upper_bound);