Optimization module

File containing optimization solvers.

pyrotor.optimization.compute_optimized_coefficients(format_model, phi, lin_const, sigma_inverse, c_weight, kappa, quadratic_model, extra_info, use_quadratic_programming=True)

Depending on quadratic_model and use_quadratic_programming, decide which solver to use.

Inputs:

• format_model: list of arrays or sklearn model

  Model of the cost; if list, the first element of the list is the integrated linear part W and the second one is the integrated quadratic part q

• phi: ndarray

  Matrix representing the initial and final linear conditions

• lin_const: scipy.optimize.LinearConstraint

  Object containing the initial and final conditions

• sigma_inverse: ndarray

  Pseudoinverse of the covariance matrix of the reference coefficients

• c_weight: ndarray

  Coefficients of a weighted trajectory

• kappa: float

  Optimisation hyperparameter

• quadratic_model: bool

  Indicate if the model is quadratic

• extra_info: dict

  Contains independent_variable, basis, basis_features and basis_dimension dictionaries

• use_quadratic_programming: Boolean

  Use or not quadratic programming solver

Return:

• c_opt: list of arrays

  Coefficients of the optimized trajectory

pyrotor.optimization.minimize_cvx(Q, W, phi, lin_const, sigma_inverse, c_weight, kappa)

Minimize a quadratic function using CVXOPT library.

The associated matrix must be positive semidefinite.

Inputs:

Q: ndarray

Matrix from the quadratic term of the quadratic model

W: ndarray

Vector from the linear term of the quadratic model

phi: ndarray

Matrix representing the initial and final linear conditions

lin_const: scipy.optimize.LinearConstraint

Object containing the initial and final conditions

sigma_inverse: ndarray

Pseudoinverse of the covariance matrix of the reference coefficients

c_weight: ndarray

Coefficients of a weighted trajectory

kappa: float

Optimisation hyperparameter

Output:

c_optimized: ndarray

Coefficients of the optimized trajectory

pyrotor.optimization.minimize_trust_constr(model, lin_const, sigma_inverse, c_weight, kappa, quadratic_model, extra_info)

Minimize a function using scipy.optimize.

Inputs:

• model: list of arrays or sklearn model

  Model of the cost; if list, the first element of the list is the integrated linear part W and the second one is the integrated quadratic part q

• lin_const: scipy.optimize.LinearConstraint

  Object containing the initial and final conditions

• sigma_inverse: ndarray

  Pseudoinverse of the covariance matrix of the reference coefficients

• c_weight: ndarray

  Coefficients of a weighted trajectory

• kappa: float

  Optimisation hyperparameter

• quadratic_model: bool

  Indicate if the model is quadratic

• extra_info: dict

  Contains independent_variable, basis, basis_features and basis_dimension dictionaries

Output:

c_optimized: ndarray

Coefficients of the optimized trajectory