Pyrotor class

class pyrotor.pyrotor.Pyrotor(model, reference_trajectories, endpoints, constraints, basis, basis_features, independent_variable, n_best_trajectory_to_use=10, opti_factor=2, sigma=None, derivative=False, quadratic_model=False, use_quadratic_programming=True, n_jobs=None, verbose=True)

Main interface of PyRotor, an optimization package based on data.

Create a new PyPotor class.

Inputs:

• model: str or list or sklearn model

  if str then it is the path to the folder containing the pickle model; else if list, the first element of the list is the constant c, the second one is the linear part w and the third one is the quadratic part q Ex: (8, np.array([2, 1]), np.array([[1, 0], [2, 3]]))

• reference_trajectories: list of DataFrame

  List of reference trajectories

• endpoints: dict

  Initial and final states that the optimized trajectory must satisfy ex: {‘Var 1’: {‘start’: 109, ‘end’: 98, ‘delta’: 10},

  …}

• constraints: list

  Constraints the trajectory must complain with. Each constraint is model as a function; if the function is negative when applied on the right variable, then the constraint is considered as not satisfied. ex: [f1, f2] and if f1(trajectory) < 0 then the constraint is not satisfied

• basis: string

  Name of the functional basis

• basis_features: dict

  Contain information on the basis for each state

• independent_variable: dict

  Describe the time-interval on which are defined the trajectories ex: {‘start’: 0, ‘end’: 1, ‘frequency’:.1}

• n_best_trajectory_to_use: int, default=10

  Number of trajectories to keep

• opti_factor: float, default=2

  Optimisation factor

• sigma: ndarray, default=None

  Matrix interpreted as an estimated covariance matrix

• derivative: boolean, default=False

  Compute the derivative or not

• quadratic_model: bool

  Indicate if the model is quadratic

• use_quadratic_programming: boolean, default=True

  Use or not quadratic programming solver

• n_jobs: int, default=None

  Number of process to use - If None, sequential

• verbose: boolean, default=True

  Display the verbose or not

compute_gains()

Compute gains/savings with respect to reference trajectories.

compute_one_iteration()

Compute an optimised trajectory which must satisfy the constraints.

compute_optimal_trajectory()

Compute the optimized trajectory.

compute_relative_gains()

Compute relative gains/savings with respect to reference trajectories.