o Eb~)@s~dZddlZddlZddlZddlmZmZm Z m Z ddl m Z ddl mZgZddZGd d d Z    dddZdS)zTrust-region optimization.N)_check_unknown_options_status_messageOptimizeResult_prepare_scalar_function)HessianUpdateStrategy) FD_METHODScs.dgdur dfSfdd}|fS)Nrcs*dd7<t|g|RS)Nrr)npcopy)xZ wrapper_argsargsfunctionZncalls=/usr/lib/python3/dist-packages/scipy/optimize/_trustregion.pyfunction_wrappersz(_wrap_function..function_wrapperr)rr rrr r_wrap_function s rc@sjeZdZdZdddZddZeddZed d Zed d Z d dZ eddZ ddZ ddZ dS)BaseQuadraticSubproblemaQ Base/abstract class defining the quadratic model for trust-region minimization. Child classes must implement the ``solve`` method. Values of the objective function, Jacobian and Hessian (if provided) at the current iterate ``x`` are evaluated on demand and then stored as attributes ``fun``, ``jac``, ``hess``. NcCsF||_d|_d|_d|_d|_d|_d|_||_||_||_ ||_ dSN) _x_f_g_h_g_magZ _cauchy_pointZ _newton_point_fun_jac_hess_hessp)selfr funjachesshessprrr__init__&s z BaseQuadraticSubproblem.__init__cCs*|jt|j|dt|||S)Ng?)rr dotr r"rprrr__call__3s*z BaseQuadraticSubproblem.__call__cC|jdur ||j|_|jS)z1Value of objective function at current iteration.N)rrrrrrrr6 zBaseQuadraticSubproblem.funcCr()z=Value of Jacobian of objective function at current iteration.N)rrrr)rrrr =r*zBaseQuadraticSubproblem.jaccCr()zz ||\}}Wn tjjyd}Ynyw||}||}||||||}|j|j}|j|}|dkrd}nV||}|dkr|d9}n|dkr|rtd||}|| kr |}|}| r|t||dur$|t||d7}|j| kr1d}n || kr9d}n|j| kstdtdddf} | r|dkrWt| |ntd| |td|jtd|tdjtdjtdj|dt||dk||j|jjjj|d|| |d }!|dur|j |!d<| r||!d<|!S) a Minimization of scalar function of one or more variables using a trust-region algorithm. Options for the trust-region algorithm are: initial_trust_radius : float Initial trust radius. max_trust_radius : float Never propose steps that are longer than this value. eta : float Trust region related acceptance stringency for proposed steps. gtol : float Gradient norm must be less than `gtol` before successful termination. maxiter : int Maximum number of iterations to perform. disp : bool If True, print convergence message. inexact : bool Accuracy to solve subproblems. If True requires less nonlinear iterations, but more vector products. Only effective for method trust-krylov. This function is called by the `minimize` function. It is not supposed to be called directly. Nz7Jacobian is currently required for trust-region methodsz_Either the Hessian or the Hessian-vector product is currently required for trust-region methodszBA subproblem solving strategy is required for trust-region methodsrg?zinvalid acceptance stringencyz%the max trust radius must be positivez)the initial trust radius must be positivez?the initial trust radius must be less than the max trust radius)r r!r cs||Sr)r!r$)r r&r Zsfrrr"sz%_minimize_trust_region..hesspr.g?rsuccessmaxiterz:A bad approximation caused failure to predict improvement.z3A linalg error occurred, such as a non-psd Hessian.z Warning: z# Current function value: %fz Iterations: %dz! Function evaluations: %dz! Gradient evaluations: %dz Hessian evaluations: %d) r rJstatusrr nfevZnjevnhevZnitmessager!allvecs)r ValueError Exceptionr ZasarrayZflattenrrZgradcallabler!r isinstancerrlenr-r=r,Z LinAlgErrorminappendr rprintrMZngevrNrr )"rZx0r r r!r"Z subproblemZinitial_trust_radiusZmax_trust_radiusZetaZgtolrKZdispZ return_allcallbackZinexactZunknown_optionsZnhesspZwarnflagr5r rPmkr&Z hits_boundaryZpredicted_valueZ x_proposedZ m_proposedZactual_reductionZpredicted_reductionZrhoZstatus_messagesresultrrGr_minimize_trust_regionts          9     r])rNNNNrCrDrErFNFFNT)rAr1Znumpyr Z scipy.linalgr+ _optimizerrrrZ'scipy.optimize._hessian_update_strategyrZ(scipy.optimize._differentiable_functionsr__all__rrr]rrrrs   X