0Ph-dZddlZddlZddlZddlmZddlmZm Z Gdde Z dd Z dd Z ddZddZdS)a Our own implementation of the Newton algorithm Unlike the scipy.optimize version, this version of the Newton conjugate gradient solver uses only one function call to retrieve the func value, the gradient value and a callable for the Hessian matvec product. If the function call is very expensive (e.g. for logistic regression with large design matrix), this approach gives very significant speedups. N)ConvergenceWarning)line_search_wolfe1line_search_wolfe2ceZdZdS)_LineSearchErrorN)__name__ __module__ __qualname__V/var/www/html/test/jupyter/venv/lib/python3.11/site-packages/sklearn/utils/optimize.pyr r sDrr c X|dk} dtjtj|jjz} | r0t dt d| t dt |||||||fi|} | r!| ddnd } t d | zd z| d|d t} |||zg| R}tj || z}||z }tj ||k}| r*t d tj |d|d||rtj |d}|||zg| R}tj |d}||k}| rt d|d|d||rd| ddz| ddz|||f} | dG| rt dt|||||||fi|} | r!| ddnd } t d| zd z| dt| S)a Same as line_search_wolfe1, but fall back to line_search_wolfe2 if suitable step length is not found, and raise an exception if a suitable step length is not found. Raises ------ _LineSearchError If no suitable step size is found. rz Line Searchz eps=16 * finfo.eps=z try line search wolfe1rNznot z wolfe1 line search was successfulargsz2 check loss |improvement| <= eps * |loss_old|:  <=  r)ordz1 check sum(|gradient|) < sum(|gradient_old|): z < ?z' last resort: try line search wolfe2z wolfe2 line search was )npfinfoasarraydtypeepsprintrgettupleabsscipylinalgnormrr )ffprimexkpkgfkold_fval old_old_fvalverbosekwargs is_verboserret_not_rfval tiny_lossloss_improvementchecksum_abs_grad_oldgrad sum_abs_grads r_line_search_wolfe12r8s AJ rx 8,,2337 7C, o ---... *+++ QBX| V Vv V VCDa&.b +e3lBCCC 1v~zz&%''**qb 4   F8c>** (?'((I5   FF+,,FF2;FF>CFF     $|00!0<< 6"r')D)))D <,,Tq,99L #33E C#CC(8CC;@CC FQJFQJ  1v~  = ; < < < vr2sHl  >D    H!!fnFF"E /%7,F G G G 1v~   Jrc Hdtjtjjz}tjt ||j}tj|}| }d} tj||} | } |dk} | |krRtj tj ||kr?| r>DKAJ w,, 6"&**   $ $ R1RR46F26"::4F4FRRLORR  WU^^veR    ) ) ) )k) ) ) ) ) ) P1PP;>PP0;PP37PP=@+W  )()))       > <<<<=== q5Ls'EE('E(c|dkrt|jdkrI|j}d||j|}||d|zz }tj|t d|t |j|}n|j}nt|S)aCheck the OptimizeResult for successful convergence Parameters ---------- solver : str Solver name. Currently only `lbfgs` is supported. result : OptimizeResult Result of the scipy.optimize.minimize function. max_iter : int, default=None Expected maximum number of iterations. extra_warning_msg : str, default=None Extra warning message. Returns ------- n_iter : int Number of iterations. lbfgsrz{} failed to converge (status={}): {}. Increase the number of iterations (max_iter) or scale the data as shown in: https://scikit-learn.org/stable/modules/preprocessing.htmlN r) stacklevel) statusmessageformatrYrZrrWnitNotImplementedError)solverresultmax_iterextra_warning_msgresult_message warning_msgn_iter_is r_check_optimize_resultr{Ks. =A  #^N% fVV]N;; !,t&777 M+'9a H H H H  6:x00HHzHH!! Or)r)r rPrQrRTTr)NN)__doc__rYnumpyrr" exceptionsrfixesrr RuntimeErrorr r8rOrjr{r rrrs  " ++++++99999999     |   =>OOOOdVVVV|    ~~~~B......r