a ߙfbY@sdZddlZddlZddlZddlmZgdZdZe e e j Z dZdZGdd d ejd ZGd d d eZGd ddeZdS)z> Optimization algorithms used in `~astropy.modeling.fitting`. N)AstropyUserWarning) OptimizationSLSQPSimplexdgHz>)lJ)lJ)c@seZdZdZgZddZeddZejddZeddZ e jd dZ ed d Z e jd d Z d dZ eddZ e jddZdS)raE Base class for optimizers. Parameters ---------- opt_method : callable Implements optimization method Notes ----- The base Optimizer does not support any constraints by default; individual optimizers should explicitly set this list to the specific constraints it supports. cCs||_t|_t|_t|_dS)N) _opt_methodDEFAULT_MAXITER_maxiter DEFAULT_EPS_eps DEFAULT_ACC_acc)self opt_methodr:lib/python3.9/site-packages/astropy/modeling/optimizers.py__init__.szOptimization.__init__cCs|jS)zMaximum number of iterationsr rrrrmaxiter4szOptimization.maxitercCs ||_dS)z Set maxiterNrrvalrrrr9scCs|jS)z=Step for the forward difference approximation of the Jacobianr rrrreps>szOptimization.epscCs ||_dS)z Set eps valueNrrrrrrCscCs|jS)zRequested accuracyr rrrraccHszOptimization.acccCs ||_dS)z Set accuracyNrrrrrrMscCs|jjd}|S)Nz()) __class____name__)rZfmtrrr__repr__RszOptimization.__repr__cCs|jS)z Return the optimization method.)rrrrrrVszOptimization.opt_methodcCs tddS)Nz'Subclasses should implement this method)NotImplementedErrorrrrr__call__[szOptimization.__call__N)r __module__ __qualname____doc__supported_constraintsrpropertyrsetterrrrrabcabstractmethodr rrrrrs(       r) metaclasscs0eZdZdZgdZfddZddZZS)ra) Sequential Least Squares Programming optimization algorithm. The algorithm is described in [1]_. It supports tied and fixed parameters, as well as bounded constraints. Uses `scipy.optimize.fmin_slsqp`. References ---------- .. [1] http://www.netlib.org/toms/733 )boundseqconsineqconsfixedtiedcs,ddlm}t|ddddd|_dS)Nr) fmin_slsqp)final_func_valnumiter exit_modemessage)scipy.optimizer/superrfit_info)rr/rrrrns  zSLSQP.__init__c sJ|d|j|d<d|vr$|j|d<d|vr6|j|d<|dd}|dfdd jD}d d |D}t|}|D]4}|ddurtd|d<|d durztd |d <qztj|td }t j } t j } |j ||f|d ||| | d|\} } } }}| |j d<| |j d<||j d<||j d<|dkr@tdt| |j fS) Run the solver. Parameters ---------- objfunc : callable objection function initval : iterable initial guess for the parameter values fargs : tuple other arguments to be passed to the statistic function kwargs : dict other keyword arguments to be passed to the solver riterepsilonr verblevelNrcsg|]}t|qSr)getattr).0nameZmodelrr z"SLSQP.__call__..cSsg|]}|js|js|jqSr)r-r.r*)r=Zparrrrr@rA)ZdtypeT)args full_outputdispr*r+Zieqconsr0r1r2r3zLThe fit may be unsuccessful; check fit_info['message'] for more information.)popr r r Z param_namesnpZasarrayDEFAULT_BOUNDSfloatZarrayr+r,rr6warningswarnr)robjfuncinitvalfargskwargsrEZparsr*ir+r, fitparamsr0r1r2Zmessrr?rr xsF              zSLSQP.__call__rr!r"r#r$rr __classcell__rrr7rr`s  rcs0eZdZdZgdZfddZddZZS)ra= Neald-Mead (downhill simplex) algorithm. This algorithm [1]_ only uses function values, not derivatives. Uses `scipy.optimize.fmin`. References ---------- .. [1] Nelder, J.A. and Mead, R. (1965), "A simplex method for function minimization", The Computer Journal, 7, pp. 308-313 )r*r-r.cs,ddlm}t|ddddd|_dS)Nr)fmin)r0r1r2num_function_calls)r4rTr5rr6)rZsimplexr7rrrs  zSimplex.__init__c Ksd|vr|j|d<d|vr.|d|_|dd|vrJ|d|_|d|dd}|j||f||j|dd|\}}}} } ||jd<||jd <| |jd <| |jd <|jd d krtd t|jd dkrtdt||jfS)r8rrxtolr;NT)rCrVrErDr0r1r2rUrBzLThe fit may be unsuccessful; Maximum number of function evaluations reached.zBThe fit may be unsuccessful; Maximum number of iterations reached.)r r rFrr6rJrKr) rrLrMrNrOrErQr0r1Zfuncallsr2rrrr s<          zSimplex.__call__rRrrr7rrs  r)r#rJr'ZnumpyrGZastropy.utils.exceptionsr__all__rZsqrtZfinforIrr r rHABCMetarrrrrrrs EQ