Ugt  dZddlZddlZddlZddlZddlZddlmZddl m Z ddl m Z ddl mZmZmZmZddl mZddlmZdd lmZmZmZmZd Zejgd jZejgd jZee Z!ed dZ"ed e"ed ded ded dzed ded dzeddzedded dzeddzgZ#de#DZ$ej%&de#dZ'dZ(ej%&de$dZ)ej%&de#dZ*ej%&de#dZ+ej%&de$dZ,ej%&de$dZ-ej%&de#dZ.ej%&de#d Z/ej%&de#d!Z0d"Z1ej%&de#ej%&d#e!ej2ej3dej4$gd%Z5d&Z6d'Z7ej%&de#d(Z8d)Z9d*Z:ej%&de$d+Z;d,Zd/Z?d0Z@ej%&de#d1ZAd2ZBej%&d3d4ejCd5ieDd6feejE ejEf7d8d9eDd:fgd;ZFd<ZGd=ZHej%&d>d?d@gej%&dAgdBdCZIej%&d>d?d@gej%&dAgdBdDZJej%&dAgdEej%&dFdGdHgdIZKej%&dAgdEdJZLdKZMGdLdMeZNdNZOdS)Oz'Testing for Gaussian process regressionN) approx_fprime)ConvergenceWarning)GaussianProcessRegressor)RBF DotProductExpSineSquared WhiteKernel)ConstantKernel) MiniSeqKernel)assert_allcloseassert_almost_equalassert_array_almost_equalassert_array_lessc0|tj|zS)N)npsin)xs f/var/www/surfInsights/venv3-11/lib/python3.11/site-packages/sklearn/gaussian_process/tests/test_gpr.pyfr#s rvayy=)?g@g@g@g@g @)@g@g@g@g@rfixed length_scalelength_scale_boundsr)MbP?@@{Gz?Y@h㈵>)r#r"皙?c(g|]}|tk |S) fixed_kernel).0kernels r r*6s"LLLV|5K5KV5K5K5Krr)cVtjdkrtjdt |t t}|t d\}}t|tttj |ddS)N#This test may fail on 32 bit Pythonr)T return_cov) sysmaxsizepytestxfailrfitXypredictr rdiag)r)gpry_predy_covs rtest_gpr_interpolationr>9s {e :;;; #& 1 1 1 5 5a ; ;CKKdK33MFE""",,,,,rc td}gd}tjgd}t|||}||d\}}t ||dd d tjt|z t ||t tj |d dS) Nr)baseline_similarity_bounds)ABC)r.Tr/ eval_gradientrDr1) r rarrayrr6r9r raveleyelenr:)r)r7r8r;r<r=s r!test_gpr_interpolation_structuredrMFs g > > >FA A "& 1 1 1 5 5a ; ;CKKdK33MFEq%%%a(..001rvc!ff~~3E2L2L2N2N""",,,,,rc$tjdkrtjdt |t t}||j j ||j ksJdS)Nr,r-r.) r2r3r4r5rr6r7r8log_marginal_likelihoodkernel_thetar)r;s rtest_lml_improvingrSUs {e :;;; #& 1 1 1 5 5a ; ;C  & &s{'8 9 9C>      rc$t|tt}t j|jjjtj }| |dt|jj|ddS)Nr.dtypeF) clone_kernel) rr6r7r8ronesrPrQshapefloat64rOr )r)r; input_thetas rtest_lml_without_cloning_kernelr`jsx #& 1 1 1 5 5a ; ;C'#++1DDDK %@@@ );:::::rct|tt}||jjd\}}tjtj |dk|jj|jj dddfkz|jj|jj dddfkzsJdS)Nr.T-C6?rrD) rr6r7r8rOrPrQrallabsbounds)r)r;lml lml_gradients rtest_converged_to_local_maximumrhts #& 1 1 1 5 5a ; ;C33CK4EtLLC 6    $ ;  2111a4 8 8 : ;  2111a4 8 8 :  rct|tt}|jj}t j|jjj j }d}||t j |dddfdf<t|dddf|jj|zt|jj|dddf|zdS)Nr.绽|=rDr) rr6r7r8rPrerfinforQrYmaxisfiniter)r)r;remax_tinys rtest_solution_inside_boundsrps #& 1 1 1 5 5a ; ;C [ F 8CK%+ , , 0D D,0FBKqqq!t % % %q ()fQQQTlCK$5$<===ck'1)<=====rct|tt|jd\}}t |jfdd}t||ddS)Nr.Tc0|dS)NF)rO)rQr;s rz#test_lml_gradient..sC$?$?u$M$MrrjrF)rr6r7r8rOrQrr )r)rfrglml_gradient_approxr;s @rtest_lml_gradientrus #& 1 1 1 5 5a ; ;C33FL$GGC' MMMMu &91=====rct|}|td\}}t|ddt |jjdkrBttj|tj |jdddSttj|dddS)Nr.Tr/rrD) rr9r7r rLr)rQrr:exp)r)r;y_meanr=s r test_priorrzs #& 1 1 1CKKdK33MFE1%%% 3: q  BGENNBF6<?,C,CQGGGGGBGENNAq11111rc&t|tt}|t d\}}|t d}t|tj |ddttj |tj | z tj |dtj | z ddS)Nr.Tr/irD) rr6r7r8r9X2sample_yr rmeanr:rlvar)r)r;ryr=sampless rtest_sample_statisticsrs #& 1 1 1 5 5a ; ;CKKtK44MFEll2v&&G 3 3Q777 ++--- wRWU^^//111 rctd}t|dtt}t j|jjdksJdS)Nrr) optimizer) rrr6r7r8rrxrPrQrRs rtest_no_optimizerrsU XXF "&D A A A E Ea K KC 6#+# $ $ + + + + + +rtargetrXctjdkrtjdt |t t}|td\}}|td\}}ttj tj ||dS)Nr,r-r.Tr/ return_std)r2r3r4r5rr6r7r8r9r|r rsqrtr:)r)rr;ryr=y_stds rtest_predict_cov_vs_stdrs {e :;;; #& 1 1 1 5 5a ; ;CKKtK44MFEKKtK44MFE//77777rctjd}|ddd}|dddfd|dddfzz}t ddg}t |||}tj|jj dtj|jj ddzksJdS) NrrD)2rEr$rr.rw) rrandom RandomStateuniformrrr6rxrPrQ)rngr7r8r)r;s rtest_anisotropic_kernelrs )   " "C B7##A !!!Q$#!!!Q$-A #s__F "& 1 1 1 5 5a ; ;C 6#+#A& ' '"&1B11E*F*F*J J J J J J Jrcd\}}tjd}|||dzdz }tj|dtjd|zdz|d|z}td d td g|zd g|z ztd dz}tj }tdD]u}t||d ||}||jj} | |tjtjjz ksJ| }vdS)N)rErrErDaxisrFr$)scalesizerr )rbr"rr#)r#$@) noise_levelnoise_level_boundsrw)r)n_restarts_optimizer random_state)rrrrandnrsumnormalrCrr infrangerr6rOrPrQrkfloat32eps) n_samples n_featuresrr7r8r)last_lmlrgprfs rtest_random_startsrs"Iz )   " "C )Z((1,q0A q 1  &Q--    # # $ **3Y* / / 0 sK 3UZ'k]Z=W$$$EEEFFwH %a %!5    #a)) (()9::X 4 4 888888rctjt}tjt}t|z |z }t |}|t |t |d}|t t|td\}}||z|z}||z}|td\}} t||t|| |td\} } | |dzz} |td\} } t| | dS)a Test normalization of the target values in GP Fitting non-normalizing GP on normalized y and fitting normalizing GP on unnormalized y should yield identical results. Note that, here, 'normalized y' refers to y that has been made zero mean and unit variance. r.Tr) normalize_yrr/rEN) rr~r8stdrr6r7r9r|r ) r)ryry_normr;gpr_normr< y_pred_std y_pred_normy_pred_std_norm_r= y_cov_norms rtest_y_normalizationrsDWQZZF F1IIE&jE !F #& 1 1 1CGGAv(v4HHHH LLARD99FJ e^f $Fe#J#+#3#3B4#3#H#H K ,,, O444{{2${//HAu E1H E$$RD$99MAzz*****rctdtz}ddi}tdi|}t|d}|t||t d\}}tjgd}tjgd}t||d d t||d d d S)a Here we test that, when noramlize_y=True, our GP can produce a sensible fit to training data whose variance is significantly larger than unity. This test was made in response to issue #15612. GP predictions are verified against predictions that were made using GPy which, here, is treated as the 'gold standard'. Note that we only investigate the RBF kernel here, as that is what was used in the GPy implementation. The following code can be used to recreate the GPy data: -------------------------------------------------------------------------- import GPy kernel_gpy = GPy.kern.RBF(input_dim=1, lengthscale=1.) gpy = GPy.models.GPRegression(X, np.vstack(y_large), kernel_gpy) gpy.optimize() y_pred_gpy, y_var_gpy = gpy.predict(X2) y_pred_std_gpy = np.sqrt(y_var_gpy) -------------------------------------------------------------------------- rrTrr)gy(V.@gV],;g6}~CgHW -@gƠKQ@)g ؼG'@gsr@g{u>5?g ?g.[ ?gQ?r)rtolatolg333333?Nr&) r8rrr6r7r9r|rrIr )y_large RBF_paramsr)r;r<r y_pred_gpyy_pred_std_gpys rtest_large_variance_yr#s21fG!#&J   :  F "&d C C CCGGAwRD99FJKKKJ XDDDNFJT:::: JTBBBBBBrcptjttdzfj}t d}t |dd}|ttt |dd}|t||td\}}|td\}}|td\}} |td\}} t||ddd ft||ddd fdz t|j d D]2} t||d | ft| | d | f3| td } | td } | j dksJ| j dksJt| | ddd ddftD]}t |d}|ttt |d}|ttjttfjt|jj|jjddS)NrErrF)r)rrTrr/rrD.rr)rwr)rwrErr)rvstackr8Trrr6r7r9r|r rr]r}kernelsrPrQ)y_2dr)r;gpr_2d y_pred_1dy_std_1d y_pred_2dy_std_2dry_cov_1dy_cov_2dr y_sample_1d y_sample_2ds rtest_y_multioutputrZs 9aQZ "Dc " " "F "&De T T TCGGAqMMM %VtQV W W WF JJq$++bT+::Ix ...==Ix++bT+22KAx...55KAx 9QQQT?333 9QQQT?Q#6777 1 &&==HhsF{&;<<<HhsF{&;<<<<,,rR,00K//"/33K   ' ' ' '   * * * * [Aqqq%9:::HH&f$GGG 1 )TJJJ 1biA'')***CK-v~/CQGGGGHHrcd}t||}|tt||jj||jjksJdS)Nc |tjd}|||d}}tdD]{}tj|tjd|dddftjd|dddf}||d}||kr||}}|||fS)NrFrGrrD)rrrr atleast_1drmaximumminimum) obj_func initial_thetarer theta_optfunc_minrrQrs rrz(test_custom_optimizer..optimizersi##A&&+XX . . . 8 r / /AM BJr6!!!Q$<88"*QqqqRSt :U:UVVEe444A8||&+Q8 (""rr)rr6r7r8rOrPrQr))r)rr;s rtest_custom_optimizerrs # # # #&I F F FCGGAqMMM  & &s{'8 9 9C$>vRV$>$W$W!NM n555 ]33333rctddtddz}t}|td\}}|td\}}t|}|td\}}|td\}}t ||t ||dS) Nrr)constant_value_boundsrTrr/r.)rCrrr9r7r)default_kernelgpr1ry_std1y_cov1gpr2y_std2y_cov2s rtest_no_fit_default_predictrss':::S >>>N $ % %D Q4 00IAv Q4 00IAv #> : : :D Q4 00IAv Q4 00IAvff---ff-----rctddg}t|}d}tjt|5|t tdddn #1swxYwYtddgtdd gz}t|}tj d 5}tj d |t tt|d ksJt|djtsJ|djjddksJt|djtsJ|djjddksJ dddn #1swxYwYt#jt d }tddgddg}t|}tj d 5}tj d ||tt|d ksJt|djtsJ|djjddksJt|djtsJ|djjddksJ ddddS#1swxYwYdS)Nr#rrr.zThe optimal value found for dimension 0 of parameter length_scale is close to the specified upper bound 0.001. Increasing the bound and calling fit again may find a better value.rrrgj@T)recordalwaysrErzThe optimal value found for dimension 0 of parameter k1__noise_level is close to the specified upper bound 0.001. Increasing the bound and calling fit again may find a better value.rDzThe optimal value found for dimension 0 of parameter k2__length_scale is close to the specified lower bound 1000.0. Decreasing the bound and calling fit again may find a better value.rrrr"rzThe optimal value found for dimension 0 of parameter length_scale is close to the specified lower bound 10.0. Decreasing the bound and calling fit again may find a better value.zThe optimal value found for dimension 1 of parameter length_scale is close to the specified lower bound 10.0. Decreasing the bound and calling fit again may find a better value.)rrr4warnsrr6r7r8r warningscatch_warnings simplefilterrL issubclasscategoryrargsrtile) r)r;warning_message kernel_sumgpr_sumrX_tile kernel_dimsgpr_dimss rtest_warning_boundsrs dD\ 2 2 2F "& 1 1 1C  ( @ @ @ 1 t === #JAAAJ'j999G   - - - h''' Aq6{{a&),.@AAAAA 1I  "1 %*1 1 1 1 1&),.@AAAAA 1I  "1 %*1 1 1 1 1 1%               4WQ]]FC:C:NNNK'{;;;H   - - - h''' VQ6{{a&),.@AAAAA 1I  "1 %*1 1 1 1 1&),.@AAAAA 1I  "1 %*1 1 1 1 1 1%                  s8!A--A14A1C FF F8CK  KKcdtdz}tddd}||z}t|tt dS)Ng@gI@rrr)r periodicityperiodicity_boundsr.)rrrr6r7r8)k1k2r)s r%test_bound_check_fixed_hyperparameterr!sf 3D))) )B cg   B"WFF+++//155555rctjtjdtj}t |d}|t||jtj dksJ| td\}}t||ttj |dd tjdd }}tj d}tj||d f tj|d fd gd }|t|| td\} } t| ddd fd ttj | ddd | j||fksJ| j|||fksJdS)aCheck that the std. dev. is affected to 1 when normalizing a constant feature. Non-regression test for: https://github.com/scikit-learn/scikit-learn/issues/18318 NaN where affected to the target when scaling due to null std. dev. with constant target. rrXTrrr/r1g& .>)rrErD)r)r] fill_valuerN).rD)rr\r7r]r^rr6 _y_train_stdr4rUr9r r:rr concatenaterfull) r) y_constantr;r<r=r n_targetsrr8Y_predY_covs rtest_constant_targetr(-s2:666J "&d C C CCGGAz  v}S11 1 1 1 1KKdK33MFEFJ'''BGENNCd333371:qyI )   " "C  JJYNJ + + G9a.Q 7 7 7     AGGAqMMMKKdK33MFEF111a4L!$$$BGE&M**Cd;;;; >>>>>rzparams, TypeError, err_msgrrzCalpha must be a scalar or an array with same number of entries as yrrE)r)rz#requires that all bounds are finitectdi|}tj||5|tt ddddS#1swxYwYdS)z0Check that expected error are raised during fit.rNr&)rr4rr6r7r8)params TypeErrorerr_msgr;s rtest_gpr_fit_errorr5s( # , ,V , ,C y 0 0 0 1 s!AAActttt}d}t jt|5|dddddS#1swxYwYdS)z7Check that we raise the proper error in the LML method.r.z.Gradient can only be evaluated for theta!=NonerTrGN) rrr6r7r8r4r ValueErrorrOr;r4s rtest_gpr_lml_errorr9s "#%% 0 0 0 4 4Q : :C>G z 1 1 188 ##$#777888888888888888888sA<<BBc tttt}d}t jt|5|tddddddS#1swxYwYdS)z4Check that we raise the proper error during predict.r.z9At most one of return_std or return_cov can be requested.rT)r0rN) rrr6r7r8r4r RuntimeErrorr9r8s rtest_gpr_predict_errorr<s "#%% 0 0 0 4 4Q : :CIG |7 3 3 399 A$4 888999999999999999999sBB BrTFr%)NrDrctjd}d\}}}|f}|||fz}|f}| |dkr||fz}|||}|||} |j|} t |} | || | | d\} } | | d\}}| j|ksJ| j|ksJ|j|f|zksJdS) aCheck the shapes of y_mean, y_std, and y_cov in single-output (n_targets=None) and multi-output settings, including the edge case when n_targets=1, where the sklearn convention is to squeeze the predictions. Non-regression test for: https://github.com/scikit-learn/scikit-learn/issues/17394 https://github.com/scikit-learn/scikit-learn/issues/18065 https://github.com/scikit-learn/scikit-learn/issues/22174 )r r[NrDrTrr/)rrrrrr6r9r])rr%rrn_samples_trainn_samples_test y_train_shape y_test_shaper+rr,modelr<rrr=s rtest_predict_shapesrFs4 )   % %C29/J$&M% 4 #$LQ#yl2 ii44G YY~z 2 2Fci'G $ = = =E IIgwMM&TM::MFE}}V}55HAu << ' ' ' ' ;, & & & & ;>+l: : : : : : :rctjd}d\}}d}d}|f}|||fz}| |dkr|||f}n||f}|||} |||} |j|} t |} | | | | | |} | j|ksJdS) a&Check the shapes of y_samples in single-output (n_targets=0) and multi-output settings, including the edge case when n_targets=1, where the sklearn convention is to squeeze the predictions. Non-regression test for: https://github.com/scikit-learn/scikit-learn/issues/22175 r>)rr?r[rwNrDr@r)rrrrrr6r}r])rr%rrrAn_samples_X_testn_samples_y_testrCrDr+rr,rE y_sampless rtest_sample_y_shapesrKs )   % %C"&J$&M% 4 Q()5EF (*:; ii44G YY' 4 4Fci'G $ = = =E IIgwv1ABBI ?l * * * * * *r)NrDrErFrrDrwcptjd}|dd}|d||nd}t |}|||j}||||||j}||ksJdS)zJCheck the output shape of `sample_y` is consistent before and after `fit`.rrFNrDr%r)rrrrrr}r]r6)r%rrr7r8rEshape_before_fitshape_after_fits rtest_sample_y_shape_with_priorrQs )   % %C "aA "9#8iia@@A $y 9 9 9E~~a9~==C IIaOOOnnQ)n<<BO  . . . . . .rc0tjd}d}||d}||||nd}t |}||d\}}||d \}} |||||d\} } ||d \}} |j| jksJ|j| jksJ| j| jksJdS) zIIAFFA $y 9 9 9E!MM!M==J ==t=44LAy IIaOOO--d-;;Ix--d-33KAx  y . . . . ?hn , , , , ?hn , , , , , ,rcPtjd}|dd}|dd}t d}t jtd5|||d d d d S#1swxYwYd S) zmCheck that an error is raised when the number of targets seen at fit is inconsistent with n_targets. rrrFrErDrNz!The number of targets seen in `y`rN) rrrrrr4rr7r6)rr7r8rEs rtest_n_targets_errorr\,s )   " "C "aA "aA $q 1 1 1E z)L M M M !Qs7BB"BceZdZdZdZdS) CustomKernelz A custom kernel that has a diag method that returns the first column of the input matrix X. This is a helper for the test to check that the input matrix X is not mutated. c|dddfS)Nrr&)selfr7s rr:zCustomKernel.diag@sAwrN)__name__ __module__ __qualname____doc__r:r&rrr^r^9s- rr^ctttt}t jt}|td\}}tt|dS)z Check that the input X is not modified by the predict method of the GaussianProcessRegressor when setting return_std=True. Non-regression test for: https://github.com/scikit-learn/scikit-learn/issues/24340 r.TrN) rr^r6r7r8rcopyr|r9r )r;X2_copyrs r#test_gpr_predict_input_not_modifiedrhDsd #,.. 9 9 9 = =a C CCgbkkG ;;rd; + +DAqB     r)Prdrr2r numpyrr4scipy.optimizersklearn.exceptionsrsklearn.gaussian_processr sklearn.gaussian_process.kernelsrrrr r rC4sklearn.gaussian_process.tests._mini_sequence_kernelr sklearn.utils._testingr r rrr atleast_2drr7r|rJr8r'rnon_fixed_kernelsmark parametrizer>rMrSrVr`rhrprurzrrr\r]r^rrrrrrrrrrrrr(r0zerosr7rr5r9r<rFrKrQrZr\r^rhr&rrrus--   ((((((111111====== ONNNNNBM000113R],,,--/AaDDJJLLsAAA CSCSk:::Ac;##3KPPPPAc;##3KPPPPakAc;##3KPPPPak ML'LLL7++ - -,+ - - - -#455657++,+7++;;,+;#455  65 #455 > >65 >7++ > >,+ >7++ 2 2,+ 27++,+",,,7++Awrwqwqz'L'L'L#MNN88ON,+8 K K K67++$+$+,+$+N4C4C4Cn+H+H+H\#455652   7++44,+4*..."F F F R 6 6 67++%<%<,+%