Resultats obetnus sur les donnees du beauty contest.

le 19 juillet 2004, avec l'algo de trust regions en yorick (et
decorrelation des parametres presque aux bornes apres inversion..)
J'ai introduit un modele constitué de trois briques:
- un objet ponctuel (dirac)
- une gaussienne elliptique
- un anneau circulaire

Au total, 14 parametres a ajuster.

Script:
<<
w=lit_load_data("~/data/data1.oifits");
lit_load_model(w, LIT_PONCT_XY, 0);
lit_grow_model(w, LIT_GAUSSELLIP_XY, 0);
lit_grow_model(w, LIT_CIRCRING_XY, 0);
z=lit_random_fit(w, [0.0,-15.0,-15.0,0.0,-15.0,-15.0,0.0,1.0,-PI,0.0,-15.0,-15.0,0.0,0.0], [1.0,15.0,15.0,1.0,15.0,15.0,15.0,5.0,PI,1.0,15.0,15.0,10.0,5.0], nbfits=10000, method=LIT_TRNEW2, normalize=1);

Parameter number 1 is bounded between 0 and 1 
Parameter number 2 is bounded between -15 and 15 
Parameter number 3 is bounded between -15 and 15 
Parameter number 4 is bounded between 0 and 1 
Parameter number 5 is bounded between -15 and 15 
Parameter number 6 is bounded between -15 and 15 
Parameter number 7 is bounded between 0 and 15 
Parameter number 8 is bounded between 1 and 5 
Parameter number 9 is bounded between -3.14159 and 3.14159 
Parameter number 10 is bounded between 0 and 1 
Parameter number 11 is bounded between -15 and 15 
Parameter number 12 is bounded between -15 and 15 
Parameter number 13 is bounded between 0 and 10 
Parameter number 14 is bounded between 0 and 5 

chi2=*z.chi2
chi2best=z.chi2_best = 22.0012
a_best=*z.a_best=
[0.0224593,-6.75886,2.46975,0.719879,-2.48148,1.33858,3.13048,1.40666,1.9759,
0.255414,-1.24753,1.84695,2.58067,2.98416]

Avec ces valeurs de parametres, on a obtenu les graphes suivants:
data1_bestfit_vis2.ps
data1_bestfit_clo.ps