[1181] | 1 | function [w x]=sidp3(sidp_parameters, compare_parameters,system,apriori)
|
---|
| 2 | eps=10^-4;
|
---|
| 3 | mode=0;
|
---|
| 4 | q=0;
|
---|
| 5 | col(1)='b';col(2)='r' ;col(3)='c' ;col(4)='m' ;col(5)='y';
|
---|
| 6 | a=0;
|
---|
| 7 | compare_parameters.student=quantile(trnd((sidp_parameters.num_of_candidates-1)*ones(1,100000)),(1-compare_parameters.alpha)^(1/(sidp_parameters.num_of_candidates-1)));
|
---|
| 8 | compare_parameters.rinott=4;
|
---|
| 9 |
|
---|
| 10 | %apriori.eta0=abs(apriori.y0)/system.sigma;
|
---|
| 11 | apriori.eta0=0;
|
---|
| 12 |
|
---|
| 13 | apriori.beta0=abs(apriori.b0)/sqrt(apriori.P0);
|
---|
| 14 | apriori.beta0_range=2*apriori.P0/sqrt(apriori.P0);
|
---|
| 15 | if apriori.beta0-apriori.beta0_range<eps
|
---|
| 16 | apriori.beta0=(abs(apriori.b0)+2*apriori.P0)/2/sqrt(apriori.P0);
|
---|
| 17 | apriori.beta0_range=apriori.beta0-eps;
|
---|
| 18 | end
|
---|
| 19 |
|
---|
| 20 | [H C]=init_hyperstate(sidp_parameters.n_grid,apriori,mode); %vytvori H a mi, nastavi H0 a mi*
|
---|
| 21 |
|
---|
| 22 | min_distance_C=0.5*sidp_parameters.search_region_init;
|
---|
| 23 | min_distance_H=(H(1,end)-H(1,1))/50;
|
---|
| 24 |
|
---|
| 25 | %[los1 los2]=mc_study(system,apriori,1000); q=q+1; w(q)=los1(1)/los1(3); x(q)=los2(1)/los2(3);
|
---|
| 26 |
|
---|
| 27 |
|
---|
| 28 | for j=1:sidp_parameters.n_iter
|
---|
| 29 | j
|
---|
| 30 | %search_region=sidp_parameters.gama^(j-1)*sidp_parameters.lambda^(i-1)*sidp_parameters.search_region_init;
|
---|
| 31 | search_region=sidp_parameters.gama^(j-1)*sidp_parameters.search_region_init;
|
---|
| 32 | %prolez mrizku a iteruj rizeni
|
---|
| 33 | for k=1:sidp_parameters.n_grid
|
---|
| 34 | candidates=generate_candidates(C(k), search_region, sidp_parameters.num_of_candidates);
|
---|
| 35 | C(k)=choose_best_control(H,C,k,apriori.eta0,sidp_parameters.horizont,candidates,compare_parameters);
|
---|
| 36 | end
|
---|
| 37 |
|
---|
| 38 | %[los1 los2]=mc_study(system,apriori,1000); q=q+1; w(q)=los1(2)/los1(3); x(q)=los2(2)/los2(3);
|
---|
| 39 |
|
---|
| 40 | vypis=H;
|
---|
| 41 | save 'Beta.txt' vypis -ASCII;
|
---|
| 42 | save 'C.txt' C -ASCII;
|
---|
| 43 |
|
---|
| 44 | %pokud je mrizka prilis hruba, zjemni ji tamkde je treba
|
---|
| 45 | for k=1:sidp_parameters.n_grid-1
|
---|
| 46 | if (abs(C(k)-C(k+1))>min_distance_C && sidp_parameters.n_grid<sidp_parameters.n_grid_max && H(k+1)-H(k)>min_distance_H)
|
---|
| 47 | % [H C]=update(H,C,k); sidp_parameters.n_grid=sidp_parameters.n_grid+1;
|
---|
| 48 | end
|
---|
| 49 | end
|
---|
| 50 |
|
---|
| 51 | %subplot(2,1,1);
|
---|
| 52 | %plot(H,C,['',col(j)]); hold on; a=max(a,max(C(:)));
|
---|
| 53 | end
|
---|
| 54 |
|
---|
| 55 | % xlabel('\fontsize{18} \beta'); ylabel('\fontsize{18} \nu^{ (2)}_0'); set(gca,'ylim',[0 a]); set(gca,'xlim',[0 H(end)]); legend('\pi_1','\pi_2','\pi_3','\pi_4');
|
---|
| 56 | % subplot(2,1,2);
|
---|
| 57 | % plot(0:sidp_parameters.n_iter,[w' x' ones(size(x,2),1)]); xlabel('\fontsize{18} iterace'); ylabel('\fontsize{18} relativn�tr�'); legend('mean','median');set(gca,'ylim',[0.99 max(w(1),x(1))]); set(gca,'xtick',0:sidp_parameters.n_iter); set( get(gcf, 'Children'), 'FontSize', 18);
|
---|
| 58 | %[w;x]
|
---|
| 59 | end
|
---|
| 60 |
|
---|