Contents
Load my model
dataModel = loadModel('OneStep');
Simulatate our "real model"
we use the nominal model to generate some data
exp1 = constructExperiment(30*60, 10); data = simulate(dataModel, exp1);
Scramble the parameters onthe model
m = dataModel; m.p = m.p.*random('log',0,0.5,size(dataModel.p)); m = postProcessModel(m); beforeFit = simulate(m, exp1); % Plot the data and the model before the fit on the same axis figure(1),clf plot(data.t, data.y(data.t), 'o'); hold on; drawnow; plot(beforeFit.t, beforeFit.y(beforeFit.t), '--'); hold off; disp('Hit any Key to continue'); pause;
Hit any Key to continue
Fit the new model to data generated from the old model
figure(2),clf obj = constructObjectiveOutputISE(m, 1:m.nY, data.y); [mFit,G] = fitParameters(m, obj, exp1); afterFit = simulate(mFit, exp1);
max Line search Directional First-order
Iter F-count f(x) constraint steplength derivative optimality Procedure
0 1 82.2703 -0.0109
1 4 58.2864 -0.01176 0.000501 3.28e+005 4.27e+005
2 6 48.6478 -0.02354 0.001 6.04e+003 2.33e+004
3 8 34.5166 -0.02663 0.0233 -200 1.25e+003
4 12 32.9864 -0.02456 0.0778 13.1 2.65e+003
5 16 32.4782 -0.02448 0.00332 55.3 2.47e+003
6 20 28.1936 -0.02332 0.25 -14.4 696
7 23 5.686 -0.01166 0.5 -27.1 5.71e+003 Hessian modified
8 27 2.64897 -0.008745 0.25 31 6.09e+003
9 30 2.62306 -0.01259 0.06 37.3 5.11e+003
10 34 0.654004 -0.01574 0.0542 2.38 800
11 40 0.48107 -0.01658 0.0388 5.2 1.27e+003
12 43 0.221215 -0.01567 0.5 1.76 690
13 45 0.00785286 -0.01599 1 -0.0668 108
14 47 0.00038813 -0.016 1 0.00108 5.76
15 49 0.00035435 -0.016 1 1.18e-006 0.924
Optimization terminated: magnitude of directional derivative in search
direction less than 2*options.TolFun and maximum constraint violation
is less than options.TolCon.
No active inequalities.
Plot the results
figure(3),clf plot(data.t, data.y(data.t), 'o'); hold on; plot(beforeFit.t, beforeFit.y(beforeFit.t), '--'); plot(afterFit.t, afterFit.y(afterFit.t), '-'); hold off;
