beta = 1 #Transmission coefficient tau = .2 #infectious period. FIXED tf = 36 y0 = [.999, 0.001, 0.0] def model(theta): beta = theta[0] def sir(y, t): '''ODE model''' S, I, R = y return [-beta * I * S, #dS/dt beta * I * S - tau * I, #dI/dt tau * I] #dR/dt y = odeint(sir, inits, np.arange(0, tf, 1)) #np.arange(t0,tf,step)) return y F = FitModel(300, model, y0, tf, ['beta'], ['S', 'I', 'R'], wl=36, nw=1, verbose=1, burnin=100) F.set_priors(tdists=[st.norm], tpars=[(1.1, .2)], tlims=[(0.5, 1.5)], pdists=[st.uniform] * 3, ppars=[(0, .1), (0, .1), (.8, .2)], plims=[(0, 1)] * 3) d = model([1.0]) #simulate some data noise = st.norm(0, 0.01).rvs(36) dt = {'I': d[:, 1] + noise} # add noise F.run(dt, 'MCMC', likvar=1e-5, pool=True, monitor=[]) #==Uncomment the line below to see plots of the results F.plot_results()
ser = st.nanmean(series, axis=0) # print series.shape, ser.shape return ser d = runModel([beta, alpha, sigma]) # ~ import pylab as P # ~ P.plot(d) # ~ P.show() dt = {'S': d[:, 0], 'E': d[:, 1], 'I': d[:, 2], 'A': d[:, 3], 'R': d[:, 4]} F = FitModel(900, runModel, inits, tf, ['beta', 'alpha', 'sigma'], ['S', 'E', 'I', 'A', 'R'], wl=140, nw=1, verbose=0, burnin=100) F.set_priors(tdists=[st.uniform] * 3, tpars=[(0.00001, .0006), (.1, .5), (0.0006, 1)], tlims=[(0, .001), (.001, 1), (0, 1)], pdists=[st.uniform] * 5, ppars=[(0, 500)] * 5, plims=[(0, 500)] * 5) F.run(dt, 'MCMC', likvar=1e1, pool=0, monitor=[]) # ~ print F.optimize(data=dt,p0=[0.1,.5,.1], optimizer='oo',tol=1e-55, verbose=1, plot=1) # ==Uncomment the line below to see plots of the results F.plot_results()
F = FitModel(1000, model, inits, tf, tnames, pnames, wl, nw, verbose=1, burnin=200, constraints=[]) F.set_priors(tdists=nt * [st.beta], tpars=tpars, tlims=tlims, pdists=[st.beta] * nph, ppars=[(1, 1)] * nph, plims=[(0, 1)] * nph) F.run(dt, 'DREAM', likvar=1e-4, pool=False, ew=0, adjinits=False, dbname=modname, monitor=['I', 'S']) #~ print F.AIC, F.BIC, F.DIC #print F.optimize(data=dt,p0=[s0,s1,s2], optimizer='scipy',tol=1e-55, verbose=1, plot=1) F.plot_results(['S', 'I'], dbname=modname, savefigs=1)