3f32776f0f58c961a1ad625f3e5e34ce31a9ebbb,examples/stochastic_volatility.py,,,#,7

Before Change

gam = Var("gam", Normal(-5., 3.**-2))
k = Var("k", Uniform(0,1), testval = .95)

tau, ltau = model.TransformedVar(
                 "tau", Tpos(10,.05,.05),
                 transform = exp, logjacobian = lambda x: x, testval = -2.3)

lvol = Var("lvol", timeseries.AR1(k, tau**-2), shape = n)

rtau = exp((lvol+gam)*-2.)

lreturns = Data(returns, Normal(0, rtau))



//fitting

start = find_MAP(model, vars = [lvol])
start = find_MAP(model,start, vars = [gam, k])

H = model.d2logpc()

def hessian(q, tau_scale): 

After Change

                 "sd", Exponential(1./.02),
                 transform = exp, logjacobian = lambda x: x, testval = -2.3)

nu = Var("nu", Exponential(1./10))

lvol = Var("lvol", timeseries.RW(sd**-2), shape = n)

lreturns = Data(returns, T(nu, lam = exp(-2*lvol)))

Instances