Initialzing NUTS with covariance matrix as mass matrix

junpenglao · October 16, 2019, 6:12am

You can follow the code here:

pymc-devs/pymc3/blob/75ec55f08b9dbe35a8d94b97d157138c856c6bf3/pymc3/sampling.py#L1568-L1579


elif init == 'nuts':
    init_trace = pm.sample(draws=n_init, step=pm.NUTS(),
                           tune=n_init // 2,
                           random_seed=random_seed)
    cov = np.atleast_1d(pm.trace_cov(init_trace))
    start = list(np.random.choice(init_trace, chains))
    potential = quadpotential.QuadPotentialFull(cov)
else:
    raise ValueError(
        'Unknown initializer: {}.'.format(init))


step = pm.NUTS(potential=potential, model=model, **kwargs)

Topic		Replies	Views
Initial mass matrix Questions	1	776	June 17, 2019
Trouble with Initilization and Posterior Sampling Questions theano	13	2225	June 15, 2018
Hamiltonian MC with different variable dimensions Questions	3	871	July 23, 2019
NUTS and zero derivative error Questions	6	7045	August 18, 2019
Initialize Once for Multiple Runs? Questions	2	722	January 10, 2020

Initialzing NUTS with covariance matrix as mass matrix

Related topics