def test_markov_chain():
from dolo import yaml_import
fname = 'examples/models/rbc_mfga.yaml'
model = yaml_import(fname)
from dolo.numeric.discretization import multidimensional_discretization
import numpy
sigma = numpy.array([[0.01]])
rho = 0.01
[P,Q] = multidimensional_discretization(rho, sigma, 3)
print(model.markov_chain)
def AR1(rho, sigma, *pargs, **kwargs):
rho_array = np.array(rho, dtype=float)
sigma_array = np.atleast_2d( np.array(sigma, dtype=float) )
try:
assert(rho_array.ndim<=1)
except:
raise Exception("When discretizing a Vector AR1 process, the autocorrelation coefficient must be as scalar. Found: {}".format(rho_array))
try:
assert(sigma_array.shape[0] == sigma_array.shape[1])
except:
raise Exception("The covariance matrix for a Vector AR1 process must be square. Found: {}".format())
from dolo.numeric.discretization import multidimensional_discretization
[P,Q] = multidimensional_discretization(rho_array, sigma_array, *pargs, **kwargs)
return P,Q
def AR1(rho, sigma, *pargs, **kwargs):
rho_array = np.array(rho, dtype=float)
sigma_array = np.atleast_2d( np.array(sigma, dtype=float) )
try:
assert(rho_array.ndim<=1)
except:
raise Exception("When discretizing a Vector AR1 process, the autocorrelation coefficient must be as scalar. Found: {}".format(rho_array))
try:
assert(sigma_array.shape[0] == sigma_array.shape[1])
except:
raise Exception("The covariance matrix for a Vector AR1 process must be square. Found: {}".format())
from dolo.numeric.discretization import multidimensional_discretization
[P,Q] = multidimensional_discretization(rho_array, sigma_array, *pargs, **kwargs)
return P,Q
def discretize(self, N=3):
rho = self.rho
Sigma = self.Sigma
try:
assert(abs(np.eye(rho.shape[0])*rho[0,0]-rho).max() <= 1)
except:
raise Exception("When discretizing a Vector AR1 process, the autocorrelation coefficient must be as scalar. Found: {}".format(rho_array))
from dolo.numeric.discretization import multidimensional_discretization
[P,Q] = multidimensional_discretization(rho[0,0], Sigma, N=N)
return DiscreteMarkovProcess(values=P, transitions=Q)
def __init__(self, rho=None, sigma=None, N=2):
rho_array = np.array(rho, dtype=float)
sigma_array = np.atleast_2d(np.array(sigma, dtype=float))
try:
assert(rho_array.ndim <= 1)
except:
raise Exception("When discretizing a Vector AR1 process, the autocorrelation coefficient must be as scalar. Found: {}".format(rho_array))
try:
assert(sigma_array.shape[0] == sigma_array.shape[1])
except:
raise Exception("The covariance matrix for a Vector AR1 process must be square. Found: {}".format())
from dolo.numeric.discretization import multidimensional_discretization
[P,Q] = multidimensional_discretization(rho_array, sigma_array)
self.P = P
self.Q = Q
self.extend([P, Q])
def discretize_mc(self, N=3):
rho = np.array([[self.ρ]])
Sigma = np.array([[self.σ]])
μ = np.array([self.μ])
try:
assert(abs(np.eye(rho.shape[0])*rho[0,0]-rho).max() <= 1)
except:
raise Exception("When discretizing a Vector AR1 process, the autocorrelation coefficient must be as scalar. Found: {}".format(rho_array))
from dolo.numeric.discretization import multidimensional_discretization
[P,Q] = multidimensional_discretization(rho[0,0], Sigma, N=N)
P += μ[None,:]
return MarkovChain(values=P, transitions=Q)
def __init__(self, rho=None, sigma=None, N=2):
rho_array = np.array(rho, dtype=float)
sigma_array = np.atleast_2d(np.array(sigma, dtype=float))
try:
assert (rho_array.ndim <= 1)
except:
raise Exception(
"When discretizing a Vector AR1 process, the autocorrelation coefficient must be as scalar. Found: {}"
.format(rho_array))
try:
assert (sigma_array.shape[0] == sigma_array.shape[1])
except:
raise Exception(
"The covariance matrix for a Vector AR1 process must be square. Found: {}"
.format())
from dolo.numeric.discretization import multidimensional_discretization
[P, Q] = multidimensional_discretization(rho_array, sigma_array)
self.P = P
self.Q = Q
self.extend([P, Q])
