def default_grid_exponential_distance_poisson(dist_scale = 1.0, rate_scale_scale = 1.0,
GRIDN = 10):
mu = util.logspace(0.1, 1.0, GRIDN) * dist_scale
rate_scale = util.logspace(0.1, 1.0, GRIDN) * rate_scale_scale
hps = []
for m in mu:
for r in rate_scale:
hps.append({'mu_hp' : m,
'rate_scale_hp' : r})
return hps
def default_grid_exponential_distance_poisson(dist_scale=1.0,
rate_scale_scale=1.0,
GRIDN=10):
mu = util.logspace(0.1, 1.0, GRIDN) * dist_scale
rate_scale = util.logspace(0.1, 1.0, GRIDN) * rate_scale_scale
hps = []
for m in mu:
for r in rate_scale:
hps.append({'mu_hp': m, 'rate_scale_hp': r})
return hps
def default_grid_logistic_distance_poisson(dist_scale = 1.0, rate_scale_scale = 1.0,
GRIDN = 10):
mu = util.logspace(0.1, 1.0, GRIDN) * dist_scale
rate_scale = util.logspace(0.1, 1.0, GRIDN) * rate_scale_scale
hps = []
for m in mu:
for r in rate_scale:
hps.append({'mu_hp' : m,
'lambda' : m,
'rate_scale_hp' : r,
'rate_min' : 0.01})
return hps
def default_grid_normal_inverse_chi_sq(mu_scale=1.0, var_scale=1.0, GRIDN=10):
mu = np.linspace(-1.0, 1.0,
GRIDN + 1) * mu_scale #+1 to always include zero
sigmasq = util.logspace(0.1, 1.0, GRIDN) * var_scale
kappa = util.logspace(0.1, 10.0, GRIDN)
nu = util.logspace(0.1, 10.0, GRIDN)
hps = []
for m in mu:
for s in sigmasq:
for k in kappa:
for n in nu:
hps.append({'mu': m, 'kappa': k, 'sigmasq': s, 'nu': n})
return hps
def default_grid_logistic_distance_poisson(dist_scale=1.0,
rate_scale_scale=1.0,
GRIDN=10):
mu = util.logspace(0.1, 1.0, GRIDN) * dist_scale
rate_scale = util.logspace(0.1, 1.0, GRIDN) * rate_scale_scale
hps = []
for m in mu:
for r in rate_scale:
hps.append({
'mu_hp': m,
'lambda': m,
'rate_scale_hp': r,
'rate_min': 0.01
})
return hps
def default_grid_normal_inverse_chi_sq(mu_scale = 1.0,
var_scale = 1.0,
GRIDN = 10):
mu = np.linspace(-1.0, 1.0, GRIDN+1) * mu_scale #+1 to always include zero
sigmasq = util.logspace(0.1, 1.0, GRIDN) * var_scale
kappa = util.logspace(0.1, 10.0, GRIDN)
nu = util.logspace(0.1, 10.0, GRIDN)
hps = []
for m in mu:
for s in sigmasq:
for k in kappa:
for n in nu:
hps.append({'mu' : m,
'kappa' : k,
'sigmasq' : s,
'nu' : n})
return hps