weights = np.log(dists.normal(mean=0.0, std=pop[:, 3]).pdf(sse))
# see if return both weights and predicted vals
if save_posts:
return weights, pred
else:
return weights
# set up the data
xData = np.array([5.357, 9.861, 5.457, 5.936, 6.161, 6.731])
yData = np.array([0.376, 7.104, 0.489, 1.049, 1.327, 2.077])
# set up the parameters
params = [
Param(name='a', prior=dists.uniform(-100, 100)),
Param(name='b', prior=dists.uniform(-100, 100)),
Param(name='c', prior=dists.uniform(-100, 100)),
Param(
name='delta',
display_name=r'$\mathbf{\delta}$',
prior=dists.exp(20),
init_prior=dists.uniform(0, 10),
),
]
# set up mod
mod = Model(name='fun',
params=params,
like_fun=eval_fun,
like_args=(xData, yData),
def eval_fun(abc, pop, *args):
res = Parallel(n_jobs=n_jobs)(delayed(eval_prop)(indiv, args[0])
for indiv in pop)
weights = np.asarray(res)
if abc._save_posts:
return weights, None
else:
return weights
# set up the parameters
params = [
Param(name='mu', prior=dists.uniform(-20, 20)),
Param(name='sd', prior=dists.uniform(0, 20)),
]
burnin = 50
iterations = 500
# set up abc
do_true = True
abc_true = DEMC(params,
eval_fun,
eval_args=(do_true, ),
num_groups=1,
group_size=25,
proposal_gen=DE(gamma_best=None, rand_base=True),
migration_prob=0.0,
# set up the params
params = [
Param(name='a',
prior=dists.trunc_normal(5.0, 20.0, lower=0.0, upper=50.0)),
Param(name='b', prior=dists.normal(0.0, 5.0)),
Param(name='c',
prior=dists.trunc_normal(5.0, 10.0, lower=0.0, upper=30.0)),
Param(name='kappa',
prior=dists.normal(0.0, 1.4),
transform=dists.invlogit),
Param(name='beta',
prior=dists.normal(0.0, 1.4),
transform=dists.invlogit),
Param(name='alpha',
prior=dists.trunc_normal(2.5, 10.0, lower=0.0, upper=30.0)),
Param(name='t0', prior=dists.uniform(0., min_rt))
]
pnames = [p.name for p in params]
# instantiate model object
m = Model('urdm_lca',
params=params,
like_fun=eval_fun_lca,
like_args=(s, ),
init_multiplier=4,
verbose=True,
purify_every=5)
# set up the run name
output_name = 'rdm_lca_tcv_both_cb_afrl_mri_subj_' + str(
int(s)) + '.tgz'
def eval_fun(abc, pop, *args):
res = Parallel(n_jobs=n_jobs)(delayed(eval_prop)(indiv, args[0]) for indiv in pop)
weights = np.asarray(res)
if abc._save_posts:
return weights, None
else:
return weights
# set up the parameters
params = [Param(name="mu", prior=dists.uniform(-20, 20)), Param(name="sd", prior=dists.uniform(0, 20))]
burnin = 50
iterations = 500
# set up abc
do_true = True
abc_true = DEMC(
params,
eval_fun,
eval_args=(do_true,),
num_groups=1,
group_size=25,
proposal_gen=DE(gamma_best=None, rand_base=True),
migration_prob=0.0,
initial_zeros_ok=False,
])
for s in dat.keys():
# Append a new model, note the use of the hyperpriors
params = [
# sig_b is the sigmoid transition point
#Param(name='sig_b',
# prior=dists.uniform(0., 15.0), # change to number of tau stars
# ),
# new item strength
Param(
name='alpha',
display_name=r'$\alpha$',
prior=dists.uniform(0, 10.0),
#init_prior=dists.trunc_normal(mean=.25,std=.5,lower=0,upper=5)
),
Param(
name='nu',
display_name=r'$\nu$',
prior=dists.uniform(0, 10.0),
#init_prior=dists.trunc_normal(mean=.25,std=.5,lower=0,upper=5)
),
# threshold
Param(
name='a',
display_name=r'$a$',
prior=dists.uniform(0, 10.0),
#init_prior=dists.trunc_normal(mean=.25,std=.5,lower=0,upper=5)
),
# calculate the weight with a normal kernel
weights = np.log(norm.pdf(sse,scale=pop[:,3]))
#weights = np.log(norm.pdf(sse,scale=.1))
# see if return both weights and predicted vals
if abc._save_posts:
return weights,pred
else:
return weights
# set up the data
xData = np.array([5.357, 9.861, 5.457, 5.936, 6.161, 6.731])
yData = np.array([0.376, 7.104, 0.489, 1.049, 1.327, 2.077])
# set up the parameters
params = [Param(name='a',prior=dists.uniform(-100,100)),
Param(name='b',prior=dists.uniform(-100,100)),
Param(name='c',prior=dists.uniform(-100,100)),
Param(name='delta',display_name=r'$\mathbf{\delta}$',
prior=dists.exp(20),
init_prior=dists.uniform(0,10),
),
]
# set up abc
abc = DEMC(params, eval_fun, eval_args = (xData,yData),
num_groups=4, group_size=30,
proposal_gen=DE_LOCAL_TO_BEST(),
migration_prob=0.1, initial_zeros_ok=False,
use_priors=True, save_posts=True)
Param(
name='nu',
display_name=r'$\nu$',
prior=dists.trunc_normal(mean=2., std=10., lower=0., upper=10.),
),
Param(name='a',
display_name=r'$a$',
prior=dists.trunc_normal(mean=2., std=10., lower=0., upper=10.)),
Param(name='w',
display_name=r'$w$',
prior=dists.normal(mean=0, std=1.4),
transform=dists.invlogit),
Param(
name='t0',
display_name=r'$t_0$',
prior=dists.uniform(0, min_RT),
),
]
# grab the param names
pnames = [p.name for p in params]
# initialize the model
m = Model(s,
params=params,
like_fun=eval_fun,
like_args=(s, data_sub, pnames),
num_chains=80,
init_multiplier=4,
verbose=True)
# set number of desired burn-in trials
display_name=r'$\beta$',
prior=dists.CustomDist(
pdf=lambda x: np.exp(-1.5 * np.log(1 + x**2)),
rvs=dists.laplace(0, 5).rvs))
# Fixed noise across subjects
# Using a custom Jeffreys' prior
sigma = Param(name='sigma',
display_name=r'$\sigma$',
prior=dists.CustomDist(pdf=lambda x: np.exp(-np.log(x)),
rvs=dists.dists.invgamma(1, 1).rvs))
# Hyperprior over intercept using a normal distribution
halpha = HyperPrior('alpha',
dists.normal,
params=[
Param(name='mu', prior=dists.uniform(-50, 50)),
Param(name='sig', prior=dists.invgamma(1, 1))
])
# set up the submodels for each participant
smods = []
for j in range(nsubj):
# Append a new model, note the use of the hyperprior for setting
# up the intercept param, and the fixed beta and sigma across
# participants
smods.append(
Model(name=str(j),
params=[
Param(name='alpha', display_name=r'$\alpha$', prior=halpha),
beta, sigma
],
