functools.partial(loggaussian, mu=mu[i], sigma2=sigma2[i])
for i in range(num_mixtures)
]
logjoint_ = functools.partial(logjointmix, logps=logps, alpha=alpha)
def logjoint(sample):
return logjoint_(sample)
true_mean, true_cov = tndmeancov(mu, sigma2, alpha)
sampler = lambda nsamples: samplergaussian(mu, sigma2, alpha, nsamples)
#%%
nsamples = 10
ninducing = 50
samples = sampling.sampling1(nsamples, dim, scale=10.0)
mu0 = torch.zeros(dim)
cov0 = (20.0 / 3)**2 * torch.ones(dim)
acquisition = "prospective"
folder_name = "example3"
try:
os.mkdir(folder_name)
except:
pass
#%% The main training class is initiated, it's gp model optimized,
# and initial component set
vb = VariationalBoosting(dim,
logjoint,
samples,
mu0,
cov0,
nu = torch.rand(dim) * (2 + 0.5 * dim - 2.5) + 2.5
A = torch.eye(dim)
logjoint_ = functools.partial(logrotated,
f=functools.partial(logtnd, nu=nu),
A=A)
device = "cpu"
def logjoint(sample):
return logjoint_(sample.to("cpu")).to(device)
true_mean = torch.zeros(dim)
true_cov = A @ torch.diag(nu / (nu - 2) * torch.ones(dim)) @ A.t()
sampler = lambda nsamples: samplertndrot(nsamples, nu, A, dim)
#%%
nsamples = 10 * dim
samples = sampling.sampling1(nsamples, dim, scale=5.0, device=device)
mu0 = torch.zeros(dim).to(device)
cov0 = (20.0 / 3)**2 * torch.ones(dim).to(device)
#%%
training_interval = 20
acquisitions = ["prospective", "mmlt"]
vb = VariationalBoosting(dim,
logjoint,
samples,
mu0,
cov0,
kernel_function="PMat",
matern_coef=2.5,
degree_hermite=60)
vb.optimize_bmc_model(maxiter=200)
vb.update_full()