def test_get_base_sample_shape(self):
sampler = PairwiseSobolQMCNormalSampler(num_samples=4)
self.assertFalse(sampler.resample)
self.assertEqual(sampler.sample_shape, torch.Size([4]))
self.assertTrue(sampler.collapse_batch_dims)
# check sample shape non-batched
posterior = _get_test_posterior(self.device)
bss = sampler._get_base_sample_shape(posterior=posterior)
self.assertEqual(bss, torch.Size([4, 3, 1]))
# check sample shape batched
posterior = _get_test_posterior(self.device, batched=True)
bss = sampler._get_base_sample_shape(posterior=posterior)
self.assertEqual(bss, torch.Size([4, 1, 3, 1]))
def test_fantasize(self):
for batch_shape, dtype in itertools.product(
(torch.Size(), torch.Size([2])), (torch.float, torch.double)):
tkwargs = {"device": self.device, "dtype": dtype}
X_dim = 2
model, model_kwargs = self._get_model_and_data(
batch_shape=batch_shape, X_dim=X_dim, **tkwargs)
# fantasize
X_f = torch.rand(torch.Size(batch_shape + torch.Size([4, X_dim])),
**tkwargs)
sampler = PairwiseSobolQMCNormalSampler(num_samples=3)
fm = model.fantasize(X=X_f, sampler=sampler)
self.assertIsInstance(fm, model.__class__)
fm = model.fantasize(X=X_f,
sampler=sampler,
observation_noise=False)
self.assertIsInstance(fm, model.__class__)
def test_forward_no_collapse(self):
for dtype in (torch.float, torch.double):
# no resample
sampler = PairwiseSobolQMCNormalSampler(num_samples=4,
seed=1234,
collapse_batch_dims=False)
self.assertFalse(sampler.resample)
self.assertEqual(sampler.seed, 1234)
self.assertFalse(sampler.collapse_batch_dims)
# check samples non-batched
posterior = _get_test_posterior(device=self.device, dtype=dtype)
samples = sampler(posterior)
self.assertEqual(samples.shape, torch.Size([4, 3, 2]))
self.assertEqual(sampler.seed, 1235)
# ensure samples are the same
samples2 = sampler(posterior)
self.assertTrue(torch.allclose(samples, samples2))
self.assertEqual(sampler.seed, 1235)
# ensure this works with a differently shaped posterior
posterior_batched = _get_test_posterior(device=self.device,
dtype=dtype,
batched=True)
samples_batched = sampler(posterior_batched)
self.assertEqual(samples_batched.shape, torch.Size([4, 3, 3, 2]))
self.assertEqual(sampler.seed, 1236)
# ensure error is rasied when number of points are < 2
posterior = _get_test_posterior(device=self.device,
n=1,
dtype=dtype)
with self.assertRaises(RuntimeError):
sampler(posterior)
# resample
sampler = PairwiseSobolQMCNormalSampler(num_samples=4,
resample=True,
collapse_batch_dims=False)
self.assertTrue(sampler.resample)
self.assertFalse(sampler.collapse_batch_dims)
initial_seed = sampler.seed
# check samples non-batched
posterior = _get_test_posterior(device=self.device, dtype=dtype)
samples = sampler(posterior=posterior)
self.assertEqual(samples.shape, torch.Size([4, 3, 2]))
self.assertEqual(sampler.seed, initial_seed + 1)
# ensure samples are not the same
samples2 = sampler(posterior)
self.assertFalse(torch.allclose(samples, samples2))
self.assertEqual(sampler.seed, initial_seed + 2)
# ensure this works with a differeantly shaped posterior
posterior_batched = _get_test_posterior(device=self.device,
dtype=dtype,
batched=True)
samples_batched = sampler(posterior_batched)
self.assertEqual(samples_batched.shape, torch.Size([4, 3, 3, 2]))
self.assertEqual(sampler.seed, initial_seed + 3)
# ensure error is rasied when number of points are < 2
posterior = _get_test_posterior(device=self.device,
n=1,
dtype=dtype)
with self.assertRaises(RuntimeError):
sampler(posterior)
# check max_num_comparisons
sampler = PairwiseSobolQMCNormalSampler(num_samples=4,
max_num_comparisons=2,
collapse_batch_dims=False)
self.assertFalse(sampler.resample)
self.assertFalse(sampler.collapse_batch_dims)
# check samples non-batched
posterior = _get_test_posterior(device=self.device, dtype=dtype)
samples = sampler(posterior)
self.assertEqual(samples.shape, torch.Size([4, 2, 2]))