Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Update polytope sampling code and add thinning capability #2358

Closed
wants to merge 1 commit into from
Closed

Update polytope sampling code and add thinning capability #2358

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
19 changes: 10 additions & 9 deletions botorch/optim/initializers.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -180,7 +180,7 @@ def sample_q_batches_from_polytope(
q: int,
bounds: Tensor,
n_burnin: int,
thinning: int,
n_thinning: int,
seed: int,
inequality_constraints: Optional[List[Tuple[Tensor, Tensor, float]]] = None,
equality_constraints: Optional[List[Tuple[Tensor, Tensor, float]]] = None,
Expand All @@ -192,8 +192,8 @@ def sample_q_batches_from_polytope(
q: Number of samples per q-batch
bounds: A `2 x d` tensor of lower and upper bounds for each column of `X`.
n_burnin: The number of burn-in samples for the Markov chain sampler.
thinning: The amount of thinning (number of steps to take between
returning samples).
n_thinning: The amount of thinning. The sampler will return every
`n_thinning` sample (after burn-in).
seed: The random seed.
inequality_constraints: A list of tuples (indices, coefficients, rhs),
with each tuple encoding an inequality constraint of the form
Expand Down Expand Up @@ -225,7 +225,7 @@ def sample_q_batches_from_polytope(
),
seed=seed,
n_burnin=n_burnin,
thinning=thinning * q,
n_thinning=n_thinning * q,
)
else:
samples = get_polytope_samples(
Expand All @@ -235,7 +235,7 @@ def sample_q_batches_from_polytope(
equality_constraints=equality_constraints,
seed=seed,
n_burnin=n_burnin,
thinning=thinning,
n_thinning=n_thinning,
)
return samples.view(n, q, -1).cpu()

Expand All @@ -250,7 +250,7 @@ def gen_batch_initial_conditions(
options: Optional[Dict[str, Union[bool, float, int]]] = None,
inequality_constraints: Optional[List[Tuple[Tensor, Tensor, float]]] = None,
equality_constraints: Optional[List[Tuple[Tensor, Tensor, float]]] = None,
generator: Optional[Callable[[int, int, int], Tensor]] = None,
generator: Optional[Callable[[int, int, Optional[int]], Tensor]] = None,
fixed_X_fantasies: Optional[Tensor] = None,
) -> Tensor:
r"""Generate a batch of initial conditions for random-restart optimziation.
Expand Down Expand Up @@ -283,8 +283,8 @@ def gen_batch_initial_conditions(
with each tuple encoding an inequality constraint of the form
`\sum_i (X[indices[i]] * coefficients[i]) = rhs`.
generator: Callable for generating samples that are then further
processed. It receives `n`, `q` and `seed` as arguments and
returns a tensor of shape `n x q x d`.
processed. It receives `n`, `q` and `seed` as arguments
and returns a tensor of shape `n x q x d`.
fixed_X_fantasies: A fixed set of fantasy points to concatenate to
the `q` candidates being initialized along the `-2` dimension. The
shape should be `num_pseudo_points x d`. E.g., this should be
Expand Down Expand Up @@ -343,6 +343,7 @@ def gen_batch_initial_conditions(
f"Sample dimension q*d={effective_dim} exceeding Sobol max dimension "
f"({SobolEngine.MAXDIM}). Using iid samples instead.",
SamplingWarning,
stacklevel=3,
)

while factor < max_factor:
Expand All @@ -367,7 +368,7 @@ def gen_batch_initial_conditions(
q=q,
bounds=bounds,
n_burnin=options.get("n_burnin", 10000),
thinning=options.get("thinning", 32),
n_thinning=options.get("n_thinning", 32),
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If someone passes in thinning, I am guessing it'll get silently ignored. Would something like this offer better backwards compatibility? Ideally we'd just error out if we got some unsupported option cc @esantorella

Suggested change
n_thinning=options.get("n_thinning", 32),
n_thinning=options.get("n_thinning", options.get("thinning", 32)),

seed=seed,
equality_constraints=equality_constraints,
inequality_constraints=inequality_constraints,
Expand Down
Loading