Skip to content

Local search: limit duplicate neighbor evaluation #954 #1299

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

Merged
mwever merged 2 commits into from
Mar 4, 2026
Merged

Local search: limit duplicate neighbor evaluation #954 #1299

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
7318ff4
Local search: prevent duplicate neighbor evaluation for discrete hype…
Feb 23, 2026
1c042b8
Merge branch 'development' into feature/limit-duplicates-ls-954
mwever Mar 4, 2026
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
52 changes: 50 additions & 2 deletions smac/acquisition/maximizer/local_search.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,6 +9,11 @@
import numpy as np
from ConfigSpace import Configuration, ConfigurationSpace
from ConfigSpace.exceptions import ForbiddenValueError
from ConfigSpace.hyperparameters import (
CategoricalHyperparameter,
OrdinalHyperparameter,
UniformIntegerHyperparameter,
)
from joblib import Parallel, delayed

from smac.acquisition.function import AbstractAcquisitionFunction
Expand Down Expand Up @@ -311,6 +316,8 @@ def _search(
if isinstance(start_points, Configuration):
start_points = [start_points]

hp_names = list(start_points[0].config_space.keys())

candidates = start_points
# Compute the acquisition value of the candidates
num_candidates = len(candidates)
Expand Down Expand Up @@ -359,6 +366,20 @@ def _search(
# Keeping track of configurations with equal acquisition value for plateau walking
neighbors_w_equal_acq: list[list[Configuration]] = [[] for _ in range(num_candidates)]

# Track already visited hyperparameter values per local search.
# Only categorical, ordinal, and small discrete integer hyperparameter are tracked.
visited_values = []
for i in range(num_candidates):
vv: dict[str, set[Any]] = {}
for hp in start_points[i].config_space.values():
if (
isinstance(hp, CategoricalHyperparameter)
or isinstance(hp, OrdinalHyperparameter)
or (isinstance(hp, UniformIntegerHyperparameter) and hp.size <= 100)
):
vv[hp.name] = set()
visited_values.append(vv)

num_iters = 0
while np.any(active):

Expand All @@ -373,6 +394,9 @@ def _search(
# Used to request a new neighborhood for the candidates of the i-th local search
new_neighborhood = [False] * num_candidates

# How many neighbors were actually obtained this iteration
actually_obtained = [obtain_n[i] for i in range(num_candidates)]

# gather all neighbors
neighbors = []
for i, neighborhood_iterator in enumerate(neighborhood_iterators):
Expand All @@ -381,6 +405,23 @@ def _search(
for j in range(obtain_n[i]):
try:
n = next(neighborhood_iterator)

# Find out in what hyperparameter neighbor differs from candidate.
diff = n.get_array() != candidates[i].get_array()
changed_hp_idx = np.flatnonzero(diff)
if len(changed_hp_idx) != 1:
continue
changed_hp_name = hp_names[changed_hp_idx[0]]
# The value of differing hyperparameter.
value = n[changed_hp_name]

# Skip neighbors that modify a hyperparameter to a value
# that has already been explored in this local search.
if changed_hp_name in visited_values[i]:
if value in visited_values[i][changed_hp_name]:
continue
visited_values[i][changed_hp_name].add(value)

neighbors_generated[i] += 1
neighbors_for_i.append(n)
except ValueError as e:
Expand All @@ -390,8 +431,9 @@ def _search(
new_neighborhood[i] = True
except StopIteration:
new_neighborhood[i] = True
obtain_n[i] = len(neighbors_for_i)
break
obtain_n[i] = len(neighbors_for_i)
actually_obtained[i] = len(neighbors_for_i)
neighbors.extend(neighbors_for_i)

if len(neighbors) != 0:
Expand All @@ -410,7 +452,7 @@ def _search(
continue

# And for each local search we know how many neighbors we obtained
for j in range(obtain_n[i]):
for j in range(actually_obtained[i]):
# The next line is only true if there was an improvement and we basically need to iterate to
# the i+1-th local search
if improved[i]:
Expand Down Expand Up @@ -441,6 +483,9 @@ def _search(
local_search_steps[i] += 1
neighbors_w_equal_acq[i] = []
obtain_n[i] = 1
# Reset visited values, as we now evaluate a new candidate.
for s in visited_values[i].values():
s.clear()
# Found an equally well performing configuration, keeping it for plateau walking
elif acq_val[acq_index] == acq_val_candidates[i]:
neighbors_w_equal_acq[i].append(neighbors[acq_index])
Expand All @@ -465,6 +510,9 @@ def _search(
if len(neighbors_w_equal_acq[i]) != 0:
candidates[i] = neighbors_w_equal_acq[i][0]
neighbors_w_equal_acq[i] = []
# Reset visited values, as we now evaluate a new candidate.
for s in visited_values[i].values():
s.clear()
n_no_plateau_walk[i] += 1
if n_no_plateau_walk[i] >= self._n_steps_plateau_walk:
active[i] = False
Expand Down
Loading