ANL-CEEESA/MIPLearn
1
0
Fork 0
mirror of https://github.com/ANL-CEEESA/MIPLearn.git synced 2025-12-06 01:18:52 -06:00
Code Issues Packages Projects Releases Wiki Activity

Parallel processing

Browse Source
This commit is contained in:
Alinson S. Xavier
2021-04-13 19:28:18 -05:00
parent bec7dae6d9
commit 77b10b9609
9 changed files with 45 additions and 25 deletions
Download Patch File Download Diff File Expand all files Collapse all files

31
miplearn/components/component.py
View File

@@ -11,6 +11,8 @@ from miplearn.features import Sample
from miplearn.instance.base import Instance
from miplearn.types import LearningSolveStats
from p_tqdm import p_umap
if TYPE_CHECKING:
from miplearn.solvers.learning import LearningSolver
@@ -159,7 +161,6 @@ class Component(EnforceOverrides):
self,
instance: Optional[Instance],
sample: Sample,
pre: Optional[List[Any]] = None,
) -> Tuple[Dict, Dict]:
"""
Returns a pair of x and y dictionaries containing, respectively, the matrices
@@ -168,6 +169,9 @@ class Component(EnforceOverrides):
"""
pass
def pre_fit(self, pre: List[Any]):
pass
def user_cut_cb(
self,
solver: "LearningSolver",
@@ -183,6 +187,7 @@ class Component(EnforceOverrides):
def fit_multiple(
components: Dict[str, "Component"],
instances: List[Instance],
n_jobs: int = 1,
) -> None:
def _pre_sample_xy(instance: Instance) -> Dict:
pre_instance: Dict = {}
@@ -195,7 +200,17 @@ class Component(EnforceOverrides):
instance.free()
return pre_instance
def _sample_xy(instance: Instance, pre: Dict) -> Tuple[Dict, Dict]:
pre = p_umap(_pre_sample_xy, instances, num_cpus=n_jobs)
pre_combined: Dict = {}
for (cname, comp) in components.items():
pre_combined[cname] = []
for p in pre:
pre_combined[cname].extend(p[cname])
for (cname, comp) in components.items():
comp.pre_fit(pre_combined[cname])
def _sample_xy(instance: Instance) -> Tuple[Dict, Dict]:
x_instance: Dict = {}
y_instance: Dict = {}
for (cname, comp) in components.items():
@@ -206,7 +221,7 @@ class Component(EnforceOverrides):
for (cname, comp) in components.items():
x = x_instance[cname]
y = y_instance[cname]
x_sample, y_sample = comp.sample_xy(instance, sample, pre[cname])
x_sample, y_sample = comp.sample_xy(instance, sample)
for cat in x_sample.keys():
if cat not in x:
x[cat] = []
@@ -216,15 +231,7 @@ class Component(EnforceOverrides):
instance.free()
return x_instance, y_instance
pre = [_pre_sample_xy(instance) for instance in instances]
pre_combined: Dict = {}
for (cname, comp) in components.items():
pre_combined[cname] = []
for p in pre:
pre_combined[cname].extend(p[cname])
xy = [_sample_xy(instances, pre_combined) for instances in instances]
xy = p_umap(_sample_xy, instances)
for (cname, comp) in components.items():
x_comp: Dict = {}

8
miplearn/components/dynamic_common.py
View File

@@ -89,16 +89,18 @@ class DynamicConstraintsComponent(Component):
self,
instance: Optional[Instance],
sample: Sample,
pre: Optional[List[Any]] = None,
) -> Tuple[Dict, Dict]:
x, y, _ = self.sample_xy_with_cids(instance, sample)
return x, y
@overrides
def pre_fit(self, pre: List[Any]) -> None:
assert pre is not None
known_cids: Set = set()
for cids in pre:
known_cids |= cids
self.known_cids.clear()
self.known_cids.extend(sorted(known_cids))
x, y, _ = self.sample_xy_with_cids(instance, sample)
return x, y
def sample_predict(
self,

7
miplearn/components/dynamic_lazy.py
View File

@@ -108,9 +108,12 @@ class DynamicLazyConstraintsComponent(Component):
self,
instance: Optional[Instance],
sample: Sample,
pre: Optional[List[Any]] = None,
) -> Tuple[Dict, Dict]:
return self.dynamic.sample_xy(instance, sample, pre=pre)
return self.dynamic.sample_xy(instance, sample)
@overrides
def pre_fit(self, pre: List[Any]) -> None:
self.dynamic.pre_fit(pre)
def sample_predict(
self,

7
miplearn/components/dynamic_user_cuts.py
View File

@@ -101,9 +101,12 @@ class UserCutsComponent(Component):
self,
instance: "Instance",
sample: Sample,
pre: Optional[List[Any]] = None,
) -> Tuple[Dict, Dict]:
return self.dynamic.sample_xy(instance, sample, pre=pre)
return self.dynamic.sample_xy(instance, sample)
@overrides
def pre_fit(self, pre: List[Any]) -> None:
self.dynamic.pre_fit(pre)
def sample_predict(
self,

1
miplearn/components/objective.py
View File

@@ -76,7 +76,6 @@ class ObjectiveValueComponent(Component):
self,
_: Optional[Instance],
sample: Sample,
pre: Optional[List[Any]] = None,
) -> Tuple[Dict[Hashable, List[List[float]]], Dict[Hashable, List[List[float]]]]:
# Instance features
assert sample.after_load is not None

1
miplearn/components/primal.py
View File

@@ -145,7 +145,6 @@ class PrimalSolutionComponent(Component):
self,
_: Optional[Instance],
sample: Sample,
pre: Optional[List[Any]] = None,
) -> Tuple[Dict[Category, List[List[float]]], Dict[Category, List[List[float]]]]:
x: Dict = {}
y: Dict = {}

1
miplearn/components/static_lazy.py
View File

@@ -154,7 +154,6 @@ class StaticLazyConstraintsComponent(Component):
self,
_: Optional[Instance],
sample: Sample,
pre: Optional[List[Any]] = None,
) -> Tuple[Dict[Hashable, List[List[float]]], Dict[Hashable, List[List[float]]]]:
x, y, __ = self._sample_xy_with_cids(sample)
return x, y

12
miplearn/solvers/learning.py
View File

@@ -391,11 +391,19 @@ class LearningSolver:
self._restore_miplearn_logger()
return stats
def fit(self, training_instances: List[Instance]) -> None:
def fit(
self,
training_instances: List[Instance],
n_jobs: int = 1,
) -> None:
if len(training_instances) == 0:
logger.warning("Empty list of training instances provided. Skipping.")
return
Component.fit_multiple(self.components, training_instances)
Component.fit_multiple(
self.components,
training_instances,
n_jobs=n_jobs,
)
def _add_component(self, component: Component) -> None:
name = component.__class__.__name__

2
tests/components/test_dynamic_lazy.py
View File

@@ -87,6 +87,7 @@ def training_instances() -> List[Instance]:
def test_sample_xy(training_instances: List[Instance]) -> None:
comp = DynamicLazyConstraintsComponent()
comp.pre_fit([{"c1", "c2", "c3", "c4"}])
x_expected = {
"type-a": [[5.0, 1.0, 2.0, 3.0], [5.0, 4.0, 5.0, 6.0]],
"type-b": [[5.0, 1.0, 2.0], [5.0, 3.0, 4.0]],
@@ -98,7 +99,6 @@ def test_sample_xy(training_instances: List[Instance]) -> None:
x_actual, y_actual = comp.sample_xy(
training_instances[0],
training_instances[0].samples[0],
pre=[{"c1", "c2", "c3", "c4"}],
)
assert_equals(x_actual, x_expected)
assert_equals(y_actual, y_expected)