Add dynamic_args_provider to Runner

Author: basnijholt

adaptive/runner.py

@@ -146,6 +146,12 @@ class BaseRunner(metaclass=abc.ABCMeta):
         the point is present in ``runner.failed``.
     raise_if_retries_exceeded : bool, default: True
         Raise the error after a point ``x`` failed `retries`.
+    dynamic_args_provider : callable, optional
+        A callable that takes the learner as its sole argument and returns additional
+        arguments to pass to the function being learned. This allows you to dynamically
+        adjust parameters of the function based on the current state of the learner.
+        If provided, the function will be called as `function(x, dynamic_args_provider(learner))`
+        instead of just `function(x)`.
     allow_running_forever : bool, default: False
         Allow the runner to run forever when the goal is None.
 
@@ -188,6 +194,7 @@ def __init__(
         shutdown_executor: bool = False,
         retries: int = 0,
         raise_if_retries_exceeded: bool = True,
+        dynamic_args_provider: Callable[[LearnerType], Any] | None = None,
         allow_running_forever: bool = False,
     ):
         self.executor = _ensure_executor(executor)
@@ -228,6 +235,8 @@ def __init__(
             next, itertools.count()
         )  # some unique id to be associated with each point
 
+        self.dynamic_args_provider = dynamic_args_provider
+
     def _get_max_tasks(self) -> int:
         return self._max_tasks or _get_ncores(self.executor)
 
@@ -432,6 +441,12 @@ class BlockingRunner(BaseRunner):
         the point is present in ``runner.failed``.
     raise_if_retries_exceeded : bool, default: True
         Raise the error after a point ``x`` failed `retries`.
+    dynamic_args_provider : callable, optional
+        A callable that takes the learner as its sole argument and returns additional
+        arguments to pass to the function being learned. This allows you to dynamically
+        adjust parameters of the function based on the current state of the learner.
+        If provided, the function will be called as `function(x, dynamic_args_provider(learner))`
+        instead of just `function(x)`.
 
     Attributes
     ----------
@@ -476,6 +491,7 @@ def __init__(
         shutdown_executor: bool = False,
         retries: int = 0,
         raise_if_retries_exceeded: bool = True,
+        dynamic_args_provider: Callable[[LearnerType], Any] | None = None,
     ) -> None:
         if inspect.iscoroutinefunction(learner.function):
             raise ValueError("Coroutine functions can only be used with 'AsyncRunner'.")
@@ -497,7 +513,14 @@ def __init__(
         self._run()
 
     def _submit(self, x: tuple[float, ...] | float | int) -> FutureTypes:
-        return self.executor.submit(self.learner.function, x)
+        args = (
+            (x,)
+            if not self.dynamic_args_provider
+            else (x, self.dynamic_args_provider(self.learner))
+        )
+        if self.dynamic_args_provider:
+            return self.executor.submit(self.learner.function, *args)
+        return self.executor.submit(self.learner.function, *args)
 
     def _run(self) -> None:
         first_completed = concurrent.FIRST_COMPLETED
@@ -582,8 +605,12 @@ class AsyncRunner(BaseRunner):
         the point is present in ``runner.failed``.
     raise_if_retries_exceeded : bool, default: True
         Raise the error after a point ``x`` failed `retries`.
-    allow_running_forever : bool, default: True
-        If True, the runner will run forever if the goal is not provided.
+    dynamic_args_provider : callable, optional
+        A callable that takes the learner as its sole argument and returns additional
+        arguments to pass to the function being learned. This allows you to dynamically
+        adjust parameters of the function based on the current state of the learner.
+        If provided, the function will be called as `function(x, dynamic_args_provider(learner))`
+        instead of just `function(x)`.
 
     Attributes
     ----------
@@ -636,6 +663,7 @@ def __init__(
         ioloop=None,
         retries: int = 0,
         raise_if_retries_exceeded: bool = True,
+        dynamic_args_provider: Callable[[LearnerType], Any] | None = None,
     ) -> None:
         if (
             executor is None
@@ -666,6 +694,7 @@ def __init__(
             shutdown_executor=shutdown_executor,
             retries=retries,
             raise_if_retries_exceeded=raise_if_retries_exceeded,
+            dynamic_args_provider=dynamic_args_provider,
             allow_running_forever=True,
         )
         self.ioloop = ioloop or asyncio.get_event_loop()
@@ -694,10 +723,15 @@ def __init__(
 
     def _submit(self, x: Any) -> asyncio.Task | asyncio.Future:
         ioloop = self.ioloop
+        args = (
+            (x,)
+            if not self.dynamic_args_provider
+            else (x, self.dynamic_args_provider(self.learner))
+        )
         if inspect.iscoroutinefunction(self.learner.function):
-            return ioloop.create_task(self.learner.function(x))
+            return ioloop.create_task(self.learner.function(*args))
         else:
-            return ioloop.run_in_executor(self.executor, self.learner.function, x)
+            return ioloop.run_in_executor(self.executor, self.learner.function, *args)
 
     def status(self) -> str:
         """Return the runner status as a string.