Skip to content

Draft update to make_event #1153

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

Open
tturocy wants to merge 1 commit into
base: master
Choose a base branch
from
Open

Draft update to make_event #1153

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
20 changes: 13 additions & 7 deletions doc/tutorials/03_stripped_down_poker.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -797,7 +797,13 @@
"cell_type": "markdown",
"id": "dc4522b5",
"metadata": {},
"source": "Here we specified an equal-probability distribution over the three actions, using `pygambit`'s `Rational` class, which is derived from Python's `fractions.Fraction`, to represent the probabilities exactly.\n\nNumerical data can be set as rational numbers.\n`make_event` forms a collection of nodes into an event with a given distribution; applied to a single node which is already an event, it resets the probabilities.\nLet's use it to change the distribution, again with `Rational` numbers:"
"source": [
"Here we specified an equal-probability distribution over the three actions, using `pygambit`'s `Rational` class, which is derived from Python's `fractions.Fraction`, to represent the probabilities exactly.\n",
"\n",
"Numerical data can be set as rational numbers.\n",
"`set_event_actions` sets the actions and probability distribution of an existing event; since the action labels are unchanged here, it simply resets the probabilities.\n",
"Let's use it to change the distribution, again with `Rational` numbers:"
]
},
{
"cell_type": "code",
Expand All @@ -806,7 +812,7 @@
"metadata": {},
"outputs": [],
"source": [
"small_game.make_event(\n",
"small_game.set_event_actions(\n",
" gbt.H.path(),\n",
" {\"a\": gbt.Rational(1, 4), \"b\": gbt.Rational(1, 2), \"c\": gbt.Rational(1, 4)}\n",
")\n",
Expand All @@ -828,7 +834,7 @@
"metadata": {},
"outputs": [],
"source": [
"small_game.make_event(\n",
"small_game.set_event_actions(\n",
" gbt.H.path(),\n",
" {\"a\": gbt.Decimal(\".25\"), \"b\": gbt.Decimal(\".50\"), \"c\": gbt.Decimal(\".25\")}\n",
")\n",
Expand All @@ -854,7 +860,7 @@
"metadata": {},
"outputs": [],
"source": [
"small_game.make_event(gbt.H.path(), {\"a\": \"1/4\", \"b\": \"1/2\", \"c\": \"1/4\"})\n",
"small_game.set_event_actions(gbt.H.path(), {\"a\": \"1/4\", \"b\": \"1/2\", \"c\": \"1/4\"})\n",
"list(small_game.get_action_probs(gbt.H.path()).values())"
]
},
Expand All @@ -865,7 +871,7 @@
"metadata": {},
"outputs": [],
"source": [
"small_game.make_event(gbt.H.path(), {\"a\": \".25\", \"b\": \".50\", \"c\": \".25\"})\n",
"small_game.set_event_actions(gbt.H.path(), {\"a\": \".25\", \"b\": \".50\", \"c\": \".25\"})\n",
"list(small_game.get_action_probs(gbt.H.path()).values())"
]
},
Expand All @@ -889,7 +895,7 @@
"metadata": {},
"outputs": [],
"source": [
"small_game.make_event(gbt.H.path(), {\"a\": .25, \"b\": .50, \"c\": .25})\n",
"small_game.set_event_actions(gbt.H.path(), {\"a\": .25, \"b\": .50, \"c\": .25})\n",
"list(small_game.get_action_probs(gbt.H.path()).values())"
]
},
Expand All @@ -909,7 +915,7 @@
"outputs": [],
"source": [
"try:\n",
" small_game.make_event(gbt.H.path(), {\"a\": 1/3, \"b\": 1/3, \"c\": 1/3})\n",
" small_game.set_event_actions(gbt.H.path(), {\"a\": 1/3, \"b\": 1/3, \"c\": 1/3})\n",
"except ValueError as e:\n",
" print(\"ValueError:\", e)"
]
Expand Down
104 changes: 47 additions & 57 deletions src/pygambit/game.pxi
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -746,7 +746,13 @@ class Game:
f"{history.__class__.__name__}"
)
resolved_node = self._resolve_node(history, "get_action_probs")
infoset_handle: c_GameInfoset = cython.cast(Node, resolved_node)._infoset_handle()
return self._action_probs(cython.cast(Node, resolved_node))

def _action_probs(self, node: Node) -> dict:
"""Return the probability of each action at `node`'s event, keyed by
label, or an empty dict if `node` does not currently belong to a
chance event."""
infoset_handle: c_GameInfoset = node._infoset_handle()
if (
infoset_handle == cython.cast(c_GameInfoset, NULL)
or not infoset_handle.deref().IsChanceInfoset()
Expand Down Expand Up @@ -1892,26 +1898,6 @@ class Game:
f"(the node is terminal)"
)

def _resolve_probs(self,
probs: typing.Sequence | typing.Mapping,
action_labels: list[str],
funcname: str) -> list:
"""Resolve a probability specification against an ordered list of action labels.

`probs` may be a sequence (positional; must have exactly one entry per action)
or a mapping from action labels to values (may be sparse; omitted labels are
assigned zero). Returns a dense list of values in action order.
"""
if isinstance(probs, typing.Mapping):
unknown = [k for k in probs if k not in action_labels]
if unknown:
raise KeyError(f"{funcname}(): no action with label '{unknown[0]}'")
return [probs.get(label, 0) for label in action_labels]
probs = list(probs)
if len(probs) != len(action_labels):
raise IndexError(f"{funcname}(): must specify exactly one probability per action")
return probs

def append_move(self, nodes: Selector | GroupedSelector,
player: str,
actions: list[str]) -> None:
Expand Down Expand Up @@ -2565,37 +2551,33 @@ class Game:

def make_event(self,
nodes: Selector | GroupedSelector,
probs: typing.Mapping,
label: str | None = None) -> None:
"""Form `nodes` into a single event with distribution `probs`.

`nodes` must all be nonterminal nodes of this game with the same actions, with the same
labels in the same order. They need not be chance nodes; personal nodes are
converted, and the move is thereafter resolved by chance. Nodes are removed from
whatever information sets or events they currently belong to; any of those which
retain members survive, keeping their labels, and those left with no members are deleted.
The resulting event's members, actions, and player are accessible via
"""Form `nodes` into a single event.

`nodes` must all be nonterminal nodes of this game currently belonging to a
chance event, with the same actions, with the same labels in the same order,
and the same probability distribution over those actions. Nodes are removed
from whatever events they currently belong to; any of those which retain
members survive, keeping their labels, and those left with no members are
deleted. The resulting event's members, actions, and player are accessible via
``Node.members``/``Node.actions``/``Node.player`` for any node in `nodes`.

To pool nodes whose actions or probabilities do not yet agree, first bring
them into conformance with `set_event_actions`.

`nodes` is a `Selector` (an `H`-built expression, evaluated against this game
and treated as a flat set of nodes) or a `GroupedSelector` (an `H`-built
`.by(...)` expression, whose groups are pooled together into the one event).

Which resolved node is treated as "first", determining the action order of
the event and the frame against which keys of `probs` are resolved, follows
`nodes`' own resolution order.
the event, follows `nodes`' own resolution order.

.. versionadded:: 17.0.0

Parameters
----------
nodes : Selector or GroupedSelector
The nonempty set of nonterminal nodes to place in the event.
probs : Mapping
The probability distribution over the actions of the event, as a mapping
from action label to probability. May be sparse; omitted actions are
assigned probability zero. Probabilities are non-negative and sum to
exactly one.
label : str, optional
The label of the new event. If specified, must be unique among the events
of the game after the operation. A label currently held by another event
Expand All @@ -2604,17 +2586,19 @@ class Game:
Raises
------
TypeError
If `nodes` is not a `Selector` or `GroupedSelector`, or `probs` is not a
mapping.
KeyError
If a key of `probs` matches no action label of the event.
If `nodes` is not a `Selector` or `GroupedSelector`.
UndefinedOperationError
If any of `nodes` is a terminal node, or the game is not a tree.
If any of `nodes` is a terminal node, does not currently belong to a
chance event, or the game is not a tree.
ValueError
If `nodes` is empty or contains a repeated node; if the nodes do not
all have the same actions in the same order; if `probs` are not
non-negative numbers summing to exactly one; or if `label` is not
unique among the game's events after the operation.
all have the same actions in the same order, or the same probability
distribution over those actions; or if `label` is not unique among the
game's events after the operation.

See Also
--------
set_event_actions : Adjust an event's actions and probability distribution.
"""
if not self.is_tree:
raise UndefinedOperationError(
Expand All @@ -2627,30 +2611,36 @@ class Game:
f"make_event(): nodes must be a Selector or GroupedSelector, "
f"not {nodes.__class__.__name__}"
)
if not isinstance(probs, typing.Mapping):
raise TypeError(
f"make_event(): probs must be a mapping, not {probs.__class__.__name__}"
)
resolved_nodes = self._resolve_nodes(nodes, "make_event")
if any(cython.cast(Node, n)._is_terminal() for n in resolved_nodes):
raise UndefinedOperationError(
"make_event(): all nodes must be nonterminal"
)
resolved_node = cython.cast(Node, resolved_nodes[0])
action_labels = list(resolved_node.actions)
if any(list(n.actions) != action_labels
for n in resolved_nodes[1:]):
raise ValueError(
"make_event(): all nodes must have the same actions, "
"with the same labels in the same order"
reference_probs = self._action_probs(resolved_node)
if not reference_probs:
raise UndefinedOperationError(
"make_event(): all nodes must currently belong to a chance event"
)
resolved_probs = self._resolve_probs(probs, action_labels, "make_event")
for n in resolved_nodes[1:]:
node = cython.cast(Node, n)
if list(node.actions) != action_labels:
raise ValueError(
"make_event(): all nodes must have the same actions, "
"with the same labels in the same order"
)
if self._action_probs(node) != reference_probs:
raise ValueError(
"make_event(): all nodes must have the same probability "
"distribution over their actions"
)
c_nodes = stdvector[c_GameNode]()
for n in resolved_nodes:
c_nodes.push_back(cython.cast(Node, n).node)
c_probs = stdvector[c_Number]()
for p in resolved_probs:
c_probs.push_back(_to_number(p))
for label_ in action_labels:
c_probs.push_back(_to_number(reference_probs[label_]))
self.game.deref().MakeEvent(c_nodes, c_probs, (label or "").encode("utf-8"))

def relabel_actions(self,
Expand Down
Loading
Loading