numba 中的 N 叉树

Question

我需要实现一个树结构，其中每个节点都有任意数量的子节点。当孩子的数量已知时（链表、二叉树等），这似乎很简单，但我还没有设法在更一般的情况下实现它。

例如，我尝试通过将next 参数更改为列表来扩展此linked list example。但是，似乎没有deferred_type 的列表。关于如何实现这一点的任何想法？

from collections import OrderedDict 
from numba import njit, jitclass, types, int32, deferred_type, optional
import numpy as np 

node_type = deferred_type()

spec = OrderedDict() 
spec['data'] = int32 
spec['next'] = optional(types.ListType(node_type))


@jitclass(spec) 
class LinkedNode:
    def __init__(self, data, next):
        self.data = data
        self.next = next

    def prepend(self, data):
        return LinkedNode(data, self)


@njit def make_linked_node(data):
    return LinkedNode(data, None)


node_type.define(LinkedNode.class_type.instance_type)

Answer 1

我找到了一个替代方案，虽然不是我想要的，但仍然可以完成工作。它基本上是二叉树的标准扩展，在左右节点之间包含更多节点（参见this article）。

这些结构通常只跟踪第一个（左）子节点，但我也有一个指向最后一个（右）子节点的指针，因为它有助于仅使用父节点来排列所有边。

from collections import OrderedDict
from numba import jitclass, int64, deferred_type, optional

node_type = deferred_type()

spec = OrderedDict()
spec['id'] = int64
spec['parent'] = optional(node_type)
spec['left_child'] = optional(node_type)  # first child
spec['right_child'] = optional(node_type) # last child
spec['sibling'] = optional(node_type)  # next sibling


@jitclass(spec)
class Node:
    def __init__(self, id, parent):
        self.id = id
        self.parent = parent
        # initialize parent and left right children as None
        self.left_child = None
        self.right_child = None
        self.sibling = None
        if parent is not None:
            parent.add_child(self)

    @property
    def children(self):
    """ A list with all children. """
        children = []
        child = self.left_child
        while child is not None:
            children.append(child)
            child = child.sibling
        return children

    def add_sibling(self, sibling):
        self.sibling = sibling

    def add_child(self, child):
        # if parent has no children
        if self.left_child is None:
            # this node is it first child
            self.left_child = child
            self.right_child = child
        else:
            # the last (now right) child will have this node as sibling
            self.right_child.add_sibling(child)
            self.right_child = child


node_type.define(Node.class_type.instance_type)