API Cheatsheet

We list all the important (abstract) functions from the base classes as follows.

Algorithm

PyXAB.algos.Algo.Algorithm.pull(): Generate a point for each time step to be evaluated
PyXAB.algos.Algo.Algorithm.receive_reward(): After receiving the reward, update the parameters of the algorithm
PyXAB.algos.Algo.Algorithm.get_last_point(): The function to retrieve the last output of the algorithm

Partition

PyXAB.partition.Partition.Partition.make_children(): Make children for one node

Objective

PyXAB.synthetic_obj.Objective.Objective.f(): Evaluate the point and return the reward (stochastic or deterministic)

Note

The general base classes and the implemented functions are listed as follows

PyXAB.algos.Algo.Algorithm

Base class for all X-armed Bandit algorithms

class PyXAB.algos.Algo.Algorithm

Bases: abc.ABC

Abstract class for X-armed bandit algorithms.

abstract __init__(): Initialization for the algorithm

abstract get_last_point()

Every algorithm needs a function to get the last point

Returns: chosen_point – The point chosen by the algorithm
Return type: list

abstract pull(time)

Every algorithm needs a function to pull a node.

Parameters: time (int) – The time step of the online process.
Returns: chosen_point – The point chosen by the algorithm
Return type: list

abstract receive_reward(time, reward)

Every algorithm needs a function to receive the reward.

Parameters

time (int) – The time step of the online process.
reward (float) – The (Stochastic) reward of the pulled point

PyXAB.synthetic_obj.Objective.Objective

Base class for any objective

class PyXAB.synthetic_obj.Objective.Objective

Bases: abc.ABC

Abstract class for general blackbox objectives

abstract f(x)

Evaluation of the chosen point

Parameters: x (list) – one input point in the form of x = [x_1, x_2, … x_d]
Returns: y – Evaluated value of the function at the particular point
Return type: float

PyXAB.partition.Partition.Partition

Base class for any partition

class PyXAB.partition.Partition.Partition(domain, node=<class 'PyXAB.partition.Node.P_node'>)

Bases: abc.ABC

Abstract class for partition of the parameter domain

__init__(domain, node=<class 'PyXAB.partition.Node.P_node'>)

Initialization of the partition

Parameters

domain (list(list)) – The domain of the objective function to be optimized, should be in the form of list of lists (hypercubes), i.e., [[range1], [range2], … [range_d]], where [range_i] is a list indicating the domain’s projection on the i-th dimension, e.g., [-1, 1]
node – The node used in the partition, with the default choice to be P_node.

deepen(): The function to deepen the partition by one layer by making children to every node in the last layer

get_depth()

The function to get the depth of the partition

Returns: depth – The depth of the partition
Return type: int

get_layer_node_list(depth)

The function to get the all the nodes on the specified depth

Parameters: depth (int) – The depth of the layer in the partition
Returns: self.node_list – The list of nodes on the specified depth
Return type: list

get_node_list()

The function to get the list all nodes in the partition

Returns: self.node_list – The list of all nodes
Return type: list

get_root()

The function to get the root of the partition

Returns: The root node of the partition
Return type: self.root

abstract make_children(parent, newlayer=False)

The function to make children for the parent node

Parameters

parent – The parent node to be expanded into children nodes
newlayer (bool) – Boolean variable that indicates whether or not a new layer is created

PyXAB.partition.Node.P_node

Base class for any node inside a partition

class PyXAB.partition.Node.P_node(depth, index, parent, domain)

Bases: object

The most basic node class that contains everything needed to be inside a partition

__init__(depth, index, parent, domain)

Initialization of the P_node class

Parameters

depth (int) – The depth of the node
index (int) – The index of the node
parent – The parent node of the P_node
domain (list(list)) – The domain that this P_node represents

get_children(): The function to get the children of the node

get_cpoint(): The function to get the center point of the domain

get_depth(): The function to get the depth of the node

get_domain(): The function to get the domain of the node

get_index(): The function to get the index of the node

get_parent(): The function to get the parent of the node

update_children(children)

The function to update the children of the node

Parameters: children – The children nodes to be updated