A random walk bandit. More...
Public Member Functions | |
| None | __init__ (self, int k) |
| Construct the class. | |
| select (self, index) | |
| Select one or several arms to obtain a reward from. | |
 Public Member Functions inherited from bandit.normal.Normal | |
| trueValues (self) | |
| Return the distribution parameters for the arms. | |
| Public Member Functions inherited from bandit.base_bandit.BaseBandit | |
| int | k (self) |
| Return the number of arms this bandit has. | |
Additional Inherited Members | |
| Protected Attributes inherited from bandit.normal.Normal | |
| _std | |
| _mean | |
| Protected Attributes inherited from bandit.base_bandit.BaseBandit | |
| _k | |
Detailed Description
A random walk bandit.
This class features k arms with rewards from the arms drawn from normal distributions. The means, when initialized, are drawn from a uniform range of [-1, 1). However, after each call to select, the means for every arm is changed. Each arm's mean is adjusted by a randomly selected value drawn from a normal distribution with mean 0 and standard deviation 0.01. These values are drawn independently for each arm.
Definition at line 5 of file random_walk.py.
Constructor & Destructor Documentation
◆ __init__()
| None bandit.random_walk.RandomWalk.__init__ | ( | self, | |
| int | k | ||
| ) |
Construct the class.
This includes defining the normal distribution parameters for each arm. There is a different distribution for each arm. The means are sampled from the uniform range [-1, 1). The standard deviations are 1.0.
- Parameters
-
k The number of arms this bandit should have. This must be an int greater than 0.
Reimplemented from bandit.normal.Normal.
Definition at line 17 of file random_walk.py.
Member Function Documentation
◆ select()
| bandit.random_walk.RandomWalk.select | ( | self, | |
| index | |||
| ) |
Select one or several arms to obtain a reward from.
- Parameters
-
index Any numpy valid indexing method to select which arms a reward should be drawn from. None can also be passed, but will only return a reward of None.
- Returns
- The rewards. The size of this will depend on the type of index. If a single integer is passed in, a single float will be returned. Otherwise, a numpy array will be returned. If None is passed in, this will also be None.
Reimplemented from bandit.normal.Normal.
Definition at line 20 of file random_walk.py.
The documentation for this class was generated from the following file:
- bandit/random_walk.py
