Space search

orion.algo.space – Objects describing a problem’s domain

There are 3 classes representing possible parameter types. All of them subclass the base class Dimension:

• Real
• Integer
• Categorical

These are instantiated to declare a problem’s parameter space. Oríon registers them in a ordered dictionary, Space, which describes how the parameters should be in order for orion.algo.base.AbstractAlgorithm implementations to communicate with orion.core.

Parameter values recorded in orion.core.worker.trial.Trial objects must be and are in concordance with orion.algo.space objects. These objects will be defined by orion.core using the user script’s configuration file.

Prior distributions, contained in Dimension classes, are based on scipy.stats.distributions and should be configured as noted in the scipy documentation for each specific implentation of a random variable type, unless noted otherwise!

class orion.algo.space.Categorical(name, categories, **kwargs)

Subclass of Dimension for representing categorical parameters.

Attributes:

name : str

See Dimension attributes.

type : str

See Dimension attributes.

prior : scipy.stats.distributions.rv_generic

shape : tuple

Return the shape of dimension.

categories : tuple

A set of unordered stuff to pick out from, except if enum

Methods

cast(self, point)	Cast a point to some category
get_prior_string(self)	Build the string corresponding to current prior
get_string(self)	Build the string corresponding to current dimension
interval(self[, alpha])	Return a tuple of possible values that this categorical dimension can take.
sample(self[, n_samples, seed])	Draw random samples from prior.
validate(self)	Validate dimension arguments

cast(self, point)

Cast a point to some category

Casted point will stay a list or a numpy array depending on the given point’s type.

Raises:

ValueError

If one of the category in point is not present in current Categorical Dimension.

get_prior_string(self)

Build the string corresponding to current prior

interval(self, alpha=1.0)

Return a tuple of possible values that this categorical dimension can take.

Warning

This method makes no sense for categorical variables. Use self.categories instead.

sample(self, n_samples=1, seed=None)

Draw random samples from prior.

See also

Dimension.sample

class orion.algo.space.Dimension(name, prior, *args, **kwargs)

Base class for search space dimensions.

Attributes:

name : str

See Dimension attributes.

type : str

See Dimension attributes.

prior : scipy.stats.distributions.rv_generic

A distribution over the original dimension.

shape : tuple

Return the shape of dimension.

Methods

cast(self, point)	Cast a point to dimension’s type
get_prior_string(self)	Build the string corresponding to current prior
get_string(self)	Build the string corresponding to current dimension
interval(self[, alpha])	Return a tuple containing lower and upper bound for parameters.
sample(self[, n_samples, seed])	Draw random samples from prior.
validate(self)	Validate dimension arguments

cast(self, point)

Cast a point to dimension’s type

If casted point will stay a list or a numpy array depending on the given point’s type.

default_value

Return the default value for this dimensions

get_prior_string(self)

Build the string corresponding to current prior

get_string(self)

Build the string corresponding to current dimension

interval(self, alpha=1.0)

Return a tuple containing lower and upper bound for parameters.

If parameters are drawn from an ‘open’ supported random variable, then it will be attempted to calculate the interval from which a variable is alpha-likely to be drawn from.

Note

Lower bound is inclusive, upper bound is exclusive.

name

See Dimension attributes.

sample(self, n_samples=1, seed=None)

Draw random samples from prior.

Parameters:

n_samples : int, optional

The number of samples to be drawn. Default is 1 sample.

seed : None | int | numpy.random.RandomState instance, optional

This parameter defines the RandomState object to use for drawing random variates. If None (or np.random), the global np.random state is used. If integer, it is used to seed a RandomState instance just for the call of this function. Default is None.

Set random state to something other than None for reproducible results.

.. warning:: Setting `seed` with an integer will cause the same ndarray

to be sampled if n_samples > 0. Set seed with a numpy.random.RandomState to carry on the changes in random state across many samples.

shape

Return the shape of dimension.

type

See Dimension attributes.

validate(self)

Validate dimension arguments

class orion.algo.space.Fidelity(name, low, high, base=2)

Fidelity Dimension for representing multi-fidelity.

Fidelity dimensions are not optimized by the algorithms. If it supports multi-fidelity, the algorithm will select a fidelity level for which it will sample hyper-parameter values to explore a low fidelity space. This class is used as a place-holder so that algorithms can discern fidelity dimensions from hyper-parameter dimensions.

Parameters:

name : str

Name of the dimension

low: int

Mininum of the fidelity interval.

high: int

Maximum of the fidelity interval.

base: int

Base logarithm of the fidelity dimension.

Attributes:

name : str

See Dimension attributes.

default_value: int

Maximum of the fidelity interval.

Methods

cast(self[, point])	Do not do anything.
get_prior_string(self)	Build the string corresponding to current prior
get_string(self)	Build the string corresponding to current dimension
interval(self[, alpha])	Do not do anything.
sample(self[, n_samples, seed])	Do not do anything.
validate(self)	Do not do anything.

cast(self, point=0)

Do not do anything.

default_value

Return high

get_prior_string(self)

Build the string corresponding to current prior

interval(self, alpha=1.0)

Do not do anything.

sample(self, n_samples=1, seed=None)

Do not do anything.

validate(self)

Do not do anything.

class orion.algo.space.Integer(name, prior, *args, **kwargs)

Subclass of Dimension for representing integer parameters.

Attributes:

name : str

See Dimension attributes.

type : str

See Dimension attributes.

prior : scipy.stats.distributions.rv_generic

shape : tuple

Return the shape of dimension.

Methods

cast(self, point)	Cast a point to int
get_prior_string(self)	Build the string corresponding to current prior
get_string(self)	Build the string corresponding to current dimension
interval(self[, alpha])	Return a tuple containing lower and upper bound for parameters.
sample(self[, n_samples, seed])	Draw random samples from prior.
validate(self)	Validate dimension arguments

cast(self, point)

Cast a point to int

If casted point will stay a list or a numpy array depending on the given point’s type.

class orion.algo.space.Real(name, prior, *args, **kwargs)

Subclass of Dimension for representing real parameters.

Attributes:

name : str

See Dimension attributes.

type : str

See Dimension attributes.

prior : scipy.stats.distributions.rv_generic

shape : tuple

Return the shape of dimension.

low : float

Constrain with a lower bound (inclusive), default -numpy.inf.

high : float

Constrain with an upper bound (exclusive), default numpy.inf.

Methods

cast(self, point)	Cast a point to float
get_prior_string(self)	Build the string corresponding to current prior
get_string(self)	Build the string corresponding to current dimension
interval(self[, alpha])	Return a tuple containing lower and upper bound for parameters.
sample(self[, n_samples, seed])	Draw random samples from prior.
validate(self)	Validate dimension arguments

cast(self, point)

Cast a point to float

If casted point will stay a list or a numpy array depending on the given point’s type.

interval(self, alpha=1.0)

Return a tuple containing lower and upper bound for parameters.

If parameters are drawn from an ‘open’ supported random variable, then it will be attempted to calculate the interval from which a variable is alpha-likely to be drawn from.

Note

Lower bound is inclusive, upper bound is exclusive.

sample(self, n_samples=1, seed=None)

Draw random samples from prior.

See also

Dimension.sample

class orion.algo.space.Space

Represents the search space.

It is a sorted dictionary which contains Dimension objects. The dimensions are sorted based on their names.

Methods

clear()
contains	alias of Dimension
copy()
fromkeys(iterable[, value])	Create a new dictionary with keys from iterable and values set to value.
get(self, key[, default])	Return the value for key if key is in the dictionary, else default.
interval(self[, alpha])	Return a list with the intervals for each contained dimension.
items(self)	Return items sorted according to keys
keys(self)	Return sorted keys
pop()	If key is not found, d is returned if given, otherwise KeyError is raised
popitem()	2-tuple; but raise KeyError if D is empty.
register(self, dimension)	Register a new dimension to Space.
sample(self[, n_samples, seed])	Draw random samples from this space.
setdefault(self, key[, default])	Insert key with a value of default if key is not in the dictionary.
update()	If E is present and has a .keys() method, then does: for k in E: D[k] = E[k] If E is present and lacks a .keys() method, then does: for k, v in E: D[k] = v In either case, this is followed by: for k in F: D[k] = F[k]
values(self)	Return values sorted according to keys

contains

alias of Dimension

interval(self, alpha=1.0)

Return a list with the intervals for each contained dimension.

Note

Lower bound is inclusive, upper bound is exclusive.

items(self)

Return items sorted according to keys

keys(self)

Return sorted keys

register(self, dimension)

Register a new dimension to Space.

sample(self, n_samples=1, seed=None)

Draw random samples from this space.

Parameters:

n_samples : int, optional

The number of samples to be drawn. Default is 1 sample.

seed : None | int | numpy.random.RandomState instance, optional

This parameter defines the RandomState object to use for drawing random variates. If None (or np.random), the global np.random state is used. If integer, it is used to seed a RandomState instance just for the call of this function. Default is None.

Set random state to something other than None for reproducible results.

Returns:

points : list of tuples of array-likes

Each element is a separate sample of this space, a list containing values associated with the corresponding dimension. Values are in the same order as the contained dimensions. Their shape is determined by dimension.shape.

values(self)

Return values sorted according to keys

orion.algo.space.check_random_state(seed)

Return numpy global rng or RandomState if seed is specified

orion.algo.space.pack_point(point, space)

Take a list of points and pack it appropriately as a point from space.

This function is deprecated and will be removed in v0.2.0. Use orion.core.utils.points.regroup_dims instead.

orion.algo.space.unpack_point(point, space)

Flatten point in space and convert it to a 1D numpy.ndarray.

This function is deprecated and will be removed in v0.2.0. Use orion.core.utils.points.flatten_dims instead.