gemseo / algos

design_space module

Design space

A design space is used to represent the optimization’s unknowns, a.k.a. design variables. A DesignSpace describes this design space at a given state, in terms of names, sizes, types, bounds and current values of the design variables. Variables can easily be added to the DesignSpace using the DesignSpace.add_variable() method or removed using the DesignSpace.remove_variable() method. We can also filter the design variables using the DesignSpace.filter() method. Getters and setters are also available to get or set the value of a given variable property. Lastly, an instance of DesignSpace can be stored in a txt or HDF file.

class gemseo.algos.design_space.DesignSpace(hdf_file=None)

Bases: object

Class that describes the design space at a given state: the names/sizes/types/bounds of the variables and the initial solution of the optimization problem

Constructor

AVAILABLE_TYPES = ['float', 'integer']

DESIGN_SPACE_GROUP = 'design_space'

FLOAT = 'float'

INTEGER = 'integer'

LB_GROUP = 'l_b'

MINIMAL_FIELDS = ['name', 'lower_bound', 'upper_bound']

NAMES_GROUP = 'names'

SEP = '!'

SIZE_GROUP = 'size'

TABLE_NAMES = ['name', 'lower_bound', 'value', 'upper_bound', 'type']

UB_GROUP = 'u_b'

VALUE_GROUP = 'value'

VAR_TYPE_GROUP = 'var_type'

add_variable(name, size=1, var_type='float', l_b=None, u_b=None, value=None)

Add a variable to the design space

Parameters

• name – param size: (Default value = 1)

• var_type – Default value = FLOAT)

• l_b – Default value = None)

• u_b – Default value = None)

• value – Default value = None)

• size – (Default value = 1)

array_to_dict(x_array)

Split the current point into a dictionary with variables names

Parameters

x_array – x array to be converted to a dict of array

check()

Check the state of the design space

check_membership(x_vect, variables_names=None)

Checks whether the input variables satisfy the design space requirements.

Parameters

• x_vect (dict or array) – design variables

• variables_names – names of the variables to be checked

dict_to_array(x_dict, all_vars=True, all_var_list=None)

Aggregate a point as dictionary into array

Parameters

• x_dict – point as dictionary

• all_vars – if True, all variables shall be in x_dict

• all_var_list – list of whole set of variables, if None, use self.variables_names

export_hdf(file_path, append=False)

Export to hdf file.

Parameters

• file_path – path to file to write

• append – if True, appends the data in the file

export_to_txt(output_file, fields=None, header_char='', **table_options)

Exports the design space to a text file

Parameters

• output_file – output file path

• fields – list of fields to export, by default all

filter(keep_variables)

Filters the design space to keep a sublist of variables

Parameters

keep_variables – the list of variables to keep

filter_dim(variable, keep_dimensions)

Filters the design space to keep a sublist of dimensions for a given variable

Parameters

• variable – the variable

• keep_dimensions – the list of dimension to keep

get_active_bounds(x_vec=None, tol=1e-08)

Determine which bound constraints of the current point are active

Parameters

• x_vec – the point at which we check the bounds

• tol – tolerance of comparison of a scalar with a bound (Default value = 1e-8)

get_current_x(variables_names=None)

Gets the current point in the design space.

Parameters

variables_names (list(str)) – names of the required variables, optional

Returns

the x vector as array

Return type

ndarray

get_current_x_dict()

Get the current point in the design space

Returns

the x vector as a dict, keys are the variable names values are the variable vales as np array

get_current_x_normalized()

Returns the current point normalized.

Returns

the x vector as array normalized by the bounds

get_indexed_var_name(variable_name)

Retuns a list of the variables names with their indices such as [x!0,x!1,y,z!0,z!1]

get_indexed_variables_names()

Retuns a list of the variables names with their indices such as [x!0,x!1,y,z!0,z!1]

get_lower_bound(name)

Gets the lower bound of a variable.

Parameters

name – variable name

Returns

variable lower bound (possibly infinite)

get_lower_bounds(variables_names=None)

Generates an array of the variables’ lower bounds.

Parameters

variables_names – names of the variables of which the lower bounds are required

get_pretty_table(fields=None)

Builds a PrettyTable object from the design space data

Parameters

fields – list of fields to export, by default all

Returns

the pretty table object

get_size(name)

Get the size of a variable Return None if the variable is not known

Parameters

name – name of the variable

get_type(name)

Get the type of a variable Return None if the variable is not known

Parameters

name – name of the variable

get_upper_bound(name)

Gets the upper bound of a variable.

Parameters

name – variable name

Returns

variable upper bound (possibly infinite)

get_upper_bounds(variables_names=None)

Generates an array of the variables’ upper bounds.

Parameters

variables_names – names of the variables of which the upper bounds are required

has_current_x()

Tests if current_x is defined

Returns

True if current_x is defined

import_hdf(file_path)

Imports design space from hdf file

Parameters

file_path –

log_me()

Logs a representation of the design_space characteristics as a table

normalize_vect(x_vect, minus_lb=True)

Normalizes a vector of the design space. Unbounded variables are not normalized.

Parameters

• x_vect (ndarray) – design variables

• minus_lb – if True, remove lower bounds at normalization

Returns

normalized vector

project_into_bounds(x_c, normalized=False)

Projects x_c onto the bounds, using a simple coordinate wise approach

Parameters

x_c – x vector (np array)

Returns

projected x_c

static read_from_txt(input_file, header=None)

Parses a csv file to read the DesignSpace

Parameters

• input_file – returns: s: the design space

• header – fields list, or by default, read in the file

Returns

the design space

remove_variable(name)

Remove a variable (and bounds and types) from the design space

Parameters

name – name of the variable to remove

round_vect(x_vect)

Rounds the vector where variables are of integer type

Parameters

x_vect – design variables to round

set_current_variable(name, current_value)

Set the current value of a single variable

Parameters

• name – name of the variable

• current_value – current value of the variable

set_current_x(current_x)

Set the current point

Parameters

current_x – the current design vector

set_lower_bound(name, lower_bound)

Set a new lower bound for variable name

Parameters

• name – name of the variable

• lower_bound – lower bound

set_upper_bound(name, upper_bound)

Set a new upper bound for variable name

Parameters

• name – name of the variable

• upper_bound – upper bound

to_complex()

Casts the current value to complex

unnormalize_vect(x_vect, minus_lb=True, no_check=False)

Unnormalizes a normalized vector of the design space.

Parameters

• x_vect (ndarray) – design variables

• minus_lb – if True, remove lower bounds at normalization

• no_check – if True, dont check that values are in [0,1]

Returns

normalized vector