base module¶

SSBJ base class¶

class gemseo.problems.sobieski.base.SobieskiBase(dtype)[source]¶

Bases: object

Class defining Sobieski problem and related method to the problem such as disciplines computation, constraints, reference optimum

Constructor

Parameters: dtype (str) – data type

DTYPE_COMPLEX = 'complex128'¶

DTYPE_DEFAULT = 'complex128'¶

DTYPE_DOUBLE = 'float64'¶

static compute_aero_center(x_1)[source]¶

Compute a wing thickness.

Parameters: x_1 (numpy array) – local design variables (structure)
Returns: aerodynamic center
Return type: numpy array

compute_half_span(x_shared)[source]¶

Compute half-span from surface and aspect ratio.

Parameters: x_shared (numpy array) – global design variables
Returns: half-span
Return type: numpy array

compute_thickness(x_shared)[source]¶

Compute a wing thickness.

Parameters: x_shared (numpy array) – global design variables
Returns: thickness
Return type: numpy array

default_constants()[source]¶

Definition of constants vector constants for Sobieski problem.

Returns: constant vector
Return type: numpy array

derive_normalize_s(s_ref, s_new)[source]¶

Derivation of normalization of input variables for use of polynomial approximation.

Parameters

s_ref (numpy array) – vector of initial values of independant variables (5 variables at max)
s_new (numpy array) – vector of current values of independant variables

Returns

derivatives of normalized value and normalized+centered value

Return type

numpy array

derive_poly_approx(s_ref, s_new, flag, s_bound)[source]¶

Compute the polynomial coefficients to characterize the behavior of certain synthetic variables and function modifiers. Compared to poly_approx, also returns polynomial coeff for linearization

Parameters

s_ref (numpy array) – vector of initial values of independant variables (5 variables at max)
s_new (numpy array) – vector of current values of independant variables
flag (int) –
indicates functional relationship between variables:
- flag = 1: linear >0
- flag = 2: nonlinear >0
- flag = 3: linear < 0
- flag = 4: nonlinear <0
- flag = 5: parabolic
s_bound (numpy array) – offset value for normalization

Returns

poly_value: value of synthetic variable or modifier

Return type

numpy array

classmethod get_bounds_by_name(variables_names)[source]¶

Return bounds of design variables and coupling variables.

Parameters: variables_names (str or list(str)) – names of variables
Returns: lower bound and upper bound

get_default_x0()[source]¶

Return a default initial value for design variables.

Returns: initial design variables
Return type: numpy array

warning: ##DO NOT CHANGE VALUE: THEY ARE USED FOR POLYNOMIAL APPROXIMATION ##

get_initial_values()[source]¶

Get initial values used by polynomial functions.

Returns: initial values
Return type: tuple(ndarray)

classmethod get_sobieski_bounds()[source]¶

Set the input design bounds and return them as 2 numpy arrays.

Returns: upper and lower bounds
Return type: numpy array, numpy array

static get_sobieski_bounds_tuple()[source]¶

Set the input design bounds and return them as a tuple of tuples.

Returns: Subtuple is build with lower bound and upper bound.
Return type: tuple(tuple)

poly_approx(s_ref, s_new, flag, s_bound)[source]¶

This function calculates polynomial coefficients to characterize the behavior of certain synthetic variables and function modifiers.

Parameters

s_ref (numpy array) – vector of initial values of independant variables (5 variables at max)
s_new (numpy array) – vector of current values of independant variables
flag (int) –
indicates functional relationship between variables:
- flag = 1: linear >0
- flag = 2: nonlinear >0
- flag = 3: linear < 0
- flag = 4: nonlinear <0
- flag = 5: parabolic
s_bound (numpy array) – offset value for normalization

Returns

poly_value: value of synthetic variable or modifier

Return type

numpy array