Coverage for /opt/hostedtoolcache/Python/3.11.10/x64/lib/python3.11/site-packages/pysagas/optimisation/optimiser.py: 0%
39 statements
« prev ^ index » next coverage.py v7.6.4, created at 2024-10-30 04:27 +0000
1import os
2import numpy as np
3from pysagas import banner
4from numpy.typing import ArrayLike
5from typing import Union, Optional, Dict, Any
6from pyoptsparse import Optimizer, Optimization
9class ShapeOpt:
10 """PySAGAS Shape Optimisation wrapper for pyOptSparse."""
12 def __init__(
13 self,
14 optimiser: Optimizer,
15 working_dir: str,
16 optimiser_options: Optional[Dict[str, Any]] = None,
17 ) -> None:
18 """Initialise PySAGAS Shape Optimiser.
20 Parameters
21 ----------
22 optimiser : Optimizer
23 The pyoptsparse Optimizer object of choice.
25 working_dir : str
26 The name of the working directory.
28 optimiser_options : dict, optional
29 The options to pass to the optimiser.
30 """
31 self.opt_problem: Optimization
32 self.optimiser = optimiser(options=optimiser_options)
34 # Prepare working directory
35 if not os.path.exists(working_dir):
36 os.mkdir(working_dir)
38 def add_variables(
39 self,
40 parameters_dict: Optional[dict] = None,
41 name: Optional[str] = None,
42 n: Optional[int] = None,
43 vartype: Optional[str] = "c",
44 initial: Optional[ArrayLike] = None,
45 lower_bounds: Optional[Union[ArrayLike, Dict]] = None,
46 upper_bounds: Optional[Union[ArrayLike, Dict]] = None,
47 **kwargs,
48 ):
49 """Adds variables to the optimisation problem.
51 Parameters
52 ----------
53 parameters_dict : dict, optional
54 The dictionary of nominal parameter values. This can be provided
55 instead of adding each parameter individually with the 'name'
56 argument.
58 name : str
59 Name of variable group. This name should be unique across all the design variable groups
61 n : int, optional
62 Number of design variables in this variable group.
64 varType : str.
65 String representing the type of variable. Suitable values for type
66 are: 'c' for continuous variables, 'i' for integer values and
67 'd' for discrete selection.
69 value : scalar or array.
70 Starting value for design variables. If it is a a scalar, the same
71 value is applied to all 'nVars' variables. Otherwise, it must be
72 iterable object with length equal to 'nVars'.
74 lower : scalar or array.
75 Lower bound of variables. Scalar/array usage is the same as value
76 keyword
78 upper : scalar or array.
79 Upper bound of variables. Scalar/array usage is the same as value
80 keyword
81 """
83 def check_bounds(
84 parameters: Dict[str, float], bounds: Dict[str, float], sign: int
85 ):
86 for p, v in parameters.items():
87 # Get bound for this parameter
88 b = bounds.get(p)
89 if np.sign(v - b) != np.sign(sign) and np.sign(v - b) != 0:
90 if np.sign(sign) < 0:
91 bound = ">"
92 else:
93 bound = "<"
94 raise Exception(
95 f"Invalid bounds on {p}: " + f"{v:.5f} {bound} {b:.5f}."
96 )
98 if parameters_dict:
99 # Check bounds
100 if lower_bounds is None:
101 lower_bounds = {}
102 else:
103 # Make sure bounds are valid
104 check_bounds(parameters_dict, lower_bounds, 1)
106 if upper_bounds is None:
107 upper_bounds = {}
108 else:
109 # Make sure bounds are valid
110 check_bounds(parameters_dict, upper_bounds, -1)
112 # Unpack parameters dict
113 for param, value in parameters_dict.items():
114 self.opt_problem.addVarGroup(
115 name=param,
116 nVars=1,
117 varType=vartype,
118 value=value,
119 lower=lower_bounds.get(param),
120 upper=upper_bounds.get(param),
121 **kwargs,
122 )
124 else:
125 self.opt_problem.addVarGroup(
126 name=name,
127 nVars=n,
128 varType=vartype,
129 value=initial,
130 lower=lower_bounds,
131 upper=upper_bounds,
132 **kwargs,
133 )
135 def add_constriants(
136 self,
137 name: str,
138 n: Optional[int] = 1,
139 lower: Optional[Union[ArrayLike, Dict]] = None,
140 upper: Optional[Union[ArrayLike, Dict]] = None,
141 scale: Optional[float] = 1,
142 ):
143 """Adds constraints to the optimisation problem.
145 Parameters
146 -----------
147 name : str
148 The name key of the constraint being added. This must appear
149 in the dictionary returned by the objective and Jacobian callback
150 returns.
152 nCon : int
153 The number of constraints in this group
155 lower : scalar or array
156 The lower bound(s) for the constraint. If it is a scalar,
157 it is applied to all nCon constraints. If it is an array,
158 the array must be the same length as nCon.
160 upper : scalar or array
161 The upper bound(s) for the constraint. If it is a scalar,
162 it is applied to all nCon constraints. If it is an array,
163 the array must be the same length as nCon.
165 scale : scalar or array
166 A scaling factor for the constraint. It is generally
167 advisable to have most optimization constraint around the
168 same order of magnitude.
169 """
170 self.opt_problem.addConGroup(
171 name=name,
172 nCon=n,
173 lower=lower,
174 upper=upper,
175 scale=scale,
176 linear=None,
177 wrt=None,
178 jac=None,
179 )
181 def add_objective(self, name: str):
182 """Adds an objective to the optimisation problem.
184 Parameters
185 -----------
186 name : str
187 The name key of the objective being added. This must appear
188 in the dictionary returned by the objective and Jacobian callback
189 returns.
190 """
191 self.opt_problem.addObj(name)
193 def run(self, hotstart_file: str = None):
194 """Run ShapeOpt.
196 Parameters
197 -----------
198 hotstart_file : str, optional
199 The filepath to the history file, used to hot start the
200 optimiser.
201 """
202 # Print banner
203 banner()
204 print("\033[4mPySAGAS Shape Optimisation\033[0m".center(50, " "))
206 # Run optimiser
207 self.sol = self.optimiser(
208 self.opt_problem,
209 storeHistory="history.hst",
210 hotStart=hotstart_file,
211 )