Optimizers¶
LISO has its own single- and multi-objective optimizers. It also provides interfaces for 3rd party optimizers.
- class liso.optimizers.optimizer.Optimizer(seed=None)¶
Bases:
abc.ABCAbstract class for optimizers.
- name¶
Name of the optimizer.
- Type
str
- seed¶
Seed for random number. Default = None.
- Type
int
- __init__(seed=None)¶
Optimizer Class Initialization.
- abstract __call__(opt_problem)¶
Run Optimizer (Calling Routine)
- Parameters
opt_problem (Optimization) – Optimization instance.
- Returns
(optimized f, optimized x, miscellaneous information ready for printout).
- Return type
(float, array-like, str)
Local unconstrained optimizers¶
Nelder-Mead¶
- class liso.optimizers.NelderMead(*args, **kwargs)¶
Bases:
liso.optimizers.optimizer.OptimizerNelderMead Optimizer Class.
- rtol¶
Relative tolerance for Lagrange function. Default = 1e-3.
- Type
float
- atol¶
Absolute tolerance for Lagrange function. Default = 1e-4.
- Type
float
- max_stag¶
Maximum number of stagnation (no improvement). Default = 10.
- Type
int
- max_iter¶
Maximum number of iterations. Default = 10000.
- Type
int
- __init__(*args, **kwargs)¶
Optimizer Class Initialization.
- __call__(opt_prob, workers=1)¶
Run Optimizer (Optimize Routine)
Override.
Local constrained optimizers¶
SDPEN (from pyOpt)¶
see SDPEN
Single-objective global optimizers¶
ALPSO¶
- class liso.optimizers.ALPSO(*args, **kwargs)¶
Bases:
liso.optimizers.optimizer.OptimizerALPSO Optimizer Class.
- swarm_size¶
Number of particles. Default = 40.
- Type
int
- topology¶
Topology of the swarm. Default = ‘gbest’.
- Type
str
- max_outer_iter¶
Maximum Number of Outer Loop Iterations. Default = 6.
- Type
int
- max_inner_iter¶
Maximum Number of Inner Loop Iterations. Default = 3.
- Type
int
- min_inner_iter¶
Minimum Number of Inner Loop Iterations.
- Type
int
- etol¶
Absolute tolerance for equality constraints. Default = 1e-4.
- Type
float
- itol¶
Absolute tolerance for inequality constraints. Default = 1e-4.
- Type
float
- rtol¶
Relative tolerance for Lagrange function. Default = 1e-3.
- Type
float
- atol¶
Absolute tolerance for Lagrange function. Default = 1e-4.
- Type
float
- dtol¶
Absolute tolerance for position deviation of all particles . Default = 5e-2.
- Type
float
- c1¶
Cognitive Parameter. Default = 1.5.
- Type
float
- c2¶
Social Parameter. Default = 1.5.
- Type
float
- w0¶
Initial Inertia Weight. Default = 0.90.
- Type
float
- w1¶
Final Inertia Weight. Default = 0.40.
- Type
float
- use_gcpso¶
Use Guaranteed Convergence Particle Swarm Optimization (F. Bergh, A.P. Engelbrecht, A new locally convergent particle swarm optimiser, Systems, Man and Cybernetics , 2002 IEEE International Conference). Default = True.
- Type
bool
- __init__(*args, **kwargs)¶
Optimizer Class Initialization.
- __call__(opt_prob)¶
Run Optimizer (Optimize Routine)
Override.