smash.default_bayesian_optimize_options#

smash.default_bayesian_optimize_options(model, mapping='uniform', optimizer=None)[source]#

Default bayesian optimization options of Model.

Parameters:

modelModel

Primary data structure of the hydrological model smash.

mappingstr, default ‘uniform’

Type of mapping. Should be one of

'uniform'
'distributed'
'multi-linear'
'multi-power'

Hint

See the Mapping section.

optimizerstr or None, default None

Name of optimizer. Should be one of

'sbs' (only for 'uniform' mapping)
'nelder-mead' (only for 'uniform' mapping)
'powell' (only for 'uniform' mapping)
'lbfgsb' (for all mappings)
'adam' (for all mappings)
'adagrad' (for all mappings)
'rmsprop' (for all mappings)
'sgd' (for all mappings)

Note

If not given, a default optimizer will be set as follows:

'sbs' for mapping = 'uniform'
'lbfgsb' for mapping = 'distributed', 'multi-linear', 'multi-power'

Hint

See the Optimization Algorithms section.

Returns:

optimize_optionsdict[str, Any]: Dictionary containing optimization options for fine-tuning the optimization process. The specific keys returned depend on the chosen mapping and optimizer. This dictionary can be directly passed to the optimize_options argument of the bayesian_optimize (or Model.bayesian_optimize) method.

Examples

>>> from smash.factory import load_dataset
>>> setup, mesh = load_dataset("cance")
>>> model = smash.Model(setup, mesh)

Get the default bayesian optimization options for 'uniform' mapping

>>> opt_u = smash.default_bayesian_optimize_options(model, mapping="uniform")
>>> opt_u
{
    'parameters': ['cp', 'ct', 'kexc', 'llr', 'sg0', 'sg1'],
    'bounds': {
        'cp': (1e-06, 1000.0),
        'ct': (1e-06, 1000.0),
        'kexc': (-50, 50),
        'llr': (1e-06, 1000.0),
        'sg0': (1e-06, 1000.0),
        'sg1': (1e-06, 10.0)
    },
    'control_tfm': 'sbs',
    'termination_crit': {'maxiter': 50},
}

Directly pass this dictionary to the optimize_options argument of the bayesian_optimize (or Model.bayesian_optimize) method. It’s equivalent to set optimize_options to None (which is the default value)

>>> model_u = smash.bayesian_optimize(model, mapping="uniform", optimize_options=opt_u)
</> Bayesian Optimize
    At iterate     0    nfg =     1    J = 7.70491e+01    ddx = 0.64
    At iterate     1    nfg =    68    J = 2.58460e+00    ddx = 0.64
    At iterate     2    nfg =   135    J = 2.32432e+00    ddx = 0.32
    At iterate     3    nfg =   202    J = 2.30413e+00    ddx = 0.08
    At iterate     4    nfg =   269    J = 2.26219e+00    ddx = 0.08
    At iterate     5    nfg =   343    J = 2.26025e+00    ddx = 0.01
    At iterate     6    nfg =   416    J = 2.25822e+00    ddx = 0.01
    CONVERGENCE: DDX < 0.01

Get the default bayesian optimization options for a different mapping

>>> opt_ml = smash.default_bayesian_optimize_options(model, mapping="multi-linear")
>>> opt_ml
{
    'parameters': ['cp', 'ct', 'kexc', 'llr', 'sg0', 'sg1'],
    'bounds': {
        'cp': (1e-06, 1000.0),
        'ct': (1e-06, 1000.0),
        'kexc': (-50, 50),
        'llr': (1e-06, 1000.0),
        'sg0': (1e-06, 1000.0),
        'sg1': (1e-06, 10.0)
    },
    'control_tfm': 'normalize',
    'descriptor': {
        'cp': array(['slope', 'dd'], dtype='<U5'),
        'ct': array(['slope', 'dd'], dtype='<U5'),
        'kexc': array(['slope', 'dd'], dtype='<U5'),
        'llr': array(['slope', 'dd'], dtype='<U5')
    },
    'termination_crit': {'maxiter': 100, 'factr': 1000000.0, 'pgtol': 1e-12},
}

Customize the bayesian optimization options and optimize the Model

>>> opt_ml["bounds"]["cp"] = (1, 2000)
>>> opt_ml["bounds"]["sg0"] = (1e-3, 100)
>>> opt_ml["descriptor"]["cp"] = "slope"
>>> opt_ml["termination_crit"]["maxiter"] = 10
>>> model.bayesian_optimize(mapping="multi-linear", optimize_options=opt_ml)
</> Bayesian Optimize
    At iterate     0    nfg =     1    J = 7.70491e+01    |proj g| = 1.05147e+04
    At iterate     1    nfg =     2    J = 6.69437e+00    |proj g| = 2.15263e+02
    At iterate     2    nfg =     3    J = 6.52716e+00    |proj g| = 2.03207e+02
    At iterate     3    nfg =     4    J = 5.08876e+00    |proj g| = 6.83760e+01
    At iterate     4    nfg =     5    J = 4.73664e+00    |proj g| = 4.19148e+01
    At iterate     5    nfg =     6    J = 4.42125e+00    |proj g| = 1.94103e+01
    At iterate     6    nfg =     7    J = 4.28494e+00    |proj g| = 9.39774e+00
    At iterate     7    nfg =     8    J = 4.19646e+00    |proj g| = 4.74194e+00
    At iterate     8    nfg =     9    J = 4.13953e+00    |proj g| = 1.74698e+00
    At iterate     9    nfg =    10    J = 4.09997e+00    |proj g| = 1.04288e+00
    At iterate    10    nfg =    11    J = 4.02741e+00    |proj g| = 4.41394e+00
    STOP: TOTAL NO. of ITERATIONS REACHED LIMIT

The optimization process was terminated after 10 iterations, the maximal value we defined.

smash.default_bayesian_optimize_options#

This Page