geoprior.params.LearnableK

class geoprior.params.LearnableK(initial_value=1.0, log_transform=True, name=None, trainable=True, **kws)

Bases: BaseLearnable

Learnable Hydraulic Conductivity (K).

Indicates that the PINN’s hydraulic conductivity \(K\) should be learned (trainable) if TensorFlow is available; otherwise behaves as a fixed NumPy‐based parameter. We learn \(\log(K)\) to ensure \(K > 0\). The user supplies an initial_value, and the object initializes:

(1)\[\log K \;=\; \log( ext{initial\_value}).\]

Ensures positivity via log-space.

See also

BaseLearnableParam

Examples

>>> k = LearnableK(1.2)
>>> :math:`K = k.get_value()`

Parameters:

• initial_value (float)

• log_transform (bool)

• name (str | None)

• trainable (bool)

__init__(initial_value=1.0, log_transform=True, name=None, trainable=True, **kws)

Parameters:

• initial_value (float)

• log_transform (bool)

• name (str | None)

• trainable (bool)

Methods

__init__([initial_value, log_transform, ...])
from_config(config)	Re-instantiate from get_config().
get_config()	Return a JSON-serialisable dict for tf.keras.
get_value()	Return \(K = \exp(log\_K)\).

__init__(initial_value=1.0, log_transform=True, name=None, trainable=True, **kws)

Parameters:

• initial_value (float)

• log_transform (bool)

• name (str | None)

• trainable (bool)

get_value()

Return \(K = \exp(log\_K)\).

Returns:

Positive conductivity.

Return type:

Union[Tensor, float]