haversine_distance

HaversineDistanceLayer ¤

HaversineDistanceLayer(
    name=None,
    input_dtype=None,
    output_dtype=None,
    lat_lon_constant=None,
    unit="km",
    **kwargs
)

Bases: BaseLayer

Computes the haversine distance operation on a given input tensor.

If lat_lon_constant is not set, inputs must be a list of 4 tensors, in the order of lat1, lon1, lat2, lon2. If lat_lon_constant is set, inputs must be a tensor of 2 tensors, in the order of lat1, lon1.

We DO NOT check if the lat/lon values are out of bounds. For lat, this is [-90, 90] and for lon, this is [-180, 180].

Initializes the HaversineDistanceLayer layer

Parameters:

Name	Type	Description	Default
`name`	`Optional[str]`	Name of the layer, defaults to `None`.	`None`
`input_dtype`	`Optional[str]`	The dtype to cast the input to. Defaults to `None`.	`None`
`output_dtype`	`Optional[str]`	The dtype to cast the output to. Defaults to `None`.	`None`
`lat_lon_constant`	`Optional[List[float]]`	The lat/lons to use in the haversine distance.	`None`
`unit`	`str`	The unit of the distance. Must be either 'km' or 'miles'. calculation. Defaults to `None`.	`'km'`

Source code in src/kamae/keras/core/layers/haversine_distance.py

def __init__(
    self,
    name: Optional[str] = None,
    input_dtype: Optional[str] = None,
    output_dtype: Optional[str] = None,
    lat_lon_constant: Optional[List[float]] = None,
    unit: str = "km",
    **kwargs: Any,
) -> None:
    """
    Initializes the HaversineDistanceLayer layer

    :param name: Name of the layer, defaults to `None`.
    :param input_dtype: The dtype to cast the input to. Defaults to `None`.
    :param output_dtype: The dtype to cast the output to. Defaults to `None`.
    :param lat_lon_constant: The lat/lons to use in the haversine distance.
    :param unit: The unit of the distance. Must be either 'km' or 'miles'.
    calculation. Defaults to `None`.
    """
    super().__init__(
        name=name, input_dtype=input_dtype, output_dtype=output_dtype, **kwargs
    )
    if lat_lon_constant is not None and len(lat_lon_constant) != 2:
        raise ValueError("If set, lat_lon_constant must be a list of 2 floats")
    self.lat_lon_constant = lat_lon_constant
    if unit not in ["km", "miles"]:
        raise ValueError("unit must be either 'km' or 'miles'")
    self.unit = unit
    self.earth_radius = 6371.0 if unit == "km" else 3958.8

compatible_dtypes `property` ¤

compatible_dtypes

Returns the compatible dtypes of the layer.

Returns:

Type	Description
`Optional[List[str]]`	The compatible dtypes of the layer.

compute_haversine_distance ¤

compute_haversine_distance(lat1, lon1, lat2, lon2)

Computes the haversine distance between two lat/lon pairs.

Parameters:

Name	Type	Description	Default
`lat1`	`KerasTensor`	Tensor of latitudes of the first point.	required
`lon1`	`KerasTensor`	Tensor of longitudes of the first point.	required
`lat2`	`KerasTensor`	Tensor of latitudes of the second point.	required
`lon2`	`KerasTensor`	Tensor of longitudes of the second point.	required

Returns:

Type	Description
`KerasTensor`	Tensor of haversine distances.

Source code in src/kamae/keras/core/layers/haversine_distance.py

def compute_haversine_distance(
    self, lat1: KerasTensor, lon1: KerasTensor, lat2: KerasTensor, lon2: KerasTensor
) -> KerasTensor:
    """
    Computes the haversine distance between two lat/lon pairs.

    :param lat1: Tensor of latitudes of the first point.
    :param lon1: Tensor of longitudes of the first point.
    :param lat2: Tensor of latitudes of the second point.
    :param lon2: Tensor of longitudes of the second point.
    :returns: Tensor of haversine distances.
    """
    lat1_radians = get_radians(lat1)
    lon1_radians = get_radians(lon1)
    lat2_radians = get_radians(lat2)
    lon2_radians = get_radians(lon2)

    lat_diff = lat2_radians - lat1_radians
    lon_diff = lon2_radians - lon1_radians

    a = ops.power(ops.sin(lat_diff / 2.0), 2.0) + ops.cos(lat1_radians) * ops.cos(
        lat2_radians
    ) * ops.power(ops.sin(lon_diff / 2.0), 2.0)
    c = 2.0 * ops.arcsin(ops.power(a, 0.5))
    # Radius of earth in kilometers or miles
    r = ops.convert_to_tensor(self.earth_radius, dtype=c.dtype)
    return c * r

get_config ¤

get_config()

Gets the configuration of the HaversineDistance layer. Used for saving and loading from a model.

Specifically, we add the lat_lon_constant and unit to the config.

Returns:

Type	Description
`Dict[str, Any]`	Dictionary of the configuration of the layer.

Source code in src/kamae/keras/core/layers/haversine_distance.py

def get_config(self) -> Dict[str, Any]:
    """
    Gets the configuration of the HaversineDistance layer.
    Used for saving and loading from a model.

    Specifically, we add the `lat_lon_constant` and `unit` to the config.

    :returns: Dictionary of the configuration of the layer.
    """
    config = super().get_config()
    config.update({"lat_lon_constant": self.lat_lon_constant, "unit": self.unit})
    return config

haversine_distance