Skip to content

epymorph.data_type

In order to assist with proper data type handling in epymorph, we adopt a set of conventions that somewhat narrow the possibilities for expressing simulation data attributes. This module describes those conventions and provides some utilities for working with them. The goal is to simplify and remove certain categories of errors, like numerical overflow when simulating reasonable numbers of individuals.

ScalarType module-attribute 

ScalarType = type[int | float | str | date]

Supported scalar value types.

StructType module-attribute 

StructType = tuple[tuple[str, ScalarType], ...]

Supported structured types.

AttributeType module-attribute 

AttributeType = ScalarType | StructType

The allowed type declarations for epymorph attributes.

ScalarValue module-attribute 

ScalarValue = int | float | str | date

Supported scalar values.

StructValue module-attribute 

StructValue = tuple[ScalarValue, ...]

Supported structured values.

AttributeValue module-attribute 

AttributeValue = ScalarValue | StructValue

The allowed values types for epymorph attribute values (most notably used as attribute defaults).

ScalarDType module-attribute 

ScalarDType = int64 | float64 | str_ | datetime64

Numpy equivalents to supported scalar values.

StructDType module-attribute 

StructDType = void

Numpy equivalents to supported structured values.

AttributeDType module-attribute 

AttributeDType = ScalarDType | StructDType

The allowed numpy dtypes for use in epymorph: these map 1:1 with AttributeType.

AttributeArray module-attribute 

AttributeArray = NDArray[AttributeDType]

A type describing all supported numpy array forms for attribute data.

CentroidType module-attribute 

CentroidType: AttributeType = (
    ("longitude", float),
    ("latitude", float),
)

Structured epymorph type declaration for longitude/latitude coordinates.

CentroidDType module-attribute 

CentroidDType: dtype[void] = dtype_as_np(CentroidType)

The numpy equivalent of CentroidType (structured dtype for longitude/latitude coordinates).

SimDType module-attribute 

SimDType = int64

This is the numpy datatype that should be used to represent internal simulation data. Where segments of the application maintain compartment and/or event counts, they should take pains to use this type at all times (if possible).

SimArray module-attribute 

SimArray = NDArray[SimDType]

Type alias for a numpy array of SimDType.

dtype_as_np 

dtype_as_np(dtype: AttributeType) -> dtype

Convert a python-style dtype to its numpy-equivalent using epymorph typing conventions.

Parameters:

  • dtype (AttributeType) –

    The attribute type in Python form, e.g., int

Returns:

  • dtype –

    The numpy equivalent, e.g., np.int64

dtype_str 

dtype_str(dtype: AttributeType) -> str

Return a human-readable description of the given attribute data type.

Parameters:

  • dtype (AttributeType) –

    The attribute type in Python form, e.g., int

Returns:

  • str –

    The friendly string representation of the type.

dtype_check 

dtype_check(dtype: AttributeType, value: Any) -> bool

Check that a singular Python value conforms to the given attribute data type. This is not intended to check numpy arrays, only scalars and tuples.

Parameters:

  • dtype (AttributeType) –

    The attribute type in Python form, e.g., int

  • value (Any) –

    A value to check.

Returns:

  • bool –

    True if the values matches the dtype.