Data handling

dynax.differentiate(ts, ys)

Approximate the derivatives of a timeseries via cubic interpolation.

PARAMETER	DESCRIPTION
ts	The time of each observation. TYPE: Float[Array, 'k']
ys	The observations themselves. Should use NaN to indicate missing data. TYPE: T

RETURNS	DESCRIPTION
T	PyTree of derivatives of ys evaluated at ts.

dynax.Trajectory

Simple dataclass for storing trajectory like data.

The fields are

• ts: Timestamps, (typically) expected shape (...,)
• ys: State vectors, (typically) expected shape (..., n)
• y_ts: State derivatives, (typically) expected shape (..., n)
• us: Inputs/Excitations, (typically) expected shape (..., m)

__init__(ts=None, ys=None, *, y_ts=None, us=None, dtype=None)

Initialize a Trajectory and converts all arrays to jax arrays.

PARAMETER	DESCRIPTION
ts	Timestamps. Defaults to None. TYPE: Any DEFAULT: None
ys	States. Defaults to None. TYPE: Any DEFAULT: None
y_ts	State derivatives. Defaults to None. TYPE: Any DEFAULT: None
us	Input/Exciataions. Defaults to None. TYPE: Any DEFAULT: None
dtype	Data type of the arrays after conversion to jax arrays. Defaults to None. TYPE: str | type[Any] | numpy.dtype | jax._src.typing.SupportsDType | None DEFAULT: None

__getitem__(index)

Index all fields simultaniously.

PARAMETER	DESCRIPTION
index	Any index into a jax.numpy array. TYPE: Index

RETURNS	DESCRIPTION
dynax.Trajectory	Trajectory with every field indexed by index.

stack(trajectories, axis=0, dtype=None) staticmethod

Broadcasted stacking of trajectories.

PARAMETER	DESCRIPTION
trajectories	Sequence of trajectories to stack. TYPE: Sequence[dynax.Trajectory]
axis	Axis along which to stack. Defaults to 0. TYPE: int DEFAULT: 0
dtype	Optional dtype of the resulting array. If not specified, the dtype will be determined via type promotion rules described in type-promotion. Defaults to None. TYPE: str | type[Any] | numpy.dtype | jax._src.typing.SupportsDType | None DEFAULT: None

RETURNS	DESCRIPTION
dynax.Trajectory	Trajectory containing the stacked arrays.