systems.base.utils.stochastic.NoiseCharacteristics

systems.base.utils.stochastic.NoiseCharacteristics(
    noise_type,
    num_wiener,
    is_additive,
    is_multiplicative,
    is_diagonal,
    is_scalar,
    depends_on_state,
    depends_on_control,
    depends_on_time,
    state_dependencies,
    control_dependencies,
)

Container for noise structure analysis results.

This dataclass stores all information about the noise structure, enabling automatic solver selection and optimization.

Attributes

Name	Type	Description
noise_type	NoiseType	Classified noise type (ADDITIVE, SCALAR, DIAGONAL, MULTIPLICATIVE, GENERAL)
num_wiener	int	Number of independent Wiener processes (nw)
is_additive	bool	True if diffusion is constant (doesn’t depend on state/control/time)
is_multiplicative	bool	True if diffusion depends on state
is_diagonal	bool	True if noise sources don’t couple (diagonal diffusion matrix)
is_scalar	bool	True if single Wiener process (nw = 1)
depends_on_state	bool	True if diffusion depends on any state variable
depends_on_control	bool	True if diffusion depends on any control variable
depends_on_time	bool	True if diffusion depends on time
state_dependencies	Set[sp.Symbol]	Which specific state variables appear in diffusion
control_dependencies	Set[sp.Symbol]	Which specific control variables appear in diffusion

Examples

>>> char = NoiseCharacteristics(...)
>>> if char.is_additive:
...     G = precompute_constant_noise()  # Optimization!
>>>
>>> solvers = char.recommended_solvers('jax')
>>> print(solvers)  # ['sea', 'shark'] for additive

Methods

Name	Description
recommended_solvers	Recommend efficient solvers based on noise structure.

recommended_solvers

systems.base.utils.stochastic.NoiseCharacteristics.recommended_solvers(backend)

Recommend efficient solvers based on noise structure.

Parameters

Name	Type	Description	Default
backend	Backend	Integration backend (‘jax’, ‘torch’, ‘numpy’)	required

Returns

Name	Type	Description
	List[str]	Recommended solver names, ordered by efficiency/accuracy

Examples

>>> if char.is_additive:
...     solvers = char.recommended_solvers('jax')
...     print(solvers)  # ['sea', 'shark', 'sra1']