Physics Constraints¶

Inject domain knowledge into MLP training.

Monotonicity¶

from prandtl import Monotonicity
Monotonicity(param_idx=0, sign=1, weight=0.1)

Parameter	Type	Default	Description
`param_idx`	int	—	Which input parameter
`sign`	int	—	+1 increasing, -1 decreasing
`weight`	float	—	Constraint strength (0.01–1.0)

from prandtl import Convexity
Convexity(param_idx=0, sign=-1, weight=0.05)

Parameter	Type	Default	Description
`param_idx`	int	—	Which input parameter
`sign`	int	—	+1 convex, -1 concave
`weight`	float	—	Constraint strength (0.01–1.0)

from prandtl import BoundaryValue
BoundaryValue({"alpha": 0.0}, {"CL": 0.0}, weight=10.0)

Parameter	Type	Default	Description
`condition`	dict	—	Parameter values at the boundary
`target`	dict	—	Output values at the boundary
`weight`	float	—	Constraint strength (1–100)