Third-party models

A third-party model can participate in many rtd-sensor workflows without being registered in the built-in catalog and without inheriting from a package class.

The object must provide the structural methods required by the workflow.

Minimal full-model example

class ExampleModel:
    def celsius_to_resistance(self, temperature_c: float) -> float:
        return 100.0 + 0.4 * temperature_c

    def resistance_to_celsius(self, resistance_ohms: float) -> float:
        return (resistance_ohms - 100.0) / 0.4

    def resistance_sensitivity_ohms_per_celsius(self, temperature_c: float) -> float:
        return 0.4

    def temperature_sensitivity_celsius_per_ohm(self, temperature_c: float) -> float:
        return 2.5

This is only an illustrative linear model, not a documented physical RTD.

Use it with batch conversion

from rtd_sensor import batch

model = ExampleModel()
values = batch.celsius_to_resistance(model, [0.0, 10.0, 20.0])

Use it with a resistance reader

from rtd_sensor import measurement


class Reader:
    def read_resistance_ohms(self) -> float:
        return 110.0


temperature_c = measurement.read_temperature_celsius(
    Reader(),
    model=model,
)

Use it with uncertainty propagation

If it provides resistance_to_celsius() and temperature_sensitivity_celsius_per_ohm(), it can satisfy the narrower uncertainty protocol as well.

What rtd-sensor does not do for third-party models

The package does not automatically verify the scientific provenance, range, monotonicity, exception semantics, or numerical accuracy of an arbitrary third-party object. Its exceptions also propagate unchanged rather than being wrapped as package-owned model errors.

Use the built-in model classes when you want rtd-sensor's own validation semantics for CVD, polynomial, piecewise-polynomial, or tabulated definitions.