pdivt is currently coerced to be > 0 by preventing it going below 1e-3. This causes a discontinuity in pdivt and anything that uses it (e.g. L-H threshold constraint) when radiating high powers (e.g. for high tungsten impurities).
Whilst a negative pdivt is mathematically acceptable (a negative potential power) in the model in which it is calculated (Physics.physics()), it causes maths errors in other models where only a physical value is accepted (Physics.calculate_density_limit()). Therefore pdivt should be kept "physical" to ensure all models work with it, but this should be done continuously.
The following equation satisfies these requirements:
$$y = \frac{x}{1 - e^{-x}}$$
keeping $x$ continuously positive whilst maintaining linearity for $x >> 0$. This modification is unphysical (in that it has no physical validity), but it is required to reconcile different models that expect pdivt to only have physical values or not.
pdivtis currently coerced to be > 0 by preventing it going below 1e-3. This causes a discontinuity inpdivtand anything that uses it (e.g. L-H threshold constraint) when radiating high powers (e.g. for high tungsten impurities).Whilst a negative
pdivtis mathematically acceptable (a negative potential power) in the model in which it is calculated (Physics.physics()), it causes maths errors in other models where only a physical value is accepted (Physics.calculate_density_limit()). Thereforepdivtshould be kept "physical" to ensure all models work with it, but this should be done continuously.The following equation satisfies these requirements:
$$y = \frac{x}{1 - e^{-x}}$$ $x$ continuously positive whilst maintaining linearity for $x >> 0$ . This modification is unphysical (in that it has no physical validity), but it is required to reconcile different models that expect
keeping
pdivtto only have physical values or not.