#include <distortion.hpp>
Static Public Attributes | |
| static constexpr double | kNewtonTolerance = 1e-9 |
| Convergence threshold normalized to the current produced by one volt through the modeled resistance. | |
| static constexpr int | kMaxNewtonIterations = 8 |
| Hard cap on inner-loop iterations per sample. | |
| static constexpr int | kMaxNewtonIterationsPerSample = 2 * kMaxNewtonIterations |
| TR-BDF2 has two implicit stages, each with the cap above. | |
| static constexpr double | kAdaaEpsilon = 1e-5 |
| ADAA fallback guard on |v[n] ā v[nā1]|. | |
|
staticconstexpr |
ADAA fallback guard on |v[n] ā v[nā1]|.
Below this the difference quotient loses precision and the plain evaluation is used. [design parameter] default 1e-5, range 1e-7 .. 1e-3.
|
staticconstexpr |
Hard cap on inner-loop iterations per sample.
An unbounded loop in process() is not RT-safe; hitting the cap clamps to the last iterate, which is a bounded worst-case degrade rather than a failure. [design parameter] default 8, range 4 .. 16.
|
staticconstexpr |
TR-BDF2 has two implicit stages, each with the cap above.
This is the total per-sample ceiling exposed to RT-contract tests.
|
staticconstexpr |
Convergence threshold normalized to the current produced by one volt through the modeled resistance.
Each solve uses kNewtonTolerance / R, making this approximately a node-voltage tolerance instead of an absolute ampere threshold that mutes quiet input. [design parameter] default 1e-9, range 1e-11 .. 1e-6.