The ``MeanFlowSpec`` interface (paper Section V.A).
Every eigensolver input passes through this one container, so mean-flow fidelity
(synthetic uniform field -> parametric generator -> warm-started SU2 RANS) can be
swapped without touching the mesh, assembly, or eigensolver code. P0 only uses the
synthetic generator at the bottom of this file.
| MeanFlowSpec engine.stability_hifi.meanflow.spec.synthetic_two_zone_duct |
( |
float | L1, |
|
|
float | L2, |
|
|
float | R, |
|
|
int | nx1, |
|
|
int | nx2, |
|
|
int | nr, |
|
|
* | , |
|
|
float | c1, |
|
|
float | c2, |
|
|
float | gamma = 1.2, |
|
|
float | rho = 4.0 ) |
Two-zone duct with a sound-speed (temperature) jump at x=L1 (verification case V2).
``c1`` applies on ``[0, L1]``, ``c2`` on ``[L1, L1+L2]`` — a step discontinuity, not
a smooth profile, which is exactly why this uses ``c_element`` (see ``MeanFlowSpec``
docstring) rather than the ordinary per-node ``c`` field. ``rho`` is passed through
uniformly for now; density does jump physically alongside sound speed at a real
temperature interface (rho = p/(R*T), same p, different T), but the passive
Helmholtz operator (Eq. 10) only ever uses ``c``, so it plays no role in this case.
Definition at line 91 of file spec.py.