Symptom: an HLL + hyperelasticity run computes its direct elastic wave-speed estimates from uninitialized G_L/G_R — hyperelasticity is not restricted to any solver, so this is a live, reachable path.
Mechanism: HLL's else if (hyperelasticity) wave-speed branch (master m_riemann_solvers.fpp, post-split m_riemann_solver_hll.fpp:364-367) uses G_L/G_R, which HLL assigns only inside its if (hypoelasticity) energy block. HLLC, by contrast, loads G under the merged elasticity gate before using it.
Introduced: the hyperelastic wave-speed branch arrived with the elasticity series around #727 (@mrodrig6); the G-loading asymmetry between HLL and HLLC has existed since. Tagging @mrodrig6 for context. Related: #1557 (the hyper tau_e formulation divergence between HLL and HLLC — these two would naturally be ruled on together).
Status: untouched by the refactoring series (correctly — physics). Needs an elasticity-author ruling on what HLL's hyperelastic branch should read.
Symptom: an HLL +
hyperelasticityrun computes its direct elastic wave-speed estimates from uninitializedG_L/G_R— hyperelasticity is not restricted to any solver, so this is a live, reachable path.Mechanism: HLL's
else if (hyperelasticity)wave-speed branch (masterm_riemann_solvers.fpp, post-splitm_riemann_solver_hll.fpp:364-367) usesG_L/G_R, which HLL assigns only inside itsif (hypoelasticity)energy block. HLLC, by contrast, loads G under the mergedelasticitygate before using it.Introduced: the hyperelastic wave-speed branch arrived with the elasticity series around #727 (@mrodrig6); the G-loading asymmetry between HLL and HLLC has existed since. Tagging @mrodrig6 for context. Related: #1557 (the hyper tau_e formulation divergence between HLL and HLLC — these two would naturally be ruled on together).
Status: untouched by the refactoring series (correctly — physics). Needs an elasticity-author ruling on what HLL's hyperelastic branch should read.