The main objective of this exercise is to analyse the convergence of the frequency computations for the different oscillation codes.

The actual models and information details on the HE and BV tests are given in Models (exercise #2)

Step 1. We select the well-known modelS, computed by Jørgen for which HE and BV tests were positively passed, ie. models 2-001 and 2-002, respectively. Using ADIPLS routines, we computed these models for a different number of mesh points (2K, 4K, 6K, 8K, and 16K) redistributed in an optimised way to properly map certain physical variable at the rapidly varying regions of the model.

Step 2. Computation of adiabatic oscillation frequencies

1. Only p-modes (up to cut-off frequency) (L=0,1,2,3).
2. No Richardson Extrapolation
3. Compute both with Lagrangian and Eulerian variables
4. Limit (surface) conditions: P → 0
5. Same gravitation constant G: 6.6743e-8 (cgs)

**Step 3. Comparison of

1. Individual frequencies around nu_max (± 10 radial orders)
2. Comparison of Δν and δν

The first results on the exercise can be read at the following online report.

The numerical convergence of the frequencies shown by GYRE and ADIPLS is robust and reliable. GraCo results are not representative (we will discuss them in detail when IS are delivered). The variations in the frequencies computed by these codes are around 0.005 μHz in the worst case, and around 0.002μHz around n= 20.

P04 seems good for the frequency computations, although P06 is the optimum case (in terms of frequency differences). It worth study the impact of choosing P06 instead of P4 on the computation time.

As expected, model 2-002 gives slightly better global behaviour in the statistics.

The choice of eulerian/lagrangian variable has no significant impact on the results.