| 4 |
|
|
| 5 |
2. cg2d.flow sets adjoint of cg2d_no_nsa to adcg2d_no_nsa -- this is the only reason cg2d_no_nsa.F even needs to exist |
2. cg2d.flow sets adjoint of cg2d_no_nsa to adcg2d_no_nsa -- this is the only reason cg2d_no_nsa.F even needs to exist |
| 6 |
|
|
| 7 |
3. solve_for_pressure.F modified to call the above, only when ALLOW_DIFFERENTIATE_CG2D_MATRIX is defined (in ECCO_CPPOPTIONS). |
3. solve_for_pressure.F modified to call the above, only when ALLOW_DIFFERENTIATE_CG2D_MATRIX is defined and ALLOW_CG2D_NSA is undefined (in ECCO_CPPOPTIONS). |
| 8 |
|
|
| 9 |
4. adcg2d_no_nsa finds adjoint of cg2d_b_ad as well as ac2d_ad etc. Needs common block from ad_taf_output.f to affect the latter. harcoded. |
4. adcg2d_no_nsa finds adjoint of cg2d_b_ad as well as ac2d_ad etc. Needs common block from ad_taf_output.f to affect the latter. harcoded. |
| 10 |
|
|
| 12 |
|
|
| 13 |
6. line 212 of obcs_fields_load.F commented; if not, there are extensive recomputation errors in taf_ad.log i cannot fix. |
6. line 212 of obcs_fields_load.F commented; if not, there are extensive recomputation errors in taf_ad.log i cannot fix. |
| 14 |
|
|
| 15 |
7. shelfice_ad_diff.list in "main_branch_files" may be required. |
7. shelfice_ad_diff.list in "main_branch_files" were modifief but do not seem to be required. |
| 16 |
|
|
| 17 |
|
8. output below for depth control. with temp0 or salt0 control, grdchk is accurate with NONLIN_FRSURF and nonlinearfreesurface=4. |
| 18 |
|
|
| 19 |
OUTPUT WITH NONLIN_FRSURF OFF: |
OUTPUT WITH NONLIN_FRSURF OFF: |
| 20 |
|
|
| 64 |
grad-res 0 6 5 14 1 1 1 1 -9.00753618943E+01 -9.00753618918E+01 -9.00753618967E+01 |
grad-res 0 6 5 14 1 1 1 1 -9.00753618943E+01 -9.00753618918E+01 -9.00753618967E+01 |
| 65 |
grad-res 0 6 6 15 0 1 1 1 2.52987938174E-07 2.44781261927E-07 3.24390020583E-02 |
grad-res 0 6 6 15 0 1 1 1 2.52987938174E-07 2.44781261927E-07 3.24390020583E-02 |
| 66 |
|
|
|
|
|
| 67 |
|
|