eb573f1 | Thorsten Becker | 27 June 2008, 00:53:30 UTC | Check in of proper version for elliptical net rotation correction. | 27 June 2008, 00:53:30 UTC |
9f7396f | Thorsten Becker | 26 June 2008, 23:56:32 UTC | Reordered loop in get_buoyancy to speed up array access. Fixed omission of one elliptical correction, added a flag. | 26 June 2008, 23:56:32 UTC |
fa90e33 | Thorsten Becker | 26 June 2008, 23:04:06 UTC | Replaced several trig function evaluations in Size_does_matter.c witht he precomputed SinCos arrays for speedup. Experimental ellipticity implementation has been improved somewhat, and made accessible only when -DALLOW_ELLIPTICAL is set. This was necessary to not slow the spherical version of the code down with if statements. | 26 June 2008, 23:04:06 UTC |
cc26d94 | Thorsten Becker | 25 June 2008, 01:13:32 UTC | Modified net rotation computation for ellipticity. Minor additions (prototypes) to make the code compile smoothly under PGI. | 25 June 2008, 01:13:32 UTC |
6ada82f | Eh Tan | 24 June 2008, 21:18:37 UTC | Modified routs_s for lines. Added some debugging output and updated comments. | 24 June 2008, 21:18:37 UTC |
a0ca2a8 | Eh Tan | 24 June 2008, 21:11:18 UTC | Only top/bottom processors need to communicate | 24 June 2008, 21:11:18 UTC |
30e6081 | Thorsten Becker | 21 June 2008, 22:28:51 UTC | Assignment of velocity boundary condition only called once at the top multigrid level during boundary condition assignment. Experimental implementation of mixed velocity/free slip boundary condition based on netcdf grids, in Ggrd_handling. | 21 June 2008, 22:28:51 UTC |
e6aaa4d | Eh Tan | 20 June 2008, 23:01:28 UTC | Enable CBF topography | 20 June 2008, 23:01:28 UTC |
9cab369 | Eh Tan | 20 June 2008, 22:51:56 UTC | Added back get_CBF_topo(), from Shijie Zhong, with some minor modifications. Minor modifications: * some double arrays -> float arrays * calling exchange_snode_d() -> calling *_exchange_snode_f() | 20 June 2008, 22:51:56 UTC |
1f78588 | Eh Tan | 20 June 2008, 22:50:51 UTC | Resurrected full_parallel_communication_routs_s(), which is needed by CBF topo. Renamed exchange_snode_f() to full_exchange_snode_f() and regional_exchange_snode_f(). | 20 June 2008, 22:50:51 UTC |
b0228c2 | Eh Tan | 20 June 2008, 22:42:30 UTC | Added get_global_1d_shape_fn_L() from Shijie Zhong | 20 June 2008, 22:42:30 UTC |
d98cafb | Eh Tan | 20 June 2008, 22:41:57 UTC | Compute E->L and E->Lx 1d shape functions (1d is a misnomer, they are 2d actually). | 20 June 2008, 22:41:57 UTC |
2e17e1d | Eh Tan | 20 June 2008, 19:08:03 UTC | The function declaration was put at a wrong place. | 20 June 2008, 19:08:03 UTC |
bf432bb | Eh Tan | 19 June 2008, 22:20:07 UTC | Change back the order of input arguments to match the manual. | 19 June 2008, 22:20:07 UTC |
fff43af | Eh Tan | 19 June 2008, 21:43:25 UTC | The comment in rheol=4 doesn't match the code and is removed. Added option rheol=7, with a complete temperature- and pressure-dependent Arrhenius law. This fixed issue152. | 19 June 2008, 21:43:25 UTC |
c3a51e5 | Eh Tan | 19 June 2008, 21:42:27 UTC | Added debugging (but disabled) function for spherical harmonics expansion. | 19 June 2008, 21:42:27 UTC |
ab74087 | Eh Tan | 19 June 2008, 01:05:12 UTC | Send geoid arrays to vertical columns of processors. After this fix, the geoid output is the same for nprocz=1 and nprocz>1. | 19 June 2008, 01:05:12 UTC |
08ab04b | Eh Tan | 19 June 2008, 01:04:06 UTC | Compute E->surf_det (top surface area, ie. surface jacobian, of an element) on a unit sphere. Use E->surf_det, instead of the surface jacobian returned by get_global_1d_shape_fn(), to expand spherical harmonics. The surface jacobian returned by get_global_1d_shape_fn() is defined at the top radius of the current mesh. When nprocz>1, the top radius is not the same as the outer radius of the sphere for processors. As a result, the spherical harmonics expansion gave incorrect result when nprocz>1 in the previous revisions. | 19 June 2008, 01:04:06 UTC |
d3025a8 | Thorsten Becker | 18 June 2008, 22:35:04 UTC | Speed up of computation by precomputing constants. | 18 June 2008, 22:35:04 UTC |
2d8894e | Eh Tan | 17 June 2008, 21:23:45 UTC | Compute geoid with self gravitation. Code contributed by Shijie Zhong. | 17 June 2008, 21:23:45 UTC |
de569a3 | Thorsten Becker | 17 June 2008, 20:31:30 UTC | Added a few more comments and made a few loops more efficient. | 17 June 2008, 20:31:30 UTC |
67ed449 | Thorsten Becker | 14 June 2008, 00:00:58 UTC | Added a few comments on Rayleigh number netcdf control. | 14 June 2008, 00:00:58 UTC |
9735d41 | Eh Tan | 13 June 2008, 23:36:53 UTC | update NEWS item | 13 June 2008, 23:36:53 UTC |
d5dc9d3 | Eh Tan | 13 June 2008, 23:35:38 UTC | Replaced MPI_COMM_WORLD by E->parallel.world. MPI_COMM_WORLD is the global communicator. E->parallel.world is the communicator for the current solver. These two are different in a coupled run. Before this fix, full-regional coupling will dead lock. | 13 June 2008, 23:35:38 UTC |
adc7607 | Eh Tan | 13 June 2008, 23:34:08 UTC | Added some comments | 13 June 2008, 23:34:08 UTC |
dbd7942 | Eh Tan | 13 June 2008, 23:33:29 UTC | To be consistent with C version, disable checkpoint point output at 0th step if restarting. | 13 June 2008, 23:33:29 UTC |
35878a7 | Eh Tan | 13 June 2008, 23:32:21 UTC | Correctly counting the global number of equations. This will slightly affect the results of the stokes solver. This avoid double counting the overlapping nodes between processors and the global norm of velocity vector. | 13 June 2008, 23:32:21 UTC |
e5d4428 | Eh Tan | 13 June 2008, 23:25:41 UTC | Using firewall channel, since error channel doesn't stop the code. | 13 June 2008, 23:25:41 UTC |
94e1195 | Eh Tan | 13 June 2008, 23:24:32 UTC | Decalred variables as double to avoid float-double conversion later in the loop. | 13 June 2008, 23:24:32 UTC |
b118cc2 | Eh Tan | 13 June 2008, 19:14:12 UTC | update the help message | 13 June 2008, 19:14:12 UTC |
d2eccfa | Thorsten Becker | 12 June 2008, 21:22:41 UTC | Added tentative implementation of a rotating, elliptical Earth. Made restart for constant viscosity cases possible. Added alternative geoid computation place holder for self-gravitating mantle, to be filled out by Shijie's old code. | 12 June 2008, 21:22:41 UTC |
6f05c7a | Thorsten Becker | 30 May 2008, 18:42:57 UTC | Fixed typo in memory allocation procedure that led to core dump. | 30 May 2008, 18:42:57 UTC |
de1b3e0 | Michael Gurnis | 12 May 2008, 21:22:42 UTC | fixed typo | 12 May 2008, 21:22:42 UTC |
1fbd9ec | Michael Gurnis | 12 May 2008, 21:16:55 UTC | Just a place holder for the next release | 12 May 2008, 21:16:55 UTC |
9a0fcce | Michael Gurnis | 05 May 2008, 19:17:31 UTC | Bug Fix | 05 May 2008, 19:17:31 UTC |
8ffc5f0 | Eh Tan | 01 May 2008, 22:59:14 UTC | Add python binding to update top surface temperatures | 01 May 2008, 22:59:14 UTC |
e7216e6 | Michael Gurnis | 01 May 2008, 22:56:12 UTC | More additions to allow top surface temperatures to be read in a each time step. Should read in every time step for C only version. Additional changes for Pyre version to follow; will test more once these changes are made. Small changes to viscosity structures when reading in materials for each element. More documentation to follow. | 01 May 2008, 22:56:12 UTC |
e137764 | Eh Tan | 01 May 2008, 00:51:25 UTC | Inputing tbcs_file & temperature_boundary_file | 01 May 2008, 00:51:25 UTC |
7499aea | Michael Gurnis | 30 April 2008, 23:11:18 UTC | Start to add new features for reading surface temperature for all top surface node points Added new variables to the CONTROL structure vbcs_file & temperature_boundary_fi le | 30 April 2008, 23:11:18 UTC |
1d4d90a | Thorsten Becker | 06 April 2008, 21:58:45 UTC | GMT/Netcdf grd input can now deal with velocity boundary conditions, material dependence, and local Rayleigh number in surface layers. | 06 April 2008, 21:58:45 UTC |
1866b6f | Eh Tan | 05 March 2008, 21:15:02 UTC | Close trace_log files when finalizing | 05 March 2008, 21:15:02 UTC |
adc1f2d | Eh Tan | 05 March 2008, 21:07:02 UTC | Clean up unused arrays and functions | 05 March 2008, 21:07:02 UTC |
b33dc3c | Eh Tan | 05 March 2008, 21:06:11 UTC | Write exclusive tracer time every 20 steps | 05 March 2008, 21:06:11 UTC |
331291e | Eh Tan | 05 March 2008, 21:04:21 UTC | Revert r11328 | 05 March 2008, 21:04:21 UTC |
1709362 | Eh Tan | 05 March 2008, 20:50:31 UTC | Write exclusive tracer time every 20 steps | 05 March 2008, 20:50:31 UTC |
37cc271 | Eh Tan | 28 February 2008, 01:03:50 UTC | Removed artificial tolerance of empty elements. This greatly decreases the minimal number of tracers per element required for a given mesh. This fixed issue140. | 28 February 2008, 01:03:50 UTC |
cc3e43d | Eh Tan | 28 February 2008, 01:00:54 UTC | Minor tweak | 28 February 2008, 01:00:54 UTC |
e15117c | Thorsten Becker | 27 February 2008, 07:42:44 UTC | Built surface velocity BC support via Netcdf grd files back in. This needs more testing, but it would be good to have the framework in place before new additions are made. Also, age control is not yet implemented, but will be soon. | 27 February 2008, 07:42:44 UTC |
332d337 | Eh Tan | 21 February 2008, 01:25:34 UTC | Removed unnecessary call to construct_mat_group() | 21 February 2008, 01:25:34 UTC |
c42f6fd | Eh Tan | 21 February 2008, 01:03:43 UTC | Finished interpolating fields onto tracers | 21 February 2008, 01:03:43 UTC |
222c941 | Eh Tan | 21 February 2008, 01:02:41 UTC | Renamed fix_phi() to fix_angle() | 21 February 2008, 01:02:41 UTC |
2bd8f42 | Eh Tan | 21 February 2008, 01:01:09 UTC | Fixed issue139. Bug for tracers near South pole. | 21 February 2008, 01:01:09 UTC |
c0e2815 | Eh Tan | 21 February 2008, 01:00:21 UTC | Get new tracers in post processing stage | 21 February 2008, 01:00:21 UTC |
346ba4e | Eh Tan | 21 February 2008, 00:56:29 UTC | Read multiple extraq | 21 February 2008, 00:56:29 UTC |
5cc21b8 | Eh Tan | 21 February 2008, 00:55:41 UTC | Write averaged composition in horiz_avg output. The HDF5 part of output is not implemented... | 21 February 2008, 00:55:41 UTC |
892dd0d | Eh Tan | 21 February 2008, 00:54:25 UTC | Removed horizontal average from output_interpolated_fields() | 21 February 2008, 00:54:25 UTC |
89dc1c4 | Eh Tan | 21 February 2008, 00:53:35 UTC | Compute horizontal average for composition fields | 21 February 2008, 00:53:35 UTC |
a9e6ba7 | Thorsten Becker | 17 February 2008, 00:14:06 UTC | Modified Citcom.c such that E->mat gets assigned based on element depth (as was default before), even if mat_control is not selected. Added (back in?) ggrd netcdf type VIP prefactor material control via ggrd_mat_control flags (as documented in Instructions.c) Renamed some of the ggrd stuff to make it more modular. | 17 February 2008, 00:14:06 UTC |
797ffec | Eh Tan | 09 February 2008, 00:14:35 UTC | Interpolate fields in post_processing() * The parameter "itracer_interpolate_fields" is the same as "model_type" in the AVM code. * If itracer_interpolate_fields is 1, 2, or 3, it interpolates temperature and compositions. The horizonal average is not removed yet. | 09 February 2008, 00:14:35 UTC |
0262f7b | Eh Tan | 08 February 2008, 23:58:57 UTC | Refactored the functions. * get shape functions on tracer location * interpolate data using acquired shape functions | 08 February 2008, 23:58:57 UTC |
a0c621f | Eh Tan | 08 February 2008, 23:56:39 UTC | Added binding for post_processing(). Exit program after post processing and saving results. | 08 February 2008, 23:56:39 UTC |
8688232 | Eh Tan | 08 February 2008, 23:54:36 UTC | * Using a different reference point for uv space. This fixed a nasty bug when some cap extends over ~pi/2. * A bunch of debugging output (disabled) * Disable more debugging output | 08 February 2008, 23:54:36 UTC |
aed8cda | Eh Tan | 08 February 2008, 23:50:30 UTC | Added new parameter: itracer_interpolate_fields When it is 0 (default), nothing is done. Otherwise, the convection fields are interpolated onto the location of tracers. | 08 February 2008, 23:50:30 UTC |
5b88643 | Eh Tan | 04 February 2008, 23:23:21 UTC | Reverted part of unwannted change | 04 February 2008, 23:23:21 UTC |
f3b65a8 | Eh Tan | 04 February 2008, 23:21:19 UTC | Fixed potential memory leak | 04 February 2008, 23:21:19 UTC |
7455cf7 | Eh Tan | 04 February 2008, 23:13:23 UTC | Fixed bugs in managing reference counts and potential double-free | 04 February 2008, 23:13:23 UTC |
61854c2 | Eh Tan | 04 February 2008, 20:40:45 UTC | Fixed a regression reported by Lydia Dicaprio. The initial temperature should be constructured in this order: 1. if tic_method=-1, read from velo files. 2. if lith_age=1, read/constructed from lith_age files. 3. otherwise, call {full,regional}_construct_tic_from_input() | 04 February 2008, 20:40:45 UTC |
8eda467 | Eh Tan | 22 January 2008, 19:52:09 UTC | Renamed functions | 22 January 2008, 19:52:09 UTC |
2c8211b | Eh Tan | 22 January 2008, 19:38:43 UTC | Computes the derivatives of shape functions and jacobians only once and stores them for later use. This reduces the total cpu time of cookbook8 by 13%. - get_global_shape_fn_sph() computes and stores the derivatives of shape functions and jacobians. - Moved get_global_shape_fn() to Obsolete.c - Calling construct_shape_function_derivatives() in initial_mesh_solver_setup(). This function call get_global_shape_fn_sph() for each element. - get_rtf_at_vpts() and get_rtf_at_ppts() are for coord. transformation matrix at vpts and ppts respectively - Passed arguments by references in PG solvers instead of by values to avoid copying. | 22 January 2008, 19:38:43 UTC |
64dc706 | Thorsten Becker | 17 January 2008, 02:06:25 UTC | Changed GNOMONIC to CITCOM_GNOMONIC to avoid namespace conflict with GMT gmt.h | 17 January 2008, 02:06:25 UTC |
07ddb1c | Eh Tan | 16 January 2008, 20:53:14 UTC | Forgot to checkin this file | 16 January 2008, 20:53:14 UTC |
97b3adf | Eh Tan | 16 January 2008, 20:34:22 UTC | Added timing information for tracer module. The timing is reported in *.tracer_log.* files. | 16 January 2008, 20:34:22 UTC |
077bc61 | Eh Tan | 16 January 2008, 20:30:09 UTC | A more memory efficient way to find shape function for tracers. - Migrated E->tracer.UV to E->gnomonic. The reference point is the 1st node of the local mesh. The 0th element of E->gnomonic stores the sine of cosine of reference theta. - Convertd shape_coefs array from size of nel to size of snel to save memory. | 16 January 2008, 20:30:09 UTC |
1637c6f | Eh Tan | 15 January 2008, 23:36:54 UTC | Added Luis Armendariz to the author list | 15 January 2008, 23:36:54 UTC |
7d95bab | Eh Tan | 15 January 2008, 23:11:38 UTC | Simplified loop over levels | 15 January 2008, 23:11:38 UTC |
a12771a | Eh Tan | 15 January 2008, 23:05:58 UTC | Reverted previous checkin | 15 January 2008, 23:05:58 UTC |
e30ef29 | Eh Tan | 15 January 2008, 23:01:26 UTC | Simplified loop over levels | 15 January 2008, 23:01:26 UTC |
6febbb9 | Eh Tan | 15 January 2008, 00:29:40 UTC | Forgot to rename variables | 15 January 2008, 00:29:40 UTC |
50251ba | Eh Tan | 14 January 2008, 23:03:21 UTC | Fixed 2 bugs in full_coord_of_cap(). 1. x and y arrays were overridden in the loop of level. 2. Choice of reference point was not optimal and may cause round off problem. | 14 January 2008, 23:03:21 UTC |
50c6d05 | Eh Tan | 03 January 2008, 19:51:53 UTC | Re-applied patches that got reverted in r8973 | 03 January 2008, 19:51:53 UTC |
b08db17 | Thorsten Becker | 29 December 2007, 07:02:49 UTC | Trying to sync back up, those should be old changes, I hope. | 29 December 2007, 07:02:49 UTC |
7d7984a | Eh Tan | 20 December 2007, 20:30:13 UTC | Generated global mesh by great circles. Tracer module can split the cap in the map view now. | 20 December 2007, 20:30:13 UTC |
aab74be | Eh Tan | 20 December 2007, 20:27:41 UTC | Fine tune the mesh so that each cap has the same surface area | 20 December 2007, 20:27:41 UTC |
1946b88 | Eh Tan | 20 December 2007, 20:26:04 UTC | Removed obsolete code | 20 December 2007, 20:26:04 UTC |
031dd16 | Eh Tan | 20 December 2007, 20:25:33 UTC | Fixed a redundant cast | 20 December 2007, 20:25:33 UTC |
a6cfbf2 | Eh Tan | 17 December 2007, 22:06:06 UTC | Fixed phase change width | 17 December 2007, 22:06:06 UTC |
726f389 | Eh Tan | 12 November 2007, 22:42:56 UTC | Updated the version number and file NEWS | 12 November 2007, 22:42:56 UTC |
8b1c234 | Eh Tan | 12 November 2007, 22:38:36 UTC | Added mishing header file to the makefile, also avoided compiling unused code. | 12 November 2007, 22:38:36 UTC |
3585011 | Thorsten Becker | 11 November 2007, 20:37:22 UTC | Synching with the new release, none of those changes should be relevant. | 11 November 2007, 20:37:22 UTC |
0612d94 | Eh Tan | 08 November 2007, 22:41:31 UTC | Updated file ChangeLog to r8244 | 08 November 2007, 22:41:31 UTC |
209f8f4 | Eh Tan | 08 November 2007, 22:31:21 UTC | Replaced a system call by std C library remove() and disabled another system call (backup input file). Partially fixed issue130. All remaining system calls are in lib/Output_gzdir.c. | 08 November 2007, 22:31:21 UTC |
d280056 | Eh Tan | 08 November 2007, 21:38:09 UTC | Updated file ChangeLog to r8242 | 08 November 2007, 21:38:09 UTC |
eb0cd74 | Eh Tan | 08 November 2007, 21:36:55 UTC | Removed binary checkpoint files from makefile, as the file size is too big for distribution. | 08 November 2007, 21:36:55 UTC |
4ffbc25 | Eh Tan | 08 November 2007, 21:17:14 UTC | Updated file ChangeLog to r8240 | 08 November 2007, 21:17:14 UTC |
2b294a0 | Eh Tan | 08 November 2007, 19:19:12 UTC | Update the version number | 08 November 2007, 19:19:12 UTC |
744e0f1 | Eh Tan | 08 November 2007, 19:18:52 UTC | Update NEWS | 08 November 2007, 19:18:52 UTC |
6e63c0b | Eh Tan | 08 November 2007, 19:18:26 UTC | Dereased the timestep size to reduce artifacts in advection | 08 November 2007, 19:18:26 UTC |
678f8c2 | Eh Tan | 07 November 2007, 19:52:39 UTC | Decreased the # of processors. This is the only way I can reproduce single-cell convection as in the manual. | 07 November 2007, 19:52:39 UTC |
ef214c1 | Eh Tan | 07 November 2007, 19:52:17 UTC | Removed comp_el from default output, since it is not required for restart anymore. | 07 November 2007, 19:52:17 UTC |
9dbf4b9 | Eh Tan | 07 November 2007, 19:51:56 UTC | Print a warning when matrix eqn solver not converging | 07 November 2007, 19:51:56 UTC |