sort by:
Revision Author Date Message Commit Date
b0f818a Strengthen the default convergence criterion for the 3rd iteration in compressible cg solver 05 October 2007, 20:00 UTC
72ad845 Disable a debugging output 05 October 2007, 19:58 UTC
4bcf0ad Add licence text to the multi-couplers 05 October 2007, 19:58 UTC
820d139 Turn off filter_temp by default since it's not in the energy equation 05 October 2007, 19:57 UTC
4e87245 Fxied material assignment bug in PDEPV (0 vs. 1 array issue...) 04 October 2007, 19:35 UTC
cc73a80 Updated the cookbook scripts 01 October 2007, 20:38 UTC
f4f2b13 Added Cookbook8 for compressible model + geoid + checkpoint + vsolver tuning 01 October 2007, 19:26 UTC
ae31e95 Fixed restart I/O bug in Output_gzdir. 28 September 2007, 23:36 UTC
e429e26 Added function prototype and fixed a few errors 27 September 2007, 19:33 UTC
a79e4a9 Shrink the size of checkpoint files 27 September 2007, 19:07 UTC
02acf94 Enable coor_refine for regional model, only radial coordinate is refined 27 September 2007, 19:07 UTC
e29d3c2 Fixed a bug in vtkio 25 September 2007, 19:40 UTC
f392355 Introducing choice of reference state: When solver.param.reference_state=0, the reference state is read from a file (solver.param.refstate_file) When solver.param.reference_state=1 (default), the reference state is calculated using Adams-Williamson equation of state. 25 September 2007, 01:31 UTC
4170440 Detect the presence of libz and enable ascii-gz at configure time 24 September 2007, 21:29 UTC
95c453f Comment out varialbe thermal conductivity. Removed unused struct members. 24 September 2007, 21:28 UTC
e4cd242 Multiply rho*cp to DT/dt 24 September 2007, 21:28 UTC
6a4ff8e Don't write to log file, since only rank-0 proc can access it 19 September 2007, 20:34 UTC
37b7dd0 Set default value for tracer/composition flags earlier 19 September 2007, 20:34 UTC
a3387ca Shrink the size of arrays 19 September 2007, 20:33 UTC
3488a3f Shrink the size of arrays 19 September 2007, 20:31 UTC
37e37f8 Fixed uninit'd heat flux and geoid arrays bugs, as reported by valgrind 19 September 2007, 20:30 UTC
dec899b Minor bug fix in Output_gzidir, output_heating function didn't work. 19 September 2007, 00:28 UTC
c89da81 Fixed a memory bug in checkpoint 18 September 2007, 23:08 UTC
995fff8 Setup composition for regional model 18 September 2007, 23:07 UTC
f23ffbf Fix for Issue129: "CitcomS configured without exchanger still links with -lExchanger". Also, use CIT_HEADER_MPI so that configuring --with-exchanger using MPICH2 works (see Issue14 and Issue57). 18 September 2007, 01:52 UTC
a2fd2a4 Generalized init. composition from neighbors for multi-component composition 17 September 2007, 20:30 UTC
c10e388 Using neighboring elements to determine the initial composition, if only a few elements are empty. 17 September 2007, 19:07 UTC
e1af673 Revert accidental commit in part of r7957 15 September 2007, 00:17 UTC
94cd779 Rearranged computation of latent heat and the test condition of phase change 14 September 2007, 23:39 UTC
fedce4a Added the contribution of latent heat to the diffusion term 14 September 2007, 23:11 UTC
ffe1a6f Replace abs() by fabs() 14 September 2007, 23:11 UTC
d4bc483 Consevert energy (sum of rho*cp*T) instead of temperature (sum of T) in the Lenardic filter, also fixed a long-standing bug (abs -> fabs) 14 September 2007, 23:10 UTC
d13bdd5 Changed the magic numbers (0.0 and 1.0) to const variables, reindent the code. 14 September 2007, 23:09 UTC
9189bab Comment out adiabatic temperature profile for now, since it is not used in the code. 13 September 2007, 22:36 UTC
fcd95c8 Fixed a problem when the input string is empty 12 September 2007, 19:24 UTC
249d790 Added heating file to optional output 12 September 2007, 19:24 UTC
cb39951 Keep the load vector intact in the bicg solver 12 September 2007, 19:24 UTC
0b02b96 Using drand48() instead of rand() to improve the quality of random tracers. Added some debugging output if the tracer count is low. 12 September 2007, 19:23 UTC
f8219a2 Add new input boolean parameter 'itracer_warnings'. Default to on. When it is off, the exit condition of tracer is relaxed. 12 September 2007, 19:22 UTC
6b934e2 Removed the test for F=0 load vector bug. The bug was fixed in r7940. 12 September 2007, 19:19 UTC
d6e4a43 Fixed tsolver. Fixed a bug in TMass and another in convert element # to nz. 07 September 2007, 22:53 UTC
0d53949 Fixed a few memory problems. Shrink the size of several arrays related to Stokes eqn. Moved jacobi() to Obsolete.c 05 September 2007, 22:35 UTC
7273549 Forgot to update surfacee_tem 05 September 2007, 22:21 UTC
c77dfd0 Set the min of 2nd strain rate invariant to 1e-16 to prevent infinite viscosity when SDEPV=on 05 September 2007, 22:20 UTC
681786d Init'd timestep variables 05 September 2007, 22:20 UTC
e3ccb79 Added test for F=0 load vector bug, moved open_qfiles to after set_elapsed time. 04 September 2007, 23:09 UTC
e3f6845 Forgot to update the makefile 29 August 2007, 23:50 UTC
4b3831f Removed unused velocity bc files 29 August 2007, 23:39 UTC
f0d9ebe Restructing the repository layout for CitcomS documentation 29 August 2007, 23:18 UTC
e06f544 Forgot to update r7902 29 August 2007, 20:55 UTC
3711a41 Replaced output_open() with output_open_mode() Append the log and time files if restarting 29 August 2007, 20:48 UTC
a981bca Minor fix 29 August 2007, 20:47 UTC
d1f0b86 Missed to merge one modification after the conflict. 29 August 2007, 01:46 UTC
13492bf When checking in my changes, I had to resolve a conflict for lib/Instructions.c by hand, which I hope I did properly. Here are my changes: - renamed CONTOL structure members ORTHO and ORTHOZ to CITCOM_ORTHO and CITCOM_ORTHOZ Those were never used and conflicted with definitions in GMT gmt.h - Added a higher frequency heat flow output option like so: write_q_files=1 # option to write heat flux to files qt.dat and qb.dat # at intervals smaller than storage_spacing (0) - Added the capability to read in initial temperatures from netcdf grd files, if -USE_GGRD is used. Lot of options, like so: # # read initial temperature conditions from grd files (default values in parentheses) # tic_method=4 # read initial temperature from netcdf GRD files (off) ggrd_tinit_scale_with_prem=off # scale the temperature with PREM densities (off) ggrd_tinit_scale=1.0 # scaling factor to apply to read in scalars f (1.0) ggrd_tinit_offset=-0.5 # offset, T = f * scale + offset + tm (0.0) # where tm is the mean between top and bottom TBC values # if the bottom is flux, will use 1 for bottom TBC value ggrd_tinit_gfile="../../data/tomography/s20a_smean_new_age/t" # prefix of grd files, will # try to read t.1.grd, t.2.grd ... t.n.grd # where n is the number of layers in the depth file ggrd_tinit_dfile="../../data/tomography/s20a_smean_new_age/tdepth.dat" # file with layer depths in km from bottom up ggrd_tinit_prem_file="../../progs/src/hc-svn/prem/prem.dat" # PREM data file ggrd_tinit_override_tbc=on # override temperature boundary conditions (off) 29 August 2007, 00:51 UTC
3ba2a04 Defaulting datadir_old to the current directory 28 August 2007, 01:10 UTC
220393d Added various heating term to the energy equation. A new input parameter: solver.surfaceT for non-dimensional surface temperature. Removed old (and unused) input parameter: solver.const.surf_temp. 28 August 2007, 01:09 UTC
557acb3 Fixed a index bug in stokes solver 27 August 2007, 19:34 UTC
8347039 Add pyre input for new parameters introduced by TWB 27 August 2007, 19:34 UTC
1a41a1d Clean up the code and declared function prototypes to avoid compiler warnings 27 August 2007, 19:34 UTC
12591d6 Compute total visc and adi heating 23 August 2007, 20:20 UTC
fce4d3d Minor refactoring 23 August 2007, 20:19 UTC
4bc0ea7 Compute phase change wrt reference density and gravity profiles. 23 August 2007, 20:18 UTC
d0098c3 Avoid reading unused input parameters 23 August 2007, 20:18 UTC
1268a26 assert solver.tsolver.monitor_max_T is off in coupled run and remove debugging output 23 August 2007, 20:17 UTC
ae9b86a Renamed E->control.filter_temperature to E->advection.filter_temperature 23 August 2007, 20:16 UTC
8233668 Added new input parameter solver.tsolver.monitor_max_T When it is on (default), if the max temperature changes too much between timestep, the temperature field is restored at the tsolver is called using half of the timestep size. 23 August 2007, 20:16 UTC
447f924 Removed unused var. in struct advection 23 August 2007, 20:15 UTC
80a070b A new way to compute TMass 23 August 2007, 20:14 UTC
6d31d51 Minor refactoring: add comments, rename variables 23 August 2007, 20:14 UTC
a169ee5 Added an option to remove net rotations of the whole model before ascii-gz output. This might be something of interest for other output options as well, but is still experimental. Obvisouly, NR motions could be removed in a post-processing step, but it is useful to monitor how those develop during a model run. (Particularly pronounced for strong lateral viscosity variations based on composition.) 22 August 2007, 01:00 UTC
7616c27 Removed the 2nd exit condition for bicg solver because the convergence criterion is hard to determine. 21 August 2007, 22:42 UTC
239522f Fixed the scaling constants in geoid calculation with reference state 21 August 2007, 22:41 UTC
eb5aedc Disable phase change in compressible flow 21 August 2007, 22:41 UTC
2473c52 Added Wei Leng to author list 21 August 2007, 20:28 UTC
1420d4e A new input parameter to specify which compressible solver If solver.vsolver.uzawa == "cg", Wei Leng's implementation is used. If solver.vsolver.uzawa == "bicg", Eh Tan's implementation is used. 21 August 2007, 20:27 UTC
dee9a51 Refactored non-newtonian stokes solver and iter-CG compressible stokes solver 21 August 2007, 20:26 UTC
9aa0d94 Forgot this file. 20 August 2007, 19:44 UTC
9370fcf Added #ifdef block around vtk_io in output_finalize() 20 August 2007, 18:11 UTC
316e0b2 Made CDEPV and Q0.enriched work only with ncomp=1 for now. 20 August 2007, 00:11 UTC
e53e0c8 - merged changes with Eh's additions - added coor=2 option (for full citcom) to assign finer radial node spacing to top and lower layers of shell. The coor_refine=0.1,0.15,0.1,0.2 parameter specifies the radius fraction of the bottom layer [0], the fraction of the nodes in this layer [1], the top layer fraction [2], and the top layer node fraction [3]. I.e. the defaults will put 15% of all nz nodes into the 10% lower layer, 20% in the top 10% upper layer, and the rest in between. - renamed gzipped output option ascii-gz - built in restart facilities for temperature and tracers when using gzdir I/O - added a composition viscosity function, CDEPV, based on two tracer flavors - for this to work, I had to move viscosity_input() *behind* tic_input() and tracer_input() in instructions - added tracer_enriched option. If tracer = on and tracer_enriched = on, will reader Q0_enriched and vary the element heat production between Q0 for C = 0 and Q0_enriched for C = 1. I.e. this only works if C varies between 0 and 1. - added an experimental option to write to a single VTK file (note caveats!) if ascii-gz is activated gzdir_vtkio = 2, will try to write VTK straight (experimental) the VTK output is of the "legacy", serial type, and requires that all processors see the same filesystem. This will likely lead to a bottleneck for larg CPU computations as each processor has to wait til the previous is done. 19 August 2007, 23:54 UTC
5cbae2e Added excutable permission to testing shell script 17 August 2007, 21:41 UTC
cac7d92 Added comments to restart example 17 August 2007, 21:39 UTC
fec6071 Converting TWB's ascii-gz and vtkio output and CDEPV and tracer_enriched for multi-component chemical convection. 17 August 2007, 18:35 UTC
5345880 Multi-component chemical convection. By setting tracer_flavors > 2, one can enable multi-component chemical convection. See tests/checkpoint/tracer.cfg for example input. 17 August 2007, 18:33 UTC
8f2682a Merging the checkpoint branch to trunk. The checkpoint file is called [datafile].chkpt.[rank].[step] and is outputed with an interval of controller.checkpointFrequency. To read back the checkpoint file, set solver.ic.restart=on. The old checkpoint file, [datafile_old].chkpt.[rank].[solution_cycles_init] will be read in. The old way of restarting (read in temperature from the velo files) will still be available through solver.ic.tic_method=-1. Setting solver.tracer.ic_method=2 will read the old *.tracer.* files as before, but the composition field will be calculated according to the tracers, not read from the *.comp_el.* files. The parameter solver.tracer.reset_initial_composition becomes obsolete. 16 August 2007, 21:36 UTC
7c82fa3 When the element is close to the poles, use a more precise method to compute the strain rate. 16 August 2007, 20:10 UTC
2343a70 Forgot to add two files to the repository 16 August 2007, 19:38 UTC
61655f0 Merging compressible branch to trunk. The TALA solver is working. Two new input parameters: dissipation_number and gruneisen. 16 August 2007, 19:28 UTC
33d700d Updated Cookbook5 so that it uses only 2 processors. 16 August 2007, 01:42 UTC
c3fdb31 Merging compressible branch (up to r6584) to trunk. 16 August 2007, 00:27 UTC
fe53297 - added coor=2 option (for full citcom) to assign finer radial node spacing to top and lower layers of shell. The coor_refine=0.1,0.15,0.1,0.2 parameters specify the radius fraction of the bottom layer [0], the fraction of the nodes in this layer [1], the top layer fraction [2], and the top layer node fraction [3]. I.e. the defaults will put 15% of all nz nodes into the 10% lower layer, 20% in the top 10% upper layer, and the rest in between. - renamed gzipped output version with sub-directory storage ascii-gz - built in restart facilities for temperature and tracers when using ascii-gz I/O with vtkio != 2 - added a composition viscosity function, CDEPV, based on two tracer flavors - for this to work, I had to move viscosity_input() *behind* tic_input() and tracer_input() in instructions - added tracer_enriched option for internal heating. If tracer = on and tracer_enriched = on, will reader Q0_enriched and vary the element heat production between Q0 for C = 0 and Q0_enriched for C = 1. I.e. this only works if C varies between 0 and 1. - added an option to write from all processros to a single VTK file, if ascii-gz is activated, and vtkio = 2. The VTK output is of the "legacy", serial, single-file type, and requires that all processors see the same filesystem. This will lead to a bottleneck for large # of CPU computations as each processor has to wait til the previous is done. More efficient I/O should be possible by using the distributed storage version of VTK, but I have no clue how this works. Anyone? 14 August 2007, 03:33 UTC
d6e512c Minor refactoring, clean-up and moving functions between files 13 August 2007, 22:55 UTC
45a5e52 Moved obsolete functions phae_change_* from Phase_change.c to Obsolete.c and declared private functions to be static 13 August 2007, 21:43 UTC
408e100 Update the author list 10 August 2007, 23:56 UTC
d603970 additions per becker, tan2 10 August 2007, 23:56 UTC
d9f152e Fixed two type-mismatch errors. Issue127 is fixed by this change and r7802. 10 August 2007, 23:44 UTC
f036e33 Prevent the original force vector being modified by Uzawa solver. This is needed for the non-Newtonian iterations. 10 August 2007, 23:43 UTC
da4b2ca - renamed gzipped output option ascii-gz - built in restart facilities for temperature and tracers when using gzdir I/O - added a composition viscosity function, CDEPV, based on two tracer flavors - for this to work, I had to move viscosity_input() *behind* tic_input() and tracer_input() in instructions 10 August 2007, 20:51 UTC
a77c186 - added a compositional viscosity function, CDEPV, based on two tracer flavors - for this to work, I had to move viscosity_input() *behind* tic_input() and tracer_input() in Instructions.c - built in restart facilities for temperature and tracers when using gzdir I/O 09 August 2007, 01:27 UTC
8e448b0 - added new routine Ggrd_handling which will take care of netcdf grd input. For now, only the tracer innitial conditions can be set by using, for example, ictracer_grd_layers=2 ictracer_grd_file="tmask.grd" for this to work, the ggrd libraries from HC, as well as the -lgmt and -lnetcdf need to be linked, and -DUSE_GGRD needs to be defined on compile time. - rewrote TDEPV viscosity options 3 and 4 slightly to make them more efficient - added TDEPV viscosity option 6 eta = N_0 exp(E(T_0-T) + (1-z) Z_0 ) - took layers subroutine in Construct_arrays.c apart to allow calling the same function in a more flexible from other subroutines - addded error handling routine "myerror" to Pan_problem_misc_functions - made the output behavior of the multigrid solver convergence parameters consisent with verbosity setting (ie. use report instead of always stderr output) - changed tracer input parameter "tracer" and "chemical_buoyancy" to boolean routine for added flexibility (I hope) - made check_initial_composition consistent with tracer advection, exits after warning only for E->trace.itracer_warnings==1 - debugging output in Full_tracer_advection.c and Tracer_setup only gets printed to file for E->control.verbose == 1 - fixed some typos in Output_gzdir (that might have been introduced by asynchronous svn updating or something like that) 08 August 2007, 22:55 UTC
back to top