| 573d06e | Eh Tan | 28 November 2006, 03:34:23 UTC | Chapter 3 updated | 28 November 2006, 03:34:23 UTC |
| 80c8616 | Leif Strand | 28 November 2006, 03:01:37 UTC | Changes related to Pythia v0.8.1.0b7, which CitcomS now requires: 'journal' is no longer a part of an application's inventory, and pyre.applications.start() is now responsible for calling run(). | 28 November 2006, 03:01:37 UTC |
| f503267 | Eh Tan | 28 November 2006, 02:57:07 UTC | Updated the dimensional const | 28 November 2006, 02:57:07 UTC |
| 2d35142 | Eh Tan | 28 November 2006, 02:55:44 UTC | Updated to be consistent with the manual | 28 November 2006, 02:55:44 UTC |
| 61249ec | Eh Tan | 27 November 2006, 23:09:42 UTC | Changed the default value of "datadir" to "." | 27 November 2006, 23:09:42 UTC |
| 944d4da | Eh Tan | 27 November 2006, 22:37:27 UTC | Changed postprocessing scripts for the new "datadir" parameter | 27 November 2006, 22:37:27 UTC |
| 510c2e5 | Eh Tan | 27 November 2006, 21:27:15 UTC | Fix for POSIX-compliant arithmetic expression The "sh" shell on my Ubuntu 6.10 doesn't have the "let" command. That's how I found that "let" is not in the POSIX standard. Use "$((expression))" instead. | 27 November 2006, 21:27:15 UTC |
| baf9708 | Eh Tan | 27 November 2006, 20:05:42 UTC | Added extern program to determine "datadir" | 27 November 2006, 20:05:42 UTC |
| 2cd498c | Eh Tan | 27 November 2006, 20:05:11 UTC | Changes how to specify output directory * "output_format" must be either "ascii" (the default) or "hdf5", the "ascii-local" option is gone * "datafile" becomes the prefix of the filename and cannot contain "/" * "datadir" becomes mandatory input parameters * If "datadir" is "%DATADIR", the output directory is determined by an external program "citcoms_datadir" * Special substrings "%RANK", "%HOSTNAME" in datadir will be substituted by mpi rank, machine's hostname, $DATADIR environment variable, respectively. * These special strings start with "%", rather than "$", becuase it's very difficult to input "$" from the command line. | 27 November 2006, 20:05:11 UTC |
| eaffb18 | Eh Tan | 22 November 2006, 21:26:09 UTC | Changed the convergence criterion of Stokes solver: * stop iterations if delta(v) or delta(p) becomes too small (i.e., v or p has converged) * this is suggested by Shijie Zhong | 22 November 2006, 21:26:09 UTC |
| 26f0c05 | Eh Tan | 22 November 2006, 21:25:47 UTC | Ignoring overlapping nodes when counting # of global nodes | 22 November 2006, 21:25:47 UTC |
| 5c19bff | Sue Kientz | 22 November 2006, 20:25:40 UTC | typo; made references into an itemized list | 22 November 2006, 20:25:40 UTC |
| 32fc3c9 | Eh Tan | 22 November 2006, 20:05:23 UTC | Removed unused codes and files | 22 November 2006, 20:05:23 UTC |
| 2d8467a | Eh Tan | 22 November 2006, 19:41:43 UTC | Refactored geoid calculation | 22 November 2006, 19:41:43 UTC |
| 8463943 | Luis Armendariz | 21 November 2006, 23:03:38 UTC | Expanded section on optimizing parallel I/O. | 21 November 2006, 23:03:38 UTC |
| 91c9c75 | Eh Tan | 21 November 2006, 20:18:02 UTC | Added scripts for 2D temperature plots, using GMT commands | 21 November 2006, 20:18:02 UTC |
| 201713b | Eh Tan | 20 November 2006, 19:24:40 UTC | Forgot to declare function prototype | 20 November 2006, 19:24:40 UTC |
| 401efaa | Eh Tan | 18 November 2006, 05:17:07 UTC | Disabled a debugging output, which is not protected by a "if(E->control.verbose)" and caused Segmentation Fault. | 18 November 2006, 05:17:07 UTC |
| 1433785 | Leif Strand | 18 November 2006, 01:29:12 UTC | The function loadObject() has been moved from Pythia to Merlin. | 18 November 2006, 01:29:12 UTC |
| 7d6ea27 | Eh Tan | 18 November 2006, 00:48:49 UTC | Updated the vbc files and input file | 18 November 2006, 00:48:49 UTC |
| a4596a5 | Eh Tan | 18 November 2006, 00:41:52 UTC | Updated the vbc files and input file. | 18 November 2006, 00:41:52 UTC |
| 63a29c8 | Eh Tan | 18 November 2006, 00:20:26 UTC | HDF5 output for geoid is working | 18 November 2006, 00:20:26 UTC |
| 4b8c00e | Luis Armendariz | 17 November 2006, 22:37:12 UTC | Cleaned up 'svn status' by updating svn:ignore property | 17 November 2006, 22:37:12 UTC |
| 1d9227a | Eh Tan | 17 November 2006, 20:25:45 UTC | HDF5 output of surf, botm, horiz_avg are working now. * Time and timestep infomation is written as scalar attributes "time" and "timestep" in / group. * Set MPI_Info according to input parameters "cb_block_size" and "cb_buffer_size" * I decided not to write surf, botm, and horiz_avg coordinates. This coordinate can be retrieved from the 3D coordinate very easily by this pseudo-code: surf_coord = coord[cap=all, x=all, y=all, z=1] z_coord = coord[cap=1, x=1, y=1, z=all] Remainging works: geoid for hdf5, update the manual | 17 November 2006, 20:25:45 UTC |
| c947d7a | Eh Tan | 17 November 2006, 06:26:37 UTC | HDF5 output reloaded. See issue4 for reasoning. Most data outputs are working. Remaining works: * How to deal with time output? (Maybe not output at all? Time info is always written in the .time file anyway.) * Where to put 2d and 1d coordinate output? (Maybe output with the data? These datasets are small and won't take too much space anyway.) * Geoid output | 17 November 2006, 06:26:37 UTC |
| 30bb98f | Leif Strand | 17 November 2006, 05:34:42 UTC | Fixed a minor bug in Output.py. | 17 November 2006, 05:34:42 UTC |
| 7da74a3 | Leif Strand | 17 November 2006, 02:24:51 UTC | Revised everything having to do with runtime configuration: launcher.xxx, scheduler.xxx, and job.xxx. Lowered the prominence/importance given to .pml files... and resisted my temptation to delete all reference to them :-) (I think I'm pretty much done with the v2.1 manual at this point. Reading the manual makes me tired, because it reminds me of how much work remains to be done on the code itself. Ease-of-use could be substantially improved...) | 17 November 2006, 02:24:51 UTC |
| 86701e5 | Leif Strand | 16 November 2006, 21:58:42 UTC | Reverted another recent decision of mine: Always build the legacy drivers, 'CitcomSFull' and 'CitcomSRegional'. I took these out of the Pyre build to avoid confusion; but now -- thinking of shared installations of CitcomS on clusters -- I think they should always be built (as they were before) so that users have the option. | 16 November 2006, 21:58:42 UTC |
| 6d6764b | Leif Strand | 16 November 2006, 21:36:50 UTC | Install examples under PREFIX/share/CitcomS/examples, as now documented in the manual. This is for the benefit of users of shared installations of CitcomS. I also sync'd the file list in examples/Makefile.am. Note: the file velbc_regional.tar.gz seems to be corrupt!!! | 16 November 2006, 21:36:50 UTC |
| d5b200b | Leif Strand | 16 November 2006, 21:25:54 UTC | Corrections with respect to r5302: the binaries/scripts are now once again built under the 'bin' subdirectory. Also reviewed all references to PREFIX. | 16 November 2006, 21:25:54 UTC |
| a1e12dc | Leif Strand | 16 November 2006, 14:40:41 UTC | Re-created 'bin' subdirectory to avoid filename conflict on case-sensitive filesystems (Mac, Windows): "CitcomS" == "citcoms". Fixed egg-related bugs which prevented 'configure' from succeeding when buildir != srcdir (another BuildBot test vector). Call addsitedir() in the 'citcoms' script. I guess this is now necessary since "site.py" is no longer in the same directory as "citcoms". (*) Note that there is currently a bug in Merlin which prevents 'citcoms' from running ("No module named pkg_resources"). The work-around is to manually add the merlin egg to the merlin-install.pth file. There are more bugs which prevent running on Mac OS X at all (still investigating). (*) I'm becoming annoyed with setuptools' use of .pth files as a sys.path/PYTHONPATH persistence mechanism, mainly because of the accompanying hacks which are used to turn ordinary directories into "site" directories (i.e., directories where .pth files work) -- these ugly, confusing hacks include the fake site.py, "virtual python", etc. I think Merlin will only use .pth files in real site directories. It will use clear, straitforward, direct sys.path manipulation everywhere else. | 16 November 2006, 14:40:41 UTC |
| 2642094 | Eh Tan | 16 November 2006, 03:04:21 UTC | Forgot to commit module/ directory | 16 November 2006, 03:04:21 UTC |
| 6602642 | Eh Tan | 16 November 2006, 03:01:59 UTC | Declared function prototype at the top | 16 November 2006, 03:01:59 UTC |
| 3cfe216 | Eh Tan | 16 November 2006, 02:07:09 UTC | Forgot to commit python files | 16 November 2006, 02:07:09 UTC |
| 785021d | Eh Tan | 16 November 2006, 02:07:00 UTC | Declared function prototype at the top | 16 November 2006, 02:07:00 UTC |
| 1bd5e68 | Eh Tan | 16 November 2006, 01:52:25 UTC | Implemented geoid calculation, see issue64. User visible changes include: * "output_optional" can include string "geoid", which will enable geoid output * new input parameters: radius, density_above, density_below * removed input parameters: layerd, wdensity, llmax, nlong, nlati * moved input parameter "output_ll_max" from Sphere.py to Output.py * sync'd the defalut values of input parameters in Const.py and Instructions.c TODO: h5output_geoid() not finished | 16 November 2006, 01:52:25 UTC |
| 4aa72aa | Leif Strand | 15 November 2006, 22:01:35 UTC | Elaborated a bit on .egg download, and clarified that CitcomS needs Pythia v0.8.1.x (instead of 0.8). | 15 November 2006, 22:01:35 UTC |
| f245f79 | Leif Strand | 15 November 2006, 21:18:52 UTC | Re-read the stuff I worked on yesterday, making minor corrections: "is in your PATH" -> "is on your PATH", inserted the word "Section" before section references, other minor changes. | 15 November 2006, 21:18:52 UTC |
| 465cdc7 | Sue Kientz | 15 November 2006, 20:40:13 UTC | a few typos fixed, http removed from htmlurl tags (not needed and lengthen name in line); pdf regen | 15 November 2006, 20:40:13 UTC |
| 0693207 | Leif Strand | 15 November 2006, 02:06:20 UTC | Corrections to the section, "Changing Parameters". | 15 November 2006, 02:06:20 UTC |
| 901d882 | Leif Strand | 15 November 2006, 01:14:49 UTC | Updated the installation chapter. | 15 November 2006, 01:14:49 UTC |
| 80a98d3 | Leif Strand | 14 November 2006, 03:25:28 UTC | Pythia v0.8.1.0b4: Fixed bugs relating to MPI launching: Fix sys.path/PYTHONPATH automatically; don't touch 'argv' until after MPI_Init(). Next, I may try to slurp 'pycitcoms.c' into the framework... it contains a lot of implementation details which have everything to do with Pyre and nothing to do with CitcomS. | 14 November 2006, 03:25:28 UTC |
| 4bd3a75 | Leif Strand | 11 November 2006, 03:34:08 UTC | Use the new merlin (the setuptools clone). Download Python dependencies automatically during 'configure' (This doesn't work for "make install" yet.) The Dynamo BuildBot might even succeed after this check-in... fingers crossed... | 11 November 2006, 03:34:08 UTC |
| 79e53b1 | Sue Kientz | 10 November 2006, 19:20:56 UTC | created formulas for sec a.1.11 | 10 November 2006, 19:20:56 UTC |
| 8cac7ae | Sue Kientz | 10 November 2006, 18:54:31 UTC | cookbook 6 graphic edited, various small tweaks | 10 November 2006, 18:54:31 UTC |
| fdd2455 | Sue Kientz | 10 November 2006, 01:59:45 UTC | finished various typos and fixes after checking with Luis on questions | 10 November 2006, 01:59:45 UTC |
| 897706c | Sue Kientz | 09 November 2006, 19:37:41 UTC | various typos and fixes | 09 November 2006, 19:37:41 UTC |
| 967ea83 | Eh Tan | 07 November 2006, 19:33:31 UTC | * Fixed a few typos * Added "TODO" in front of all unfinished text. | 07 November 2006, 19:33:31 UTC |
| f89ab8a | Leif Strand | 07 November 2006, 02:30:04 UTC | Moved CitcomS HDF5 macros to cs/autoconf. | 07 November 2006, 02:30:04 UTC |
| 02fbe2f | Leif Strand | 07 November 2006, 01:33:06 UTC | Took a step backwards so that I can take a more conservative approach to build/install. Also, temporarily deleted 'Exchanger' and all references to it, to avoid confusion. We can always go back in time and retrieve it later. | 07 November 2006, 01:33:06 UTC |
| c20d5e5 | Sue Kientz | 07 November 2006, 00:49:42 UTC | pdf generated | 07 November 2006, 00:49:42 UTC |
| bd49eb5 | Sue Kientz | 07 November 2006, 00:49:04 UTC | gurnis edits, new pdf | 07 November 2006, 00:49:04 UTC |
| b93974f | Sue Kientz | 06 November 2006, 18:53:08 UTC | added authors to cover | 06 November 2006, 18:53:08 UTC |
| 2dbb7e7 | Sue Kientz | 06 November 2006, 18:33:16 UTC | added new section 1.7 per gurnis and supporting graphic | 06 November 2006, 18:33:16 UTC |
| 8bab2a8 | Luis Armendariz | 01 November 2006, 20:47:53 UTC | Re-typed most of the changes from my copy of the citcoms.lyx file | 01 November 2006, 20:47:53 UTC |
| b472059 | Eh Tan | 27 October 2006, 21:26:23 UTC | Removed legacy file .cvsignore | 27 October 2006, 21:26:23 UTC |
| 6cdac9f | Eh Tan | 27 October 2006, 21:25:51 UTC | Removed legary file .cvsignore | 27 October 2006, 21:25:51 UTC |
| ff62a3e | Eh Tan | 20 October 2006, 19:36:03 UTC | This file (examples/.reset-temporary) was checked into svn at the initial import by me. I don't remember what this file was used for. | 20 October 2006, 19:36:03 UTC |
| 47c4c2b | Eh Tan | 20 October 2006, 19:00:27 UTC | * Removed extra output_time(), which was used for profiling * Converted tab to 4 spaces | 20 October 2006, 19:00:27 UTC |
| caf7f85 | Leif Strand | 20 October 2006, 08:19:58 UTC | Fail if MPI lib/header is not found. | 20 October 2006, 08:19:58 UTC |
| 19f0a1b | Leif Strand | 17 October 2006, 11:00:23 UTC | Fixed legacy build. | 17 October 2006, 11:00:23 UTC |
| 2e201da | Leif Strand | 17 October 2006, 09:18:15 UTC | Transitioning to Merlin. (Warning: still under construction.) | 17 October 2006, 09:18:15 UTC |
| addc0ee | Eh Tan | 14 October 2006, 00:29:17 UTC | Undid the accidental checkin at r4998 | 14 October 2006, 00:29:17 UTC |
| d6ccdc2 | Eh Tan | 14 October 2006, 00:27:09 UTC | Renamed reference of "average" to "horiz_avg" | 14 October 2006, 00:27:09 UTC |
| a16042e | Eh Tan | 14 October 2006, 00:06:56 UTC | Added description on new features and bug fixes for v2.1.0 | 14 October 2006, 00:06:56 UTC |
| 73c5c0d | Luis Armendariz | 13 October 2006, 23:59:48 UTC | Settled on /horiz_avg/ instead of /have/ | 13 October 2006, 23:59:48 UTC |
| 2d20c72 | Sue Kientz | 13 October 2006, 23:46:36 UTC | updated url handling | 13 October 2006, 23:46:36 UTC |
| bfdbda6 | Eh Tan | 13 October 2006, 23:18:11 UTC | * Updated for the new layout of HDF5 group * Changed the order of nodes, so that the output is the same as the cap file from autocombine.py | 13 October 2006, 23:18:11 UTC |
| d87840e | Eh Tan | 13 October 2006, 07:39:03 UTC | * Fixed a typo "PHDF HOME" -> "PHDF_HOME" * Added a section on using none-pyre version * Undid a xref in section 3.1, which should link to (not-checked-in-yet) "Working with HDF5" chapter * Updated Appendix A and changed the title * Added description on new output files (stress, pressure, average) | 13 October 2006, 07:39:03 UTC |
| ea97058 | Eh Tan | 13 October 2006, 06:43:36 UTC | Gave default values to several parameters | 13 October 2006, 06:43:36 UTC |
| b3ab31f | Eh Tan | 13 October 2006, 04:15:53 UTC | Removed obsolete input parameter "maxadvtime" from the input samples | 13 October 2006, 04:15:53 UTC |
| 0e734d9 | Eh Tan | 13 October 2006, 04:10:32 UTC | Moved input parameters "aug_lagr" and "aug_number" from tsolver to vsolver, where they really should belong to. These parameters are for augmented lagragian stiffness matrix (I have no idea on what it is...), as indicated in Element_calculations.c. | 13 October 2006, 04:10:32 UTC |
| 701b1ec | Eh Tan | 12 October 2006, 21:43:09 UTC | 'half_space_age' was not used when 'tic_method'=1. So, fixed the comment and removed the conditional input | 12 October 2006, 21:43:09 UTC |
| fa4793e | Leif Strand | 12 October 2006, 14:48:43 UTC | Moved 'launcher' and '.cfg' support out of CitcomS and into the framework. As I was revising the "Installation" chapter of the manual (still in progress), I got frustrated with how complicated everything was -- and how difficult it was to document that complexity -- so I did something slightly violent: Converted the CitcomS extension module, 'CitcomSLib', from C++ to C. This is made possible by Pythia 0.8.1.0b2, which replaces _mpimodule with _mpi.pyx, which has a C interface. It is now possible to build CitcomS -- and all its prerequisites (e.g. Python) -- entirely from source, starting with nothing but a C compiler. Exchanger still requires C++, of course. I have yet to test these changes with Exchanger enabled, but I don't anticipate any serious problems. This may have been overkill, but I also cleaned the source of C++-style comments (even though most modern C compilers probably accept them... it became standard in C99 or something). I'll bet you can install CitcomS with GCC 2.x now :-) | 12 October 2006, 14:48:43 UTC |
| 37db3e3 | Sue Kientz | 11 October 2006, 20:01:04 UTC | took care of some todo items | 11 October 2006, 20:01:04 UTC |
| 4791fdb | Eh Tan | 10 October 2006, 23:48:38 UTC | * Added subsection on installing OpenDXutils package * Aligned figures to the center * Changed the layout of the Appendices | 10 October 2006, 23:48:38 UTC |
| 21245be | Sue Kientz | 10 October 2006, 20:37:07 UTC | added hdf5 install and addt tools chapters | 10 October 2006, 20:37:07 UTC |
| 5717745 | Luis Armendariz | 10 October 2006, 18:42:58 UTC | Inserted a few TODO items | 10 October 2006, 18:42:58 UTC |
| 55ddf81 | Luis Armendariz | 10 October 2006, 13:00:23 UTC | 1. Reorganized routines in a more logical order. 2. Consolidated routines from PyTables.[ch] into Output_h5.c | 10 October 2006, 13:00:23 UTC |
| 509de08 | Luis Armendariz | 10 October 2006, 11:54:13 UTC | Renamed local function to conform with h5 prefix. | 10 October 2006, 11:54:13 UTC |
| 7f010dc | Luis Armendariz | 10 October 2006, 07:29:28 UTC | 1. Added PID to metadata to make it easier to identify the corresponding .cfg file 2. Wrapped most calls to set_attribute() to avoid using addresses as arguments 3. Added set_attribute_{int,float,double}_vector() functions | 10 October 2006, 07:29:28 UTC |
| 24fcb93 | Luis Armendariz | 10 October 2006, 00:52:53 UTC | Forgot to delete reference to obsolete data_group. | 10 October 2006, 00:52:53 UTC |
| 3ee3b7e | Eh Tan | 10 October 2006, 00:52:45 UTC | Set MIME type to pdf | 10 October 2006, 00:52:45 UTC |
| 2fc3b57 | Eh Tan | 10 October 2006, 00:51:15 UTC | Set MIME type to image | 10 October 2006, 00:51:15 UTC |
| b992618 | Eh Tan | 10 October 2006, 00:48:41 UTC | Set MIME type to image | 10 October 2006, 00:48:41 UTC |
| 5f2b697 | Luis Armendariz | 10 October 2006, 00:45:25 UTC | Shortened /average/ group into /have/ | 10 October 2006, 00:45:25 UTC |
| a7bc5d4 | Eh Tan | 10 October 2006, 00:42:50 UTC | Figure for HDF5 visualization | 10 October 2006, 00:42:50 UTC |
| 282370e | Eh Tan | 10 October 2006, 00:41:19 UTC | Added section on visualizing HDF5 output | 10 October 2006, 00:41:19 UTC |
| 3729221 | Luis Armendariz | 10 October 2006, 00:38:19 UTC | Fix mispelling in input parameters (allignment options). | 10 October 2006, 00:38:19 UTC |
| bdd582d | Eh Tan | 09 October 2006, 23:25:06 UTC | * Added documentaion of .pml file, which comes from DANSE FAQ * Added section on visualization of full version * Moved the discussion on .log and .time files from Section 4.2 to Section 3.5.3 * Updated links to equations | 09 October 2006, 23:25:06 UTC |
| 64710e9 | Sue Kientz | 09 October 2006, 23:14:47 UTC | change equation numbering | 09 October 2006, 23:14:47 UTC |
| 5248caf | Eh Tan | 09 October 2006, 20:41:31 UTC | Changed the order of input parameters | 09 October 2006, 20:41:31 UTC |
| 45b6126 | Eh Tan | 09 October 2006, 19:58:46 UTC | Updated for the new layout of hdf5 output | 09 October 2006, 19:58:46 UTC |
| f99228b | Michael Gurnis | 09 October 2006, 00:14:42 UTC | For Viscosity_structures.c, added a new rheology, rheol=5, that is a variant of rheol=3. The difference lay in the way that plate boundaries are weakened. Must set mat_control=1 and enter materials into the VIP array. Removed debug statements from Problem_related.c in preparation of new release. | 09 October 2006, 00:14:42 UTC |
| 5a1d516 | Leif Strand | 08 October 2006, 12:04:43 UTC | Pointed CitcomS egg-downloading at the new "Eggs" section on the web site, to insulate CitcomS from the imminent chaos in pythia-dev. | 08 October 2006, 12:04:43 UTC |
| db95387 | Eh Tan | 07 October 2006, 01:15:53 UTC | Updated for the new layout in the h5 file | 07 October 2006, 01:15:53 UTC |
| 9b4a1e9 | Luis Armendariz | 07 October 2006, 00:01:54 UTC | Eliminated data group in HDF5 file. Main datasets are now on the root-level. | 07 October 2006, 00:01:54 UTC |
| 731ed51 | Eh Tan | 06 October 2006, 21:10:34 UTC | The type of elt_g should be "higher_precision" instead of "double". | 06 October 2006, 21:10:34 UTC |
| e54baed | Eh Tan | 06 October 2006, 20:18:08 UTC | Changed the signature of the signal handler from (void)(interuption(void)) to (void)(interuption(int)) to conform with the C standard | 06 October 2006, 20:18:08 UTC |
| 15b4a0f | Eh Tan | 06 October 2006, 20:02:56 UTC | Missing braces around initializers for array structs node_point[] and offset[] | 06 October 2006, 20:02:56 UTC |
