sem2dpack 2.3.0 beta

Download here (external)

Release Notes

SEM2DPACK is a spectral element method solver for 2D wave propagation and fracture dynamics, with emphasis on computational seismology and earthquake source dynamics. The main additional features in this new version (2.3.0) of SEM2DPACK are: (1) non-linear constitutive equations (continuum damage and Coulomb plasticity), (2) regularization for bimaterial faults, (3) linear slip faults for SH, (4) improved utilities for plotting and manipulating outputs, including a GUI. The documentation has been improved and a major code refactoring now allows easier implementation of new materials, source time functions, distributions and boundary conditions.

Change Log

Notation: *+ fixed bug
++ new feature
o+ optimization
+ minor modifications

Tracker IDs at SourceForge.net:
BR xxxxxxx : Bug Reports
FR xxxxxxx : Feature Requests
SR xxxxxxx : Support Requests


Thu Mar 6 18:11:49 JST 2008 -- v2.3.0

++ continuum damage (see MAT_DMG)
++ Coulomb plasticity (see MAT_PLAST)
++ normal stress response regularization for bimaterial faults
(see BC_DYNFLT_NORMAL block and EXAMPLES/Bimaterial_fault)
++ major refactoring, more commented code and guidelines for programmers in documentation
allow easier implementation of new materials, source time functions,
distributions and boundary conditions
++ Dirichlet and Neumann boundary conditions (see BC_DIRNEU block, supersedes BC_DT0TN0)
++ export triangulation file for GMT's "pscontour -T" (see PLOTS block)
++ monochromatic source time function (see STF_HARMONIC)
++ linear slip fault also for SH waves (see BC_LSF)
++ more and better documented Matlab utilities for plotting outputs,
including a GUI for snapshot views (see POST/*.m)
+ improved documentation
+ the syntax of the following input blocks has changed:
MATERIAL, SNAP_DEF (renamed PLOTS), SNAP_PS (renamed PLOTS_POSTSCRIPT),
DIST_GAUSSIAN, BC_ABSO, BC_DYNFLT (renamed BC_SWFFLT), GENERAL and SRC_DEF.
(ECHO block was removed)
+ KELVIN_VOIGT block renamed, new name is MAT_KV
+ ELASTIC block renamed, new name is MAT_ELAST
*+ fixed writing of file MeshNodesCoord_sem2d.tab
*+ fixed reference point in wave sources with incidence from above
*+ fixed postscript plotting with interpolation
SR 1746891
o+ more memory optimizations for box grids with homogeneous materials
o+ optimized matrix products


Wed Jun 6 17:32:34 CEST 2007 -- v2.2.11

*+ fixed obliquely incident plane waves (see SRC_WAVE input block)
FR 1676280
++ moment tensor sources (see new SRC_DOUBLE_COUPLE and SRC_MOMENT input blocks)
++ source time function interpolated (spline) from a user supplied file
+ allow source locations at interelement nodes
+ SRC_RICKER input block name changed to STF_RICKER
+ argument "Periodic" deleted from BC_XXX blocks (now handled internally)
+ interpolation of scalar snapshot plots
+ POST/movie.csh now crops the figure only upon request (-crop option)

Fri Jun 1 00:21:22 CEST 2007 -- v2.2.10
++ built-in mesh generation for layered medium
possibly with surface and sub-surface topography
See MESH_LAYERED input block
See also PRE/set_mesh_layers.m, EXAMPLES/TestLayers and
EXAMPLES/TestLayers_2
++ scripts POST/read_seis.m and POST/plot_seis.m to read
and plot seismogram outputs on Matlab
++ script for automated tests EXAMPLES/run_tests.csh
*+ fixed intent(in) bug and handling of Q9 elements
in fem_grid.f90 (FE_SetConnectivity)
*+ fixed array allocation bug in spline_dist.f90
o+ optimization settings in module constants.f90
should be modified by user before compilation
+ file format changed in DIST_SPLINE, DIST_LINEAR,
DIST_GRADIENT, REC_LINE and SRC_DEF :
the first line (which used to contain the number of data lines)
is not needed anymore

Wed Apr 4 10:21:55 CEST 2007 -- v2.2.9
*+ fixed reading of HETE_DIST1 input block
and more robust interpolation at the edges
BR 1692766
++ template for user-supplied source time function (src_user.f90)
See SRC_USER input block
FR 1691682
+ new example "LambsProblem", as in Mercerat et al. (2006)
SR 1681080

Mon Mar 12 05:06:31 CET 2007 -- v2.2.8
++ receivers can be located at arbitrary positions, read from a file,
not necesarily at computational nodes. See the REC_LINE block
FR 1639079
++ multiple sources, with location and delay-time read from a file.
See SRC_DEF block
FR 1675299

Thu Feb 8 16:52:09 CET 2007 -- v2.2.7
++ materials can have heterogeneous distributions using DIST_XXXXX blocks
(see EXAMPLE/InaBox)
SR 1631862
++ new DIST_HETE1 interpolates linearly from a regular grid (see EXAMPLE/InaBox)
*+ fixed bug when writing "Stability_sem2d.tab"
+ added a FAQ to the user's guide
SR 1653849 1648483

Thu Jan 18 16:21:36 CET 2007 -- v2.2.6
+ Kelvin-Voigt viscosity: fixed wrong format in initialization outputs
+ PRE/ViewMeshQuality.m: modified, see user's guide section 2.4
+ EXAMPLES/TestFlt1D/ : corrected Par.inp

Wed Nov 1 22:45:39 CET 2006 -- v2.2.5
++ symplectic time-stepping algorithms, PFR and PEFRL (4th order)
o+ Reverse Cuthill-McKee element reordering
+ PostScript snapshots for each component in P-SV
*+ abort if undefined boundary tag
*+ first sample in seismograms is t=0
*+ added Kelvin-Voigt to Newmark solver

Wed Jan 18 10:51:01 CET 2006 -- v2.2.4a
*+ Kelvin-Voigt routines fixed for g95 compiler

Mon Aug 22 22:55:56 CEST 2005 -- v2.2.4
++ SH waves (anti-plane, scalar wave equation), see "ndof=1" in &GENERAL
++ leapfrog (staggered centered-difference) is now the default time scheme
It is faster, uses less memory and allows accurate explicit treatment
of dynamic fault boundaries (see "kind" in &TIME)
++ Kelvin-Voigt material, only with leapfrog (see &KELVIN_VOIGT)
++ bc_abso: new implementation, absorbing boundaries are now second order in dt
*+ faults: mapping to the fault coordinate system was handled incorrectly
(this bug might have corrupted curved faults in P-SV)
*+ time scheme: consistent check for stability in dissipative schemes
o+ faster initialization (changed the node numbering algorithm)
o+ optimization: up to 50% performance gain by declaring NGLL as parameter
in core subroutines (ELAST_KD). Before compiling, the user must edit elastic.f90
setting OPT_NGLL to a frequent value.
o+ memory optimization: jacobian and local weights not stored
+ faults: input initial tractions renamed + new input initial stress
+ outputs: PostScript snapshots of color scalar fields
(set "vector=F, color=T" in PLOT_POSTSCRIPT)
+ post: sem2d_view.m for Matlab
+ post: "Snapshot_*.ps" renamed to "x_*.ps" where x=d,v or a.

Mon Feb 7 23:38:44 CET 2005 -- v2.2.3
+ outputs: binary formats, see "bin" option in PLOTS
field selector "fields"
stress and strain outputs
+ pre- and post-process: tools for visualization under gnuplot and Matlab,
for conversion of output to ascii (see PRE/ and POST/)
+ post: movie.csh runs now with animate (ImageMagick) instead of xanim
and gifsicle instead of gifmerge
+ doc: more about the method, mesh verification, output formats
*+ bc_swfflt: NSAMP in output header file was inaccurate
*+ mesh_cartesian, mesh_emc2 and init: bugs higlighted by g95 compiler (G.Wimpey)

lun Aug 9 20:15:09 EDT 2004 -- v2.2.2
*+ Intel fixed their bug in the "transfer" intrinsic,
the code now works with Intel compiler ifort, version 8.0.046_pe047.1
*+ fixed bug in the numbering algorithm
+ fixed POST/movie.csh

jeu nov 13 20:02:09 EST 2003 -- v2.2.1
*+ emc2, cracks and faults: bug fixed in the handling of crack tip nodes.
As a new convention, crack tip nodes must be assigned a special tag: -1.

mar oct 28 17:08:10 EST 2003 -- v2.2
++ boundary conditions: new kind BC_SWFFLT, slip weakening fault
*+ boundary conditions: bug fixed in the numbering subroutine
+ seismograms: fixed the scripts (don't need ShowSeis_sem2d.csh anymore)
+ time scheme: parameterized dissipation in explicit Newmark-alpha
o+ solver: optimized for memory
+ distributions: new kind PWCONR

jeu oct 23 15:13:48 EDT 2003 -- v2.1beta
++ new boundary condition: linear slip fault (BC_LISFLT input block)
+ "Periodic" flag in BC_ABSORB
+ doc: explanation of seismograms data format

lun oct 13 19:58:59 EDT 2003 -- v2.0