The Gerris Flow Solver

File Release Notes and Changelog

Release Name: 0.9.2

Notes:
The most significant improvements in this release are:

- Important bug fixes and improvements for the multilevel Poisson
  solver can reduce the overall computational cost by a factor of five
  or more depending on the problem.
- Variable mesh resolution along solid boundaries is now allowed and
  has been well tested (note however that time-adaptive variable
  resolution is not allowed on solid boundaries yet).
- The robustness of the solution procedure has been much improved in
  the case of very complex solid boundaries.

Other improvements include:

- Debian package snapshots.
- Adaptive refinement of VOF-advected tracers.
- Preliminary implementation of CSF surface tension using Renardy El
  Ab. "proper discretisation".
- Solid boundaries can be refined according to the local curvature.
- Implicit Coriolis terms work with the Navier-Stokes solver.
- Support for "thin" 3D domains.


Changes:
Fri Mar 31 12:02:06 NZST 2006  Stephane Popinet <popinet@users.sf.net>
  * Removed obsolete pdflatex doc stuff

Thu Mar 30 17:40:12 NZST 2006  Stephane Popinet <popinet@users.sf.net>
  * Cells with solid volume fraction of 0 are removed

Fri Mar 24 16:40:02 NZST 2006  Stephane Popinet <popinet@users.sf.net>
  * Bug fix for relaxation on lowest level

Fri Mar 24 12:13:50 NZST 2006  Stephane Popinet <popinet@users.sf.net>
  * Implicit Coriolis should now work with the Navier-Stokes solver

Fri Mar 24 12:10:56 NZST 2006  Stephane Popinet <popinet@users.sf.net>
  * RefineDistance uses a lower-bound for the distance...
  
  ... rather than the distance from the center of the cell.

Wed Mar 15 17:20:29 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * Thin cells are removed
  
  "Thin" cells are topologically complex cut cells which lead to
  inaccurate volume-of-fluid representation of solid surfaces.  In some
  cases the presence of these cells could lead to instabilities in the
  projection. In all cases they would lead to inaccurate velocities.

Wed Mar 15 08:43:33 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * Updated tests for modified initial approx projection

Tue Mar 14 16:17:14 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * RefineSolid defines a new derived variable "SolidCurvature"

Tue Mar 14 16:15:46 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * New function gfs_solid_is_thin()

Tue Mar 14 16:12:21 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * New interface for gfs_domain_add_derived_variable()

Tue Mar 14 16:05:06 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * Fix for gradient computation at coarse/fine solid boundaries
  
  The previous version could use information from the wrong side of the
  solid surface when constructing interpolants for cells close to a
  solid boundary i.e. information was "leaking through" the solid
  surface. New weighting and checks with solid surface fractions should
  now avoid this.

Tue Mar 14 16:02:42 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * Merged cells can be formed using children cells at coarse/fine interfaces

Fri Mar 10 17:51:08 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * Time step is set before the first projection
  
  Previously the scaling for the tolerance on the divergence of the
  first projection was based on a timestep of unity (i.e. usually much
  larger than the subsequent timesteps).

Fri Mar 10 10:42:39 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * Debian snapshot packages were not correctly versioned

Wed Mar  8 17:37:00 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * Leave "specks" alone

Wed Mar  8 17:33:41 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * Updated boundaries and channel tests for weighted-centered pressure-gradient

Wed Mar  8 17:32:45 NZDT 2006  Stephane Popinet <s.popinet@niwa.co.nz>
  * Approximate projection uses fraction-weighted average pressure gradients
  
  The previous unweighted averages were unstable for the "thin plate"
  test case which has been added to the test suite.

Wed Mar  8 16:45:26 NZDT 2006  Stephane Popinet <s.popinet@niwa.co.nz>
  * Topology was not computed correctly for solid fractions computation in 3D

Sun Mar  5 20:59:59 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * Fix for special cases when computing solid fractions
  
  Solid fractions of cells cut more than once by the surface were not 
  consistently computed by the approximate "VOF" technique. This was causing
  convergence problems in some cases.
  
  These cells (rare) are now treated using a simple approximation.

Wed Mar  8 16:26:27 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * Removed obsolete "GtsInterface" stuff

Fri Mar  3 12:12:59 NZDT 2006  Stephane Popinet <s.popinet@niwa.co.nz>
  * Examples tarball did not include animations

Fri Mar  3 10:26:11 NZDT 2006  Stephane Popinet <s.popinet@niwa.co.nz>
  * Updated test cases for changes in projection

Tue Feb 14 21:01:54 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * gfs_domain_norm_residual() was not computing a dimensionless divergence
  
  It now returns the error in volume conservation per timestep relative
  to the volume of the cell.

Thu Feb  9 02:53:10 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * OutputSimulation has an optional "text" format

Thu Feb  9 02:52:41 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * Updated Debian dependencies

Mon Jan 30 14:47:49 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * New functions gfs_vof_plane() and gfs_vof_facet()

Sun Feb  5 14:21:11 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * Bug fix for inconsistent calls for divergence calculation

Sun Feb  5 14:20:02 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * OutputTiming also reports the maximum number of variables allocated

Sun Feb  5 14:15:27 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * Bug fix for "Pmac"
  
  Pmac was allocated as a temporary variable and thus was not
  interpolated from one timestep to the next when using adaptive
  refinement (causing trouble with the convergence rate of the Poisson
  solver). It is now a default variable.

Sun Feb  5 14:10:24 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * GfsAdapts should be treated as standard events
  
  ... in particular to avoid problems due to a changed order of events
  in output files.

Fri Feb  3 09:19:10 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * New "nitermin" param for multilevel solvers
  
  This is optional but useful in some cases to prevent noise building up in a 
  stationary solution (spurious currents test case for example).

Wed Feb  1 03:33:01 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * OutputTiming computes the total running time
  
  i.e. does not exclude the time taken by Outputs and Events.

Mon Jan 30 23:18:56 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * Bug fix for GfsVariable half-events
  
  Half-events were never processed for GfsVariables.

Mon Jan 30 22:38:17 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * VariableFiltered works with odd numbers of iterations

Fri Feb  3 23:50:52 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * Updated macros for automake 1.8

Sat Jan 28 06:43:49 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * Generalised gfs_correct_normal_velocities() for CSF surface tension

Sat Jan 28 14:23:57 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * New object GfsVariableCurvature

Sat Jan 28 10:33:58 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * MAC pressure is saved
  
  Both the MAC and approximate projections thus have better initial guesses. When
  looking for stationary solutions this makes a big difference.

Sat Jan 28 10:33:13 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * Simplification of Boussinesq example

Sat Jan 28 06:47:49 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * Fixed typo in spurious currents test case description

Sat Jan 28 06:46:55 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * Bug fix for GfsSourceDiffusionExplicit

Sat Jan 28 11:06:45 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * Bug fix for generic coarse_fine() method for GfsVariable

Fri Jan 27 23:08:33 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * Bug fix for special case of solid boundary
  
  When children mixed cells are all empty, a divide-by-zero would occur when
  computing the center of "mass" of the parent cell.

Fri Jan 27 23:07:10 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * Better checks for construction of merged cells lists

Sat Jan 21 12:29:32 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * VOF tracers are adaptively refined consistently

Sat Jan 21 10:45:13 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * Generic coarse_fine() method for GfsVariable

Thu Jan 19 02:55:04 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * New criterion for "small" cells
  
  Each direction is tested independently and the presence or absence of 
  neighbors is taken into account.

Tue Jan 17 03:42:08 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * S now returns normalised surface fraction

Tue Jan 17 02:27:31 NZDT 2006  Stephane Popinet <popinet@users.sf.net>
  * New object GfsOutputScalarMaxima

Wed Dec  7 13:29:57 NZDT 2005  Stephane Popinet <s.popinet@niwa.co.nz>
  * Fixed typo in tutorial

Tue Dec  6 10:26:46 NZDT 2005  Stephane Popinet <s.popinet@niwa.co.nz>
  * Updated test references
  
  GCC version 4 seems to produce slightly different results.

Mon Dec  5 11:45:18 NZDT 2005  Stephane Popinet <s.popinet@niwa.co.nz>
  * Updated debian packaging and docs

Fri Dec  2 18:06:32 NZDT 2005  Stephane Popinet <s.popinet@niwa.co.nz>
  * Debian package suggests gfsview

Fri Dec  2 16:33:19 NZDT 2005  Stephane Popinet <s.popinet@niwa.co.nz>
  * cvs2dist user proper versioning

Fri Dec  2 16:22:30 NZDT 2005  Stephane Popinet <s.popinet@niwa.co.nz>
  * Debian package snapshots

Wed Nov 30 22:01:47 NZDT 2005  Stephane Popinet <popinet@users.sf.net>
  * Debian package stuff (from Marcelo's official debian archive)

Tue Nov 29 16:23:49 NZDT 2005  Stephane Popinet <s.popinet@niwa.co.nz>
  * Fix for use of uninitialised heap info in adaptive refinement

Tue Nov 29 10:35:08 NZDT 2005  Stephane Popinet <s.popinet@niwa.co.nz>
  * Workaround for bug in latex2html

Mon Nov 28 16:17:25 NZDT 2005  Stephane Popinet <s.popinet@niwa.co.nz>
  * Updated release instructions