Hi Klaus, I was unable to find any additional information in the user guide on this point in the release notes:<div><br></div><div><blockquote class="gmail_quote" style="margin-top:0px;margin-right:0px;margin-bottom:0px;margin-left:0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">
<span style>The use of Gravitational acceleration is handled more accurately in Hydrodynamics.</span></blockquote><div><br></div><div>Could you clarify as to what the improvements are?</div><div><br></div><div>Thanks!</div>
<div>- James </div><br><div class="gmail_quote">On Wed, Feb 1, 2012 at 11:33 AM, Klaus Weide <span dir="ltr"><<a href="mailto:klaus@flash.uchicago.edu">klaus@flash.uchicago.edu</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<br>
The Flash Center is pleased to announce the beta release of the next<br>
version of the FLASH code, version 4 beta. FLASH 4 beta includes<br>
several new capabilities and resolves many bugs found in<br>
previous releases up to 4-alpha.<br>
<br>
<br>
The capabilities added since the alpha release are :<br>
<br>
Capabilities added since the FLASH4-alpha include:<br>
<br>
* 3T capabilities, support for non-Cartesian geometry, and support for<br>
stationary rigid bodies in the unsplit hydrodynamics solver<br>
<br>
* Upwind biased constrained transport (CT) scheme in the unsplit staggered mesh MHD solver<br>
<br>
* Full corner transport upwind (CTU) algorithm in the unsplit hydro/MHD solver<br>
<br>
* Multi-Temperature Multi-Material Multi-Table Eos<br>
<br>
* Ability to use cold opacities<br>
<br>
* Several solvers have been threaded with two different degrees of parallelism. The<br>
block level threading is coarser grained with one FLASH block being given to one<br>
thread, there is a finer grain threading which threads performance critical loops.<br>
Both modes use OpenMP.<br>
<br>
* Laser drive is reconfigured and threaded.<br>
<br>
* Several new HEDP simulation setups have been added.<br>
<br>
<br>
Improvements to capabilities since the alpha release include:<br>
<br>
* The HEDP capabilities have been more extensively tested, and are therefore more robust.<br>
<br>
* Eos related runtime parameters are clearer.<br>
<br>
* Several new shortcuts have been added to make it easier to use the<br>
new HEDP capabilities.<br>
<br>
* The use of Gravitational acceleration is handled more accurately in Hydrodynamics.<br>
<br>
* The interface with Hypre is more robust for diffusion.<br>
<br>
* More boundary conditions in PFFT, some Poisson solvers can now handle<br>
combinations of different boundary conditions (DIRICHLET, NEUMANN, PERIODIC)<br>
on different domian boundaries.<br>
<br>
<br>
Known Limitations:<br>
<br>
* The new multipole solver is missing the ability to treat<br>
a non-zero minimal radius for spherical geometries, and the<br>
ability to specify a point mass contribution to the potential.<br>
<br>
* Unsplit diffusion does not work in 3D.<br>
<br>
* HYPRE needs to be installed to use Unsplit Diffusion, Unsplit diffusion<br>
uses hypre-2.7.0b (might not be backward compatible).<br>
<br>
* Usage of HYPRE Split solver in FLASH4 is only for internal debugging purposes<br>
and is not generally recommended for usage.<br>
<br>
* Split diffusion has not been tested with PARAMESH.<br>
<br>
* Some configurations of hydrodynamic test problems with Chombo grid<br>
show worse than expected mass and total energy conservation. Please<br>
see the Chombo section in Hydro chapter of the FLASH user guide for<br>
details.<br>
<br>
* We have experienced the following abort when running<br>
IsentropicVortex problem with Chombo Grid:<br>
"MayDay: TreeIntVectSet.cpp:1995: Assertion `bxNumPts != 0' failed. !!!"<br>
We are currently in contact with the Chombo team to resolve this<br>
issue.<br>
<br>
* The absoft compiler produces erroneous results with optimization in<br>
gr_markRefineDerefine routine. The results are fine without<br>
optimization. There may be other files with similar behavior, but<br>
this is the only one we are aware of.<br>
<br>
* The PG compiler fails to compile source files which contain OpenMP<br>
parallel regions that reference threadprivate data. This happens in<br>
the threaded versions of the Multipole solver and the within block<br>
threaded version of split hydro. A workaround is to remove<br>
"default(none)" from the OpenMP parallel region.<br>
<br>
* The unsplit hydro solver has an issue with preserving symmetry in<br>
non-fixed block size mode in UG.<br>
<br>
* The ReineckeMeyer setup will not compile with the NAG compiler because<br>
of the presence of f77 like file used computing analytical solution<br>
<br>
* We have been unable to use Hyper with OpenMPI in the FLASH code, the<br>
runs result in execution failures.<br>
<br>
* The -index-reorder setup flag does not work in all the configurations.<br>
If you wish to use it please contact the FLASH team.<br>
<br>
* A longer list of changes is contained in the User's Guide in Section 1,<br>
What's New in This Release.<br>
<br>
The release is available at:<br>
<br>
<a href="http://flash.uchicago.edu/site/flashcode/" target="_blank">http://flash.uchicago.edu/site/flashcode/</a><br>
<br>
A stripped down version of FLASH3 that may be downloaded without a<br>
license is also available at the same site. This is essentially the<br>
FLASH framework without any implementations. Additionally a collection<br>
of mini-apps that represent the FLASH workload in many different<br>
configurations is also availble under the general FLASH license<br>
agreement.<br>
<br>
<br>
The Flash Center continues to provide support for "add-ons" to the<br>
code.<br>
<br>
Additionally, the FLASH testing software FlashTest, which became<br>
available with the FLASH3, continues to be available for<br>
download at:<br>
<br>
<a href="http://flash.uchicago.edu/site/flashcode/" target="_blank">http://flash.uchicago.edu/site/flashcode/</a><br>
<br>
The code architecture and most interfaces of FLASH4 are compatible<br>
with FLASH3. Many, but not all parts are also backward compatible<br>
with FLASH2. The Flash Center has written extensive documentation<br>
detailing how to make the transition from FLASH2 to FLASH3 as smooth<br>
as possible. The user should look to:<br>
<br>
<a href="http://flash.uchicago.edu/site/flashcode/user_support/" target="_blank">http://flash.uchicago.edu/site/flashcode/user_support/</a><br>
<br>
The website also contains other documentation including a User's Guide<br>
and a developer's section. A new feature from FLASH3 onward is the<br>
online description of the public interface routines to various code<br>
units. A new feature from the current release is a list of available<br>
setup shortcuts.<br>
<br>
<br>
Development of the FLASH Code was funded by the DOE-supported<br>
ASC/Alliance Center for Astrophysical Thermonuclear Flashes,<br>
and continues to be funded by DOE NNSA and NSF. We<br>
acknowledge support received from Lawrence Livermore National<br>
Laboratory and the University of Chicago.<br>
<br>
All publications resulting from the use of the FLASH Code must<br>
acknowledge the Flash Center for Computational Science.<br>
Addition of the following text to the paper acknowledgments<br>
will be sufficient.<br>
<br>
"The software used in this work was in part developed by the<br>
DOE-supported Flash Center for Computational Science<br>
at the University of Chicago."<br>
</blockquote></div><br><br clear="all"><div><br></div>-- <br>James Guillochon<br>Department of Astronomy & Astrophysics<br>University of California, Santa Cruz<br><a href="mailto:jfg@ucolick.org" target="_blank">jfg@ucolick.org</a><br>
</div>