source: inundation/wiki/ANU meeting 27072005.txt @ 1951

AnuGA planning meeting 27 July 2005
-----------------------------------


Present: 
From ANU: Stephen, Matt, Linda
From GA:  Ole, Andrew


1: Parallelisation of pyvolution

The current contract is essentially completed. 
However, a few remaining loose ends will be given some further 
attention over the next months. They are

- fluxes in ghostcells should not affect timestep computation.
- a function for re-assembling model output should be made available
    
    
Ideas for a second contract include:

- scoping of methodologies for automatic domain decomposition
- implementation of automatic domain decomposition 
  (using C extensions for maximal sequential performance in order to minimise 
  performance penalties due to Amdahl's law)
- in depth testing and tuning of parallel performane. This may require 
  adding wrappers for non-blocking MPI communication to pypar if needed.   
- ability to read in precomputed sub-domains. Perhaps using caching.py  

We will iterate on these ideas and arrive at a mutually beneficial
contract to be carried out in early 2006



2: Validation of pyvolution

Laboratory Data:
Benchmark data from wave tank experiments were deemed suitable for
initial validation exercises and we should proceed with these
immediately using the existing acquired datasets.  In addition, data
from vertical wall run-ups will be relevant.  Andrew's letter to key
players regarding other existing laboratory datasets is deemed
extremely valuable and should go ahead.


Historical Data:
Andrew has contacted relevant organisations regarding sources and
availability of existing post-tsunami field survey data.  We agreed
that a scenario from e.g. Banda Aceh or Thailand (Phuket)
demonstrating the detailed effect of bathymetric variations would be
an ideal study and could form the basis of a scientific paper. Matt is
considering whether this would fit into his research priorities.
  

Other:
Validation against analytical solutions is well under way and 
is considered an integral component of the validation process.
The same is true for the Lake Merimbula tidal validation project.
Stephen and Chris are looking after these.


Contacts:

Ole will take contact with Lex Nielsen from SMEC in order to explore
how they can get involved

We will also establish a closer relationship with Lutz fram ACcESS

Andrew to contact key researchers/organisation to enquire about and
potentially acquire further laboratory and field survey datasets.




3: Other collaborations

Ole to investigate if GA would fund a PhD student: Project could be
"Algorithms for augmentation of unstructured grids" or "Algorithms for
unification of geodesic systems in unstructured grids" the latter
capturing the notion of being able to model the synoptic, deep water
wave propagation in latitudes and longitudes with a seemless connection 
to detailed modelling using UTM projections.
A PhD scholarship is worth about $20K per annum. 



4 Long term aim

We agreed that having a detailed, reproducible and validated 
inundation scenario from the Boxing Day Tsunami running in parallel
(and being faster than the sequential code) would be a worthwhile goal.