NPAC's activity in the Binary Black Hole 
Grand Challenge II

Geoffrey Fox, Tom Haupt, Scott Klasky



We propose to develop a domain-specific environment which 
leverages our current work on the alliance ADM code and other 
applications. This work will be centred around DAGH, and in 
order to develop such an application we will be collaborating 
with the University of Texas at Austin (Browne, Choptuik) and 
Rutgers (Parashar). This work will be done by Haupt, Klasky, 
and 2 graduate students.

The majority of this effort will be to provide the alliance DAGH 
codes with a computational steering front end. This front end 
will allow researchers and end-users easy access to the DAGH 
AMR code, and will allow for more interactive control of the 
code. This will then provide a vital link between the numerical 
aspects of AMR, the underlying physics, and visualization.

Visualization of 3D AMR codes is conceptually difficult, and is 
an active area of research. We feel that in order to develop such 
tools for the alliance effort, we must become experts in the 
underlying physics as well as the computer science. Thus, we 
feel it is necessary that we must do some computational physics 
in order to properly develop the application. 

Since this is a large collaboration between geographically 
diverse universities we also feel that is necessary to develop 
collaborative visualization tools which can be used during 
phone conferences and other collaborative interactions. These 
tools have already been in development from other NPAC 
activities, and we feel that further development is necessary in 
order to investigate 3D AMR codes from such an application. A 
major advantage in using our model is that it is developed in 
Java and is thus available on all platforms that support Java. We 
have found this to be a powerful model because it also allows 
independent researchers to develop their own filters which are 
easily integrated into our system. 

The development of a 3D AMR code necessitates the need for a 
"computational steering" front-end. One of the features of this 
will allow developers access to all of the data on-demand. This 
will allow users to see all or part of the 3D AMR grid functions. 
Another feature is that it will allow constructed quantities to be 
computed on-the-fly, while the simulation is proceeding. This 
takes away excess computation from the supercomputing and 
places it on another system. Part of the difficulty in developing 
such an application is working with the hierarchy on parallel 
processes. Since we are developing this with DAGH it will 
become part of the "DAGH-system". 

During the course of this project, and integral to our efforts, will 
be the development of "general" computational tools, that will 
be able to be utilized widely in the black hole community. In 
designing our system, we will keep it flexible enough so that we 
will be able to steer and visualize results from our financial 
modelling codes which uses a Monte-Carlo, path-integration 
kernel. We feel that it is necessary to use diverse applications in 
order to develop a flexible front-end to large computational 
codes.

Another major part of this effort will be the further development 
of Haupt's DAGH interpreter, which has been shown to turn 
non-DAGH black hole codes into parallel AMR DAGH codes. 
We will be working to make this interpreter work more general 
and feel this is a necessary and central component to NPAC's 
overall contribution. Vital components that will be added will 
serve to automatically insert Cray directives and calls to our 
visualization system which will allow the DAGH code to be 
steered. 

This project will encompass 50% of Klasky's time, 25% of 
Haupt's time, and 2 graduate students.