Entrevue avec Andreas Hornig, administrateur de Constellation - English Version

Could you please introduce yourself ?

My name is Andreas Hornig and I'm the head of the Constellation Platform. I'm an aerospace engineer with a Diplom (FH) from University of Applied Sciences Bremen and doing another Diplom with specialisation in Space Systems and Space Applications (Raumfahrtsysteme und Raumfahrtanwendungen) at the University of Stuttgart.



Could you introduce the team that is in charge of the project ?

The Constellation team consists of an interdisciplinary mixture from aerospace related professions and topics, from server admins to earth and space enthusiasts. A complete list can be found here http://aerospaceresearch.net/constellation/contactimpressum.php and feel free to contact each of us according our expertises. Withouth the team, this project wouldn't be as progressed, as it is today.

We are a rather small team, but we are also a young academics group at the university and are always looking for new team-members, who want to be part of our team and let our team grow.

And we feel, that you, the crunching volunteers, are part of our team, too. You can take part in decisions, like we asked you what duration our workunits should have to please you, or with what frequency you set your check pointing to implement it in our application.
And of course participation is so important for us, that you can also create your own sub-project that can run on the platform.



Where is the project localized ?

The Project was started in Stuttgart and our server is hosted on the campus at Selfnet.de. Some of us are also living in Stuttgart Area but the team is distributed all over Germany. A distributed Computing project can be done with distributed team-members.



When and how was the idea of the project born ?

The project originates from my first Diploma thesis „Computer-aided method for optimizing a 2D ascend trajectory of a multistage launch-vehicle in consideration of mass-ratio between propellant and payload“ where I already describe in a short paragraph the usage of BOINC for such a purpose. For my computation runs, I hijacked several university's pool-PCs and let them optimise the tasks during night-times, so no computer was blocked and students could use them. I also woke up early in the morning to be the first at those PCs to save the results. A thought manifested, that it would be great to have a BOINC project dedicated to aerospace issues. I had a discussion with Sayandeep Khan, who is a friend of mine and was an Earth and Space Science student at Jacobs University Bremen, and he also liked the idea and wanted to use BOINC for some of his projects. The begin of Constellation as a platform.
It took some time to ask on Rechenkraft.net for help with hat an be described as first Step http://www.rechenkraft.net/phpBB/viewtopic.php?f=12&t=9866&p=107156 and a short visit to BOINC Community Meeting 2010 at Wieda (Lower Saxony, Germany) http://www.boinctreffen.de.
But the real start was at the regulars table at the Stuttgart chapter of the German Society for Aeronautics and Astronautics (DGLR) (German: Deutsche Gesellschaft für Luft- und Raumfahrt) where I proposed my idea and where we became a young academic's group.
So we already had walked some distance on a long road to the stars, but haven't reached the end, luckily.



In a few words, how would you define it ?

Distributed Computing for Mankind!
We intend Constellation to be a platform for research in various aerospace related sciences and engineering.
This catchy slogan describes, that everyone can be part, be it as a volunteer donating idle PC-time and by providing a sub-project hosted on the platform. That's why we call our platform Constellation, because like a astronomical Constellation, our overall system shall be formed by several individual stars (applications).



What is the percentage of progress of the project, how many years expected ?

That's hard to set, but we are right at the beginning. Our first sub-project "TrackJack" is used for a thrustcurve optimization to find the best curve that results in the highest altitude of hybrid rocket designed by young academics group HyEnD http://www.hybrid-engine-development.de. That first batch is progressed to 14% from Constellation's public start in April until when this interview will be published. So there will be workunits for quite some time, because this this is only one batch of parameters from one HyEnD Design. (http://aerospaceresearch.net/constellation/server_status.php graph "TrackJack Progess 50 days")



Do you have any other ideas of projects ?

There are two projects under development by Sayandeep Khan.
"Extreme Machine" simulates exploration rover dynamics and "On The Moon" aims to simulate the process occurring at moon's near the surface.

We already collected possible ideas during our beta-test event "invite codes for ideas" where we received a lot of nice candidates for sub-projects. [to the editot: Perhaps you could set a link into your own forum]
Perhaps and hopefully a member of L'Alliance Francophone will contact us to propose his/her project to us. We definitively have open space for new applications.

We have some ideas for changes in BOINC itself and further progressions in distributed computing systems, and an idea about a hardware sensor array.
Because we don't have the manpower to realise those ideas by our own. We currently discuss to set up a wiki to post all our ideas for project and ask for help from the community. It would be too bad to keep all those ideas for our own. And that will be a place for new ideas send in by users.



Does the application have any issue ?

The biggest issue and the biggest advantage is JAVA. "TrackJack" is coded in JAVA and on the one hand it tends to be slower as a native compiled programming language. On the other hand it is way easier to port it to different operating systems. As long as there is a JRE for a specific operating system, we can use the same "TrackJack" JAR-File and don't have to change any line of code, we just have to replace the JRE.
That makes available even more volunteers PCs that more than compensates the slower run-time. "TrackJack" could even run on smart phones with JAVA support, a still unused resource.



Do you plan to create a badge system, like does Primegrid, in order to stimulate computation and competition?

Badges are a great way to award the efforts of volunteers and the team disuses the integration of badges and the way we'll do that. When real astronauts show their affiliation with badges on their uniforms, we'll will also do it. Any wishes?



What do you think about the distributed computing and it's development ?

We are so grateful, that distributed computing and volunteer communities of citizen scientists exist. Without the technical system and the people donating computing capacity, we and all projects couldn't succeed in our tasks. This thank you is for all of you.

The distributed computing system BOINC is a powerful tool and the wrapper technology is an easy way to integrate applications like TrackJack.
Unfortunately it isn't suitable for all aerospace purposes. Especially for computational fluid dynamics (CFD), a big branch of daily aerospace work, real parallelity is needed, where a set of cluster-nodes work together on one workunit. BOINC-Clients are not able to work together as nodes on one workunit. To teach an old dog a new trick is one of our ideas we are working on.



In some days, our team will make a run on your project : What do you think about it ?

I'm learning French but it's still very basic (A2 level) but perhaps (Do your game!) expresses it best. Have fun with your raid, we support such (space) races and we hope, that our server will survive your rush. Otherwise we know where to find you and to send the bill. ;)