Sorry to all of you who looked in vain for the meeting notes last week. The meeting last week was rather small, and there was not too much going on— Matt and Eric just spent all day doing server cleanup. And well, to be perfectly honest, the meeting note taker had some Physics homework to attend to.

But here's what happened this week:

Bob reported that we had a weird thing happen in the database world. So we did have to turn off the replication for the week and we hope to restart it by tomorrow. Everything else seems to be going OK

Eric put in a lot of work on the RFI rejection this week. He is playing around with the changing the range frequencies that his algorithm looks at. He is trying to see if wider zones can detect more RFI without consuming too much of the band. But from the first couple of tests, it looks like narrower actually is working better.

Dan asked if this would help us create fixed RFI zones, like we used to do in the original SETI@Home?

Well, Eric said yes, it is possible. But only now we are much more sensitive to the data, so instead of having 5 denoted zones, we could have a maximum of 5,000 zones that are narrower.

When Eric ran the Zone RFI in a 20 Hz zone width it found fewer zones but knocked out less RFI. So now he is trying narrower ranges (maybe around 5 Hz). As of right now the running these tests takes a lot of time, but eventually when we find the right range we hope to run it over every single spike in the database. Also at this point he is only running it on 60 million spikes-we have something like a billion in the database.

Dave asked, how many of the current top 10 NitPCkr candidates are knocked out by the Zone RFI?

Well, Eric said that all of them were knocked out. They were all pretty obviously RFI.

Will the same thing work on Gaussians and Triplets?

Yeah, a lot of them will be defined in Zone RFI. Eventually we want to end up at a point were the top 10 of the NITPckr Candidates are not obviously RFI.

Will this process be automated? No, not really necessary, Eric will just run it periodically; it’s not too much of a problem in that respect.

Dave brought up the point that as soon as Zone RFI and Radar blanking are up and running, we need to start thinking about ways to make the interface for users who are interested in Identifying RFI.

A big discussion about this ensued —and it ultimately came down to three points we need to consider in terms of this project.

1. What do we want to present to the public?

It would be really cool to get different sorts of candidates (including spikes, Guassians, Pulses, etc.) up on the NITPCkr website. Currently all the signals are presented with spike waterfall plot and it takes around 20 minutes to pull out a tenth of a days worth of spikes. People aren’t going to wait that long for a webpage to load.

We will also need a lot of training materials (such as explanations of different types of RFI) in order to give people justification for their choices of interesting candidates.

2. What should the infrastructure be for this project?

We need to consider the following types of RFI: Zone RFI that has been persistent over months, Zone RFI that has been persistent over a couple of hours, and Drift RFI, which have yet to fully understand the source of.

3. What can we handle database-wise?

A good question.

Dave reported that the Einstein at Home guys who were at the NVIDIA conference were here last week. If we could find anybody who is interested in this kind graphics computing of stuff, we could optimize our cuda application. Maybe it could be a research project for an undergrad in Astro or EE and CS? There is a student in the Casper group who may be interested. There is also a guy at LBL that wants to get into the GPU world. Maybe he’d like to volunteer?

(At this point during the meeting, a mysterious box arrived in the mail.)

Matt’s software radar blanking is working really well, and we’re going to start using it any minute now. It is working even better than the hardware blanking. Good news!

Just back from South Africa, Andrew and Dan had a lot of exciting news to report.

Andrew will be meeting with a potential new student on Wednesday, a third year Astronomy student who has some programming knowledge and is interested in working with SETI. He’ll come back next week with more information on what types of projects she is interested in and can work on.

We got funding to host locally 24 terabytes of Fly’s Eye data. Andrew has been working on what that system would look like, so he’ll forward his ideas to Matt and Eric to discuss them later.

Dan reported that the CASPER Workshop he held in South Africa was really well attended: people from 70 countries were present. It was probably the most successful workshop we’ve ever had. Also, Laura gave a nice talk in South Africa on the roach application she is working on for SERENDIP 5.

The South African government is building a huge telescope out there called meercat, with 50-70 disks that will be 12-15 m in diameter. It will be a really nice telescope to do SETI stuff with, and they are very interested in collaborating and will likely send people over here to work with us. They have a really good site for building the telescope —the Karoo desert. It is huge, empty, has no water, no people, and no RFI. Perfect.

KAROO DESERT

That is all for this week from the Space Sciences Lab, and Thanks for Reading!