 Okay, so we'll make a start now. Thank you for coming along to the FOSS in Research Lightning Talks, which is hopefully going to be the first of two sessions of Lightning Talks we'll be doing today. Our first presenter, start when you're ready. Yeah, thanks everyone for coming. My name's Jeremy Murphy. This is my first ever Lightning Talk, so please be gentle. I've done a bit of research in the biomedical sciences and being a free software fanatic, that's always tried to use it. This is my supervisor for my computational neuroscience honours year. And he wrote his own neural network modeling software called Parplex, which if you're familiar with neural network modeling, I guess the main package is called Neuron, which I think comes out of Duke or Yale University or something. But yeah, he decided to write his own, I think, because he wanted to have some particular parallel processing or something. And it was open source, but I think I was a third ever user of it. So you can imagine that it still had a few bugs. And so even though it was open source and free software, because he hadn't actually put it on the web, like on SourceForge or something, no one else has ever used it. So it was a bit frustrating. I spent a lot of my year finding bugs, asking him to fix them. Yeah, that was a bit annoying. I sort of, looking back, I wish I'd used Neuron. I would have had more confidence that it was, I don't know, that the main features would have been more reliable. I'm glad I put my phone on silent to just ring. And so for the plotting of the data that I got from that neural network modeling, I used Matplotlib, which is a Python modeling library. That was pretty good. I didn't really know what else to use. I'm not sure whether it's a MATLAB style one or whether that's something else. Maybe you know. So that was that project. And I also did some QSAR work, which in pharmacology is quantitative structure activity relationship. So that's about, you know, you've got a set of molecules, a set of drugs. You do some analysis on their structure. A lot of them have, and you haven't, and they're known biological activity. And then you've got a new molecule with an unknown biological activity, and you can make a statistical inference about what its activity will be. Which is quite fun. Initially I used this, I think it's free, where I had like a demo version or something of this chem sketch software. So just for drawing molecules and saving them into a file format. If I was doing the same thing again, yes. I used Avogadro, which is free open source software. For some of the molecular dynamics, we used Hyperchem, which, well, as an aside, is now available in Linux. That's definitely commercial software. And I think Avogadro actually has a lot of the features of Hyperchem now. Probably not all of them. And I didn't have time to look up all the different free molecular modeling, I mean, molecular dynamics software. So again, this is a QSAR process. I sort of had to get all the drug structures as files, and then run them through, we wrote our own software just to actually do the algorithm that is, you know, the protected intellectual property. But I realized afterwards that a lot of it was published publicly in the literature. And so I made a plug-in for Open Babel, which is a program for converting between molecular file formats. And I have the feeling that no one will ever use it, but it was just something to do. And then we used Weka for the data mining at the end. Of course, I use OpenOffice for writing my actual papers and so forth. And Citation, I use Bebus, which is okay. I don't know. There are other things like something called Zotero for Firefox. I didn't have time to really look at it. I'm really interested to know what other people use for Citation in free and open source software. Because Bebus, it kind of did the job. Where is it added open here? You know, it's kind of biomedically focused, because you can do a search through PubMed for what you're looking for. But yeah, it wasn't as great as, what's the big commercial citation package in note, yeah. And that's my story of using FOS in research. Thanks. Okay, so who's next? Oh, if you want. Yep. With the open source bibliography management, I've used Bibsononomy. And then there's another program called Jabriff. So you can just highlight the Bibtech on the web. Okay. And then it'll just go straight on their server. And then you can sync it with your local client. Jabriff has a module that you can pull all the references down and you can just tag them. So if you're writing a particular paper or there for your thesis, you can download a particular tag. Okay, so we'll switch to our next presenter. Feel free to grab a lollipop or something like that from the front if you want. They've given us a whole bunch of food for speakers to take. So all right, good afternoon, everyone. My name is Conrad Sanderson. I work for our company called Nikta and that's its logo there. It's an Australian Research Institute. And what we do is research in machine learning, artificial intelligence, computer vision, life sciences, et cetera. One of the labs is based at UQ. And what we're doing in that particular lab is working on surveillance, or automating surveillance to be more precise. During our research, we were doing a lot of stuff under MATLAB. And then we try to use a bit of Octave. The problem with those two approaches is that when you write stuff in MATLAB, if you want to get any speed out of it, you eventually have to convert it into some other language. And usually the choice is C++. The problem with this is that you need to redo a lot of the matrix handling. You need to redo a lot of the matrix decomposition handling of your own memory allocations, delegations, et cetera. We were actually quite frustrated with that. And we decided to build our own matrix library, but not just your usual run of the mill one. We use a lot of template techniques in order to combine a lot of operations into one. And the end result is what you see on the screen. This is a project on Sourceforge. The address is rmr.sourceforge.net. This is licensed under the LGPL, which means if you want to use it in our commercials or commercial or proprietary programs, you're more than welcome to. This has had about roughly 17, 18,000 downloads. It has an active community of people contributing bug reports and patches. If you guys are familiar with LA pack, or the linear algebra package, that's the standard, use or de facto standard. We provided interface to that as well. And this also allows you to drop in high performance replacements for LA pack. And some of these includes the Intel math library or the AMD math library. So that's it. If you're interested, let me know. I think there's a switch on that. What about now? It's more a question of whether or not the people with headphones on can hear you. Can the people with the headphones on hear me? You ready?