ScienceI find the creation of black holes to be an extremely interesting process. SN 2006gy, whether or not it formed a black hole, is an awesome source (literally inspiring awe), and I like both the idea that it was a single supermassive star (a loose term in context) like Eta Carinae and a more-than-massive star formed from collisions to be ~100's of solar masses (e.g. Portegies Zwart et al. 2007). I could easily become very interested in stellar collisions, if convinced that they are as common as claimed. Update: SN 2006gy has been a hot topic lately, and the most common scenario is no longer the supermassive collisionally-formed star mentioned in that paper, but the paper is still quite interesting and the new ideas are even more interesting.OutreachIn general, I like sharing pretty pictures with the community, which is the most basic level of outreach. Some of my images can be seen at L111 , W345 , and the NRAO Photo Contest winner and related at GCHowever, that's not enough. Ideally, an outreach project should allow the community to interact, understand how astronomers work, and develop a deeper interest in the topic. One interesting idea would be the creation of a flash game that could provide useful scientific results. I'll probably never make that, but it's a fun idea. SoftwareThrough my work on the SAGE survey, I have also become interested in developing data processing software. I wrote my own scripts using the Perl Data Language to display combinations of spectral energy distributions and cutout images, sort through databases searching for a combination of parameters, and interpolate individual velocity cut images at a number of position angles into a complete data cube. I use Python instead of PDL now because it's better supported and (somewhat) easier to install.My codes are not particularly fast, but they weren't too difficult to write and allow me to analyze large data sets with little difficulty. Data analysis and reduction tools are presently somewhat antiquated. Command line tools like IRAF and AIPS are tremendously powerful, but have rather obtuse interfaces. Python's wrapper for IRAF, pyraf, has partially remedied the problem for IRAF, and CASA will presumably do the same for radio reduction software (although in beta, its graphics are slow). However, past the reduction stage, analysis is not always intuitive. Graphical analysis tools are useful and getting better, DS9 is a wonderful tool that is underutilized by the majority of astronomers who know little of its capabilities past simple image display. The folks at CFA who support it are also very helpful and responsive. GAIA, part of the UK's Starlink package, is a powerful data-cube analysis tool that allows instantaneous extraction of the third axis at any point on the image. The UK group has excellent support even for their unfunded, unsupported software - I think they're a wonderful example of what astronomical computing SHOULD be but isn't. Most astronomers aren't very good with computers, so a well-funded and competent support group can greatly enhance their efficiency. The problems in the near future are going to be an excess of data. With ALMA's coverage of far too many spectral lines, there is no simple solution: they need to be identified somehow. However, for data sets that will come out of Pan-STARRs and (if funded) the LSST, hunting through the archives will be inefficient or impossible. Scripting in perl, python, or whatever language is available may be the most efficent way to get through the data once the raw images have already been processed through some sort of source extraction algorithm. Thankfully, the excess of data will probably make scientific verification much simpler, and there will be no image reduction required for the extremely large-scale surveys (I don't know if I really believe that). Because the work of a researcher can be reduced to analysis only, science goals will be achieved more efficiently. Computer simulation and modeling is also becoming more efficient, and it seems that modeling groups have had problems of excess data for over a decade. I hope, though, that the smaller scale models (e.g. DUSTY, STARBURST99, 2DUST, and their ilk) can be wrapped and made more useable by the general community. As they stand, any of these codes has a very steep learning curve and requries a very specialized understanding to use effectively. Maybe I'm interested in astronomical computing. |