> As you can see Grant is here so I'll begin with a brief introduction and then ask a few questions. Then we can open it up to the floor.Hello to you all. > I'm pleased to introduce our first Janet A. Mattei Research Fellow: > Grant Foster. Grant is well-known to most of our membership, > having been a longtime staff member at HQ from 1989-1997. > He is the author of many of our core data analysis programs at > HQ. > Currently, Grant has an active music career, but earns his "real" money > working on text categorization for Island Data Corporation, > teaching computers how to classify natural language text. > While at the AAVSO during the summer he will be working > on many projects involving cepheids, dwarf novae, > semiregulars and more.I'm especially looking forward to "more" > Grant, during your time at the AAVSO from 1989-1997, what major projects did you work on? > there will be "lots more" :)I worked on several "theoretical" projects and several "astronomical" ones > Some of the theoretical projects resulted in software used by staff such as TS and WWZ. These two packages are now available to the public via our we b site. Can you give a brief description of both?Theoretically, mainly the analysis of variable star time series. That includes refining Fourier analysis, and creating the first wavelet method that works well with irregularly spaced data. Astronomically, there were lots of things... Probably my two favorites were the Hipparcos project (they used our data to calibrate their photometry, and... a study with Janet (and others) on the multiperiodic character of semiregulars. We established (I think for the first time) that *most* semiregulars are multiperiodic. But -- they generally show only one period at a time. Often the periods will "switch" (mode switching) Sure... > What is the CLEANEST method?My favorite is "TS" -- it's a time series analysis program. It does very sophisticated Fourier analysis (the CLEANEST method), and also does polynomial fitting, averages, visual exploration... It's a cool program! But it runs under DOS, so unless you're on a Windows PC ... > The polynomial fitting, averaging and light curve tools are available in our Windows XP tool MagPlot. *** ReplaceThisWithAName51 (ReplaceThi@rox-38E6211E.aavso.org) has joined channel #aavsoThe CLEANEST method is a multi-periodic Fourier analysis. It enables you to scan for multiple periods, at each step refining each period to give the best fit. It's also optimized for unevenly spaced data (the bane of astronomy) The WWZ program is the "weighted wavelet Z-transform" *** Signoff: ReplaceThisWithAName51 (Client exited)It's a wavelet method which compensates for uneven time spacing. It's quite adept at detecting changes in period and amplitude of a pseudo-periodic system. > Can you give an example of uneven time spacing in variable star data?Sure. Many variables (maybe most) show an annual gap in the data stream -- when the sun comes close enough in the sky to obstruct viewing > Two more questions.. > one general.. > Q: Why is math so important to astronomy?Then we see an annual "cycle" of data density -- a gap every year. Sometimes, it's just hard to get time at an observatory, so we'll see one data set of a few days, followed by a week or two gap, followed by an other set of data. Gaps in the data are a real problem for analysis methods. For a lot of reason!!! *** Vyvyan (Vyvyan@rox-38E6211E.aavso.org) has joined channel #aavsoFirst, a lot of astronomy is physics -- astrophysics -- and physics means math Second, we take a lot of numerical data -- that means analysis -- and that means math again. so you can deduce the first to "for reason " In fact, a lot of astronomers are natural-born mathematicians But then, a lot of 'em aren't. Q: why observe , what do we learn ? > Okay last question, What kind of music do you compose and perform? Two questions! First, we observe because we learn about the nature of stars. Especially variable stars tell us about how stars change through out their lifetimes. > How did you get into Irish music?Second question: I compose classical music, I perform Irish traditional music. And Boston is a great town for Irish music! (for classical music too) I met my partner, and we decided to get together to do renaissance lute songs. Then we started doing Irish ballads, and decided we liked that better. > Okay thanks for answering those opening questions. Now let's take questions from the rest of the chat room and begin an open discussion!Then I started going to "sessions" (musicians gathering to play Irish music), and fell in love with the tunes. Fire away! Since you were last at HQ CCD's have come to play a much more proment role.... any comments on the effect of your data analysis programs Two spring to mind: 1. the data are of much greater precision, that enables us to look for more structure (e.g., cataclysmics in outburst) *** Signoff: mike (Client exited) > Any tips for observers to avoid aliases in their data from observing patterns?2. CCD data are often plagued by *words* gaps than visual data. CCD observers tend to go from target to target, while visual observers tend to make a long-term list and follow it regularly. BTW, we're hoping to get some good CCD data for some Cepheid variables ... especially TU Cas But *ONLY* visual data have the very long time baseline which is crucial for certain analyses. A couple of things. 1. Try not to observe the same time every night ... that leads to bad aliasing with frequencies /- 1 cycle/day. D'oh! 2. Observe steadily. Don't observe a target for 2 months, take six months off, then return for another two months. Don't beat yourself up if you observe the same time every night. For some variables (Miras, other LPVs) it makes little difference. For short-period stuff (Cepheids, etc.), just randomize your observing times. Are we talking a few minutes, a few hours? We're talking a decent fraction of the period -- so mix 'em up by hours if you can. > If an observer wants to get started in data analysis, where are some places to begin? What would be the first thing you'd advise? *** mike (mike@rox-38E6211E.aavso.org) has joined channel #aavsoI realize some targets have to be observed when you *can* -- go for it. But if you have a choice, randomize your times. > Any good stars to start with?To get started with data analysis, get some of the programs from AAVSO and start playing with data! Expect it to take a while to get used to things. A year from now, you'll wonder how you could be so naive at the start. It happens to everybody -- even the pros! > What AAVSO data are you working on right now?I'd start with LPVs, Miras specifically. > (FYI: Get programs here: http://www.aavso.org/data/software/ and data here: http://www.aavso.org/data/download/ ) They have large amplitude and reasonably regular pulsation light curves. There's a lot of structure there. And a lot of ways to go wrong! But don't be afraid of going wrong, that's how you learn. Right now, a couple of small things (RU Vul, BZ UMa), and one large project: Cepheid variables. > What are you doing with the Cepheids?Cepheids were, for a long time, thought to be unapproachable visually because of their low amplitude. But AAVSO observers did 'em anyway! > Sounds like us...And it's a good thing, because visual data is plenty good enough. We're looking for two things: 1. changes in period and amplitude over time (this is where visual data are most important, because no other sour ce gives enough time coverage) > Any results yet?and 2. studying the shape of the light curves, to look for clues to Cepheid structure. I've found a few with changing periods and amplitudes. I think Cepheid periods are more changeable than most people think. > Excellent! > Can't wait to see the papers :) > Let's say you have a data set and want to look for any periodicity in it. What is a good rule of thumb for what range to look for? That is, how many cycles needs to be in the data set for Fourier analysis to be > reliable?I've also devised a new way to "view" the light curve shape, which I'm hoping will clear up some outstanding problems. I'd say you need a minimum of about 3 cycles. > Often we have CCD time series runs that span just a few orbital cycles here and there, so it has been difficult for me to judge whether to believe th e fourier results or not. > That's very helpfull.If there's only one cycle -- how can you know it's periodic?!!! With only two cycles, it's still hard to be sure. With 4 or more cycles, you're on safe ground. Indeed! Of course, with AAVSO data you often get hundreds of cycles (some LPVs for example) Grant, can you explain in more detail the '"view" the light curve shape' comment? If you're just starting out with data analysis, pick a star with many cycles. *** Jim_BJS (Jim_BJS@rox-38E6211E.aavso.org) has joined channel #aavso Also, pick one with NO periodicity -- just to get a feel for what the results are like when periodicity is absent > great idea!Another piece of advice for data analysis: one of the most powerful methods, and one of the most underrated, is VISUAL INSPECTION > Use the force, Luke...In other words, LOOK at the light curve. Look at different timescales. Use your intuition. You don't need to analyze this data... *** TravisS (TravisS@rox-38E6211E.aavso.org) has joined channel #aavso > These aren't the periods we're looking for.These aren't the periods you're looking for... You can go about your business... lol > These periods can go about their business. I don't fall for Jedi mind tricks > Boy it's hot in these suits. Back to the wavelet method, AAVSO developed a very nice Windows version, "WinWWZ" > Thanks mostly to Geir Klingenberg, who volunteered his time to the project.It has a nice interface, draws pretty pictures, and is *very* easy to use. BTW, you *don't* have to be a mathematician to do data analysis -- let the computer do the math for you. > That's my kind of talk!Any CCD observers out there? Yes yup Yes... a TU Cas, too What are your favorite targets? I've been watching a lot of CVs. I'm new, so I have been doing campaign stars, mainly. Campaigns and Requests USing that as my guide I tend to stick with the same ones. > We need more SS Cyg!! Q: what should an amature know about math , what is the min requirement to just contribute data to more advanced analyzing part lol i went to all the trouble writing it Question, Grant: What would be a good tool to use on data like ASAS, with seasonal gaps and perhaps not followed all the way to minimum? A s eries of parts of a light curve, in other words? Talk to the weather Gods, Aaron! I thought SS Cyg was well covered? Yes Aaron..... we need more ss cyg (I can now go about my business) > I'm looking for nightly CCD time series until August. You *really* don't need much math to do serious data analysis. > heheMost of the math work is done, when the program is written. but if you wished where should you look > And when people like Grant invent new and amazing methods for us (CLEANEST!) There's a lot of material online. Try "mathworld" (a google search will turn it up) > soul, try here: http://www.aavso.org/aavso/meetings/spring03present/templeton.shtmlAlso, find a decent book on Fourier analysis. That's a good start for period analysis. on the Cepheids... any periodicity to the amplitude variations? I haven't found any yet. There are hints of this in the literature, but nothing confirmed yet. What are your favourite memories from your years working at HQ? no blazhko pretenders then ;) ? It's tough using visual data for amplitude, because each observer's eyes respond differently. No, no Blazhko pretenders (yet!) My favorite memories? Lots! Pick a couple... One of the best, was the spring meeting in Switzerland. I used Wayne Lowder's data on Cepheids to *prove* that visual data were plenty good enough. He was *thrilled*. He also resumed his campaign on TU Cas. a lot of people forget that the original cepheid stuff was often done on gappy photographic plates, which ain't that accuate photometrically at times I also have enjoyed meeting, and working with, some prominent astronomers who were friends of Janet's (who wasn't?). Notably, Lee Ann Willson and Martha Hazen that's why I'm always a bit cynical when period changes are found over fifty year gaps in data, sort of thing Another thing -- a typical Cepheid analysis in print is based on < 100 data points. From AAVSO, I can get > 25,000 for delta Cep! Establishing a *genuine* period change is difficult -- I'm always skeptical myself, until I see the numbers. And you're quite right, a lot can happen in fifty years -- especially if nobody's looking! ;) BTW, if any of you CCD observers want to monitor TU Cas -- I won't stop you! It's a "beat Cepheid," meaning it pulsates with two different periods simultaneously. That makes the light curve quite complicated. I think Arne is planning to organize a TU Cas campaign -- so I hope I haven't let the cat out of the bag. I'm on it, Grant...... at Arne's request. But I'm a bit spotty, so help from others would be appreciated > We announced the campaign in Alert Notice 318 > http://www.aavso.org/publications/alerts/alert318.shtml It's always best to have multiple observers. For one thing, they're a useful check on each other. For another thing, it greatly improves the coverage > We'll follow up with details in the next CCD Views. But it seems like we'll be looking for precision CCD photometry until the end of the season, arou nd next March or so.When it's raining in Boston, it's often sunny in San Diego. TU Cas is a project for our Sourthern Friend right now... Friends It's also good to have decent observer scatter in longitude. The folks in Australia can fill in the gaps from those in Iowa. not for tu cas they can't :^)) I was going to say that > Well it has been almost an hour. Any outstanding questions for Grant? Oops! The folks in Korea... yep, remember to analyse aavso W Tau data it's unique > Okay I'd like to thank Grant for stopping by. > nickname, it is on the list! :) You're welcome. It's been a pleasure. good May the Force Be with You *** Signoff: StarBoyCTX (Quit: CGI:IRC 0.5.5 (2005/01/08)) > I hope we can talk about it in the future here.. And with you all! Thanks, Grant! I still think I'm safe smoking on the HEA porch with a statistician :) Bye! > Grant will be at the AAVSO through the end of July. If you have any follow up questions you can send it to aavso@aavso.org and we'll make sure he get s it. *** Signoff: Grant (Quit: CGI:IRC 0.5.5 (2005/01/08))