Jump to content


Photo

Research project - Limit for uneven Double Stars

  • Please log in to reply
34 replies to this topic

#26 nrivard

nrivard

    Lift Off

  • -----
  • Posts: 1
  • Joined: 25 Sep 2012

Posted 26 September 2012 - 10:15 PM

For those who are interested to participate in this project, I built a web site that gives all the information you need. Please note that the list of pairs published in S&T has some errors and is incomplete. Sissy has sent many updates since then. I also prepared some sky maps for those who dont have GOTO scope like me.

http://unevendoubles...spaces.com/home

I also hope you become friends of the facebook page, so you can stay up to date with the latest changes on the wiki site.

http://www.facebook....226255637503049

Enjoy!

#27 Rich (RLTYS)

Rich (RLTYS)

    Fly Me to the Moon

  • *****
  • Moderators
  • Posts: 5249
  • Joined: 18 Dec 2004
  • Loc: New York (Long Island)

Posted 27 September 2012 - 05:13 AM

Thanks, excellent site.

Rich (RLTYS)

#28 drollere

drollere

    Surveyor 1

  • *****
  • Posts: 1584
  • Joined: 02 Feb 2010
  • Loc: sebastopol, california

Posted 27 September 2012 - 09:24 AM

The pedants always say the same thing: there is no way you can do that accurately, particularly for things near the limits. And they are right! But here's the thing: just being in the ballpark is enormously useful in real world applications. ...

well, as one of those "pedants" i'm happy to concur with your agreement that "we" (i) are correct in our skepticism. however, you seem to concede a point while denying it, as prediction that doesn't work is still "enormously useful", which leaves me unclear as to what you want to say.

How do most people pick easy pairs? By using pre-compiled lists of stars that others have observed. But this is usually just the tip of the iceberg of objects available. So everyone ends up observing the same subset of the available pairs again and again. ...

i wouldn't agree that everyone ends up in a rut ... i believe wilfried is embarked on a campaign of rummaging through obscure stars, and there was another discussion here of jonckheere doubles, a catalog that rarely make it into the "top 500" lists. but this is really a digression from your point, which is a defense of "predictive" algorithms as a statistical vade mecum that can lead observers away from the hackneyed into the wide unknown.

really? if you only date those few persons that you "predict" will give you a good night kiss, then you've taken a hell of a lot of excitement and adventure out of dating; and similarly if you only look for double stars that you "predict" in advance will be "kissing", then you've merely fallen into a more elaborate and ritualistic form of routine, and you have not really addressed what seems to be your complaint, which is that the observer should challenge himself with something genuinely novel or unknown. you've just replaced a rut of consensus with a rut of "prediction", which seems to mean a consensus as to how doubles look without actually looking at them. at least, looking was how those lists you disparage were compiled.

the crux is that you seem embarked on the project of telling me what i will see before i've actually looked at it, rather than telling me why something may be difficult to perceive when i try to look for it, which is a very different type of guidance.

The algorithms tend to break down at the edges of perception; in other words they break down for the pairs that are difficult to split. This is a happy coincidence. After all, if we could accurately predict whether or not a challenging pair can be split under every conceivable circumstance, then there would be little point in going out to the telescope to try. For the challenging pairs the very point of observing them is to try to push your limits, not knowing if you are going to be successful. If our predictive algorithm says a pair is going to be very difficult to split, sure, it may not be perfectly accurate. In truth the pair may be easier than predicted, or it may not be splittable at all. But if used correctly, the algorithm can still predict which pairs are challenging, regardless of whether or not the detailed prediction is accurate.

ok, the predictive algorithms break down exactly where they are needed most. you must see here that the problem is not with the algorithms but with your use of the term "predictive", which merely begs the question and muddles the issue. any categorization runs up against problematic cases, where one must shrug and toss the case into one category or another. the case isn't difficult to predict, it's difficult to categorize or sort. and categorization, as we all know, is just a shorthand form of description. creating a new category that you call "challenging" doesn't change the basic procedure.

what seems beguiling to the layperson is the ritualistic method by which the "predictive model" is developed. you're only creating a database sorting rule. yes, it's a "consensus" database sorting rule, because you have a lot of individual observers "voting" as to which stars shall be classified this or that way by it. but one could (as i have) create a purely theoretical or judgmental sorting algorithm to combine aperture, magnitude, etc. into a double star "difficulty score", and it amounts to exactly the same thing -- "challenging" pairs can be identified, which are usually either "unsplittable" or "easy". as long as everything stays in a database, nothing is "predicted".

if the goal is to identify, as you say, stars that will be easy or difficult given the instrument and conditions, then i'll just assert that most experienced observers already do that with pretty rudimentary, rough guess rules based on magnitude, magnitude limit and separation, given personal experience with their instrument and observing conditions. since a "challenging" pair is likely to turn out "unsplittable" or "easy" anyway, nothing more elaborate seems useful.

the fundamental question is simple: what makes a double star "difficult"? the magnitude difference by itself is meaningless, because on the one hand the primary can be so bright that it produces substantial glare or scatter, and on the other hand the primary can be so faint that the companion is pushed below the visual limit. simply articulating that explanation of why a pair *may be* difficult seems to me far more valuable than using those factors to "predict" that the pair *will be* difficult. the explanation nurtures an observer's skill, while the ritual of pseudoprediction just paves a new landscape of boredom.

it's not a "happy coincidence" that your descriptive models break down when put to the task of prediction rather than sorting data: it's an outcome that any psychometrician can predict will occur, with a high probability of being right.

#29 blb

blb

    Aurora

  • -----
  • Posts: 4510
  • Joined: 25 Nov 2005
  • Loc: Piedmont NC

Posted 27 September 2012 - 02:51 PM

The pedants always say the same thing: there is no way you can do that accurately, particularly for things near the limits. And they are right! But here's the thing: just being in the ballpark is enormously useful in real world applications. ...

well, as one of those "pedants" i'm happy to concur with your agreement that "we" (i) are correct in our skepticism. however, you seem to concede a point while denying it, as prediction that doesn't work is still "enormously useful", which leaves me unclear as to what you want to say.


Well I am one of those that thinks that if enough people will try to make the observations on enough nights, that there will probably be enough observations with good conditions, that the formula will be a good predictor and with good optics and looking on enough nights that you will split it to, if the formula says it is possible. In other words we need as many observers as possible and fewer nay sayers

#30 blb

blb

    Aurora

  • -----
  • Posts: 4510
  • Joined: 25 Nov 2005
  • Loc: Piedmont NC

Posted 27 September 2012 - 03:06 PM

really? if you only date those few persons that you "predict" will give you a good night kiss, then you've taken a hell of a lot of excitement and adventure out of dating; and similarly if you only look for double stars that you "predict" in advance will be "kissing", then you've merely fallen into a more elaborate and ritualistic form of routine, and you have not really addressed what seems to be your complaint, which is that the observer should challenge himself with something genuinely novel or unknown. you've just replaced a rut of consensus with a rut of "prediction", which seems to mean a consensus as to how doubles look without actually looking at them. at least, looking was how those lists you disparage were compiled.

the crux is that you seem embarked on the project of telling me what i will see before i've actually looked at it, rather than telling me why something may be difficult to perceive when i try to look for it, which is a very different type of guidance.


Really, with only the difference in magnitude at a limit of only 4, I think that there are still many stars to challenge us. When planning a nights observing I like the idea of knowing, if the sky cooroporates, I have a chance of seeing an object. If the formula works properly, we will be hunting those nights with good observing conditions and not hunting for something we do not have a chance to see. I see this as something that will tell me if this is an object that if I have good optics, observing conditions, etc., it will or will not be worth spending my time on. We all do this mentaly anyway based on our experences observing.

#31 drollere

drollere

    Surveyor 1

  • *****
  • Posts: 1584
  • Joined: 02 Feb 2010
  • Loc: sebastopol, california

Posted 29 September 2012 - 11:19 AM

well, i'm trying to be constructive and not rain on the parade. there's a lot to be learned for all involved by the exercise of looking for a better rule than is already out there. based on long experience collecting and analyzing performance and attitude data, and observing the way this parade is conceived ... the parade is the thing, not the prediction.

a point i omitted before that is crucial: the database itself is corrupt. not the observer data, but the WDS data that you want to use as your basis for "prediction". please skim the topic here on jonckheere doubles, and see what you conclude.

greg made the point that a rule that actually doesn't reliably "predict" anything is still useful to entice you to *look at* something -- as buddy says, to tell the observer in advance that something "is worth spending my time on." but a blindly applied *categorization rule* can't tell you that, only a consensus of observers can tell you that. you're pretending a rule leads you somewhere that you wouldn't go on your own. that to me is the crux.

everyone can make a pretty good guess about what is "worth looking at" using only primary magnitude, magnitude difference, binary separation and telescope aperture -- they just need to form an expectation, look, and then think for a minute about why their expectation was or was not met. in my obstreperous opinion, that is both a faster way to develop your observing skills, and a more rewarding form of observing.

but ... this is a hobby of diff'rent strokes, and good on us all.

#32 stray1

stray1

    Mariner 2

  • -----
  • Posts: 275
  • Joined: 03 Sep 2012
  • Loc: SW Ohio

Posted 30 October 2012 - 01:32 AM

I'm in.

#33 cphk96

cphk96

    Explorer 1

  • -----
  • Posts: 85
  • Joined: 01 Nov 2007
  • Loc: North Hollywood, CA

Posted 01 November 2012 - 09:40 AM

I want to take part in this project.

#34 azure1961p

azure1961p

    Voyager 1

  • *****
  • Posts: 10430
  • Joined: 17 Jan 2009
  • Loc: USA

Posted 02 November 2012 - 06:56 PM

I invite you all to read the September 2012 issue of Sky and Telescope(It's at page 68).

Sissy Haas is asking our help to complete a very interesting project. I invite you all to participate!


Well that's a nice invite but really, if its that important to her and these results through this forum could be that valuable why are you sitting in for her with the request? Wouldn't it serve her interests to get involved here like you.



Pete

#35 Cotts

Cotts

    Just Wondering

  • *****
  • Posts: 4896
  • Joined: 10 Oct 2005
  • Loc: Toronto, Ontario

Posted 05 November 2012 - 09:17 AM

Whether Ms. Haas is a member here or posts here or not is irrelevant. It has come to the attention of many of us (through this thread or via S&T) that she is requesting help and some of us are volunteering to do so.

Dave






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics