No subject
Mon Jan 26 18:22:44 EST 2009
if it were my project. But in spite of my value being very close to one of
them but not the other doesn't necessarily mean that the latter value is
"wrong". The total spread among the 3 estimates is about 2.5 sigma's (just
comparing the values and their uncertainties without doing a formal
statistic). Although this is small, I would tend to use one of the two
values that agree. But you should run some more test cases.
I can offer you some overall advice but nothing specific about software that
I didn't write.
In general, using the same algorithm in different software does not assure
that the result will be the same. There is a difference between "best
results" and "correct results". Seldom is there a "correct" result unless
we're dealing with simple arithmetic. When using any but the simplest, most
straightforward algorithms (like adding images), there is a lot of numerical
analysis and coding involved. Differences in software implementations can
lead to results that may---or may not---indicate "errors" or bugs. A common
misconception is that if product A has feature X and software B has feature
X, then both products are equivalent. This misconception applies to software
as well as other products and it forms the basis of a lot of marketing
strategies! Getting different results may just be a consequence of the
"fuzziness" of the mathematical concepts or numerical techniques involved.
As a classic example, the Mode of a distribution is mathematically well
defined, yet the computed mode of an arbitrary data set will vary among
software implementations and among people who estimate it manually! In the
case at hand, the KvW algorithm involves interpolation and binning, a
minimum value hunt, and other numerical techniques where small differences
could lead to differing results that aren't necessarily bugs. Or they might
indicate bugs---you just don't know unless you do some further study.
Here are a few general suggestions for getting "good" results:
-- First and foremost, use software implementations for which you trust that
the authors are experts in the area. This usually give the best results but
is not a guarantee.
-- Test by running cases where you know what the answer should be and/or by
looking at numbers and graphs and/or working out the answer yourself---only
then can you evaluate whether the "black box" gave you what you would
expect. This is why I sent you a graph of your observations with the
calculated T(min) marked. On the other hand, even the most trusted experts
can make mistakes or may later discover better techniques.
-- Compare results from multiple programs (ones that you would otherwise
trust) and see who agrees and who doesn't. This often exposes problems but
is no guarantee. You still have to judge which results agree with what you
determine the result should be. My advisor in grad school, Bob Kirshner,
imparted a valuable bit of simple wisdom that applies here: Get at least 3
independent results whenever possible: One tells you nothing and, unless 2
are identical, you don't know which is the better of the two.
Michael Newberry
----- Original Message -----
From: "Yenal Ogmen" <yenalogmen at yahoo.com>
To: "Michael Newberry" <mnewberry at mirametrics.com>
Sent: Tuesday, February 17, 2009 10:34 AM
Subject: Re: [Aavso-photometry] Peranso vs Minima V2.3 question
Michael thank you very much. I run both Minima and Peranso with the same
data set with the same margins. Results are:
Minima: JD = 2454865.33322 +/-0.00007
Peranso: JD = 2454865.333388 +/-0.000045
As you see, Minima's value agree with yours. I didnt understand why they
have different results with same margins. They both use the same algorithm.
What can you say about this?
Yenal Ögmen
Green Island Observatory (B34)
www.greenislandobservatory.com
Geçitkale
North Cyprus
----- Original Message ----
From: Michael Newberry <mnewberry at mirametrics.com>
To: Yenal Ogmen <yenalogmen at yahoo.com>
Cc: aavso-photometry at mira.aavso.org
Sent: Tuesday, February 17, 2009 5:48:01 PM
Subject: Re: [Aavso-photometry] Peranso vs Minima V2.3 question
Yenal,
Thanks! That is a fine data set. You have about twice as many observations
on the T(min-) side as on the T(min+) side, but that should not be a
problem. Using your entire data window (125 observations over 0.118 days), I
get time of minimum JD = 2454865.333224 +/- 0.000065. I attached a plot
showing your observations with T(min) marked. Your observations are spaced
82 seconds on average, and my uncertainty is +/- 6 sec.
I don't know what values you got with the other two KvW implementations, but
I hope my result helps you sort out the problem.
Regards,
Michael Newberry
----- Original Message ----- From: "Yenal Ogmen" <yenalogmen at yahoo.com>
To: "Michael Newberry" <mnewberry at mirametrics.com>
Sent: Monday, February 16, 2009 1:40 PM
Subject: Re: [Aavso-photometry] Peranso vs Minima V2.3 question
Mr. Newberry,
Thank you for your reply. Can you send your estimate for the attached minima
please?
Best regards,
Yenal Ögmen
Green Island Observatory (B34)
www.greenislandobservatory.com
Geçitkale
North Cyprus
----- Original Message ----
From: Michael Newberry <mnewberry at mirametrics.com>
To: Yenal Ogmen <yenalogmen at yahoo.com>; aavso-photometry at mira.aavso.org
Cc: bob.nelson at shaw.ca
Sent: Monday, February 16, 2009 10:32:11 PM
Subject: Re: [Aavso-photometry] Peranso vs Minima V2.3 question
I have a Kwee-Van Woerden program lying around somewhere in my archives that
is not related to either Minima or Peranso. If I could get hold of the same
data set, I would be happy to provide an independent estimate of the time of
minimum to see how it compares.
Michael Newberry
----- Original Message ----- From: "Yenal Ogmen" <yenalogmen at yahoo.com>
To: <aavso-photometry at mira.aavso.org>
Cc: <bob.nelson at shaw.ca>
Sent: Sunday, February 15, 2009 4:29 AM
Subject: Re: [Aavso-photometry] Peranso vs Minima V2.3 question
Thank you all who replied. It really helped. As I am the new user of both
programs, I did some mistakes and these mistakes lead me to wrong
conclusions. Now I am ok.
I tried Minima 25d but when I click on GO button after loading the file it
gives "Run-time error 5, Invalid procedure call or argument". Mr. Renz, did
you use it before? If not, may be this version is not ready and may be thats
why Mr. Nelson didnt put link for 25d on his page
http://members.shaw.ca/bob.nelson/software1.htm but for 25c.
Yenal Ögmen
Green Island Observatory (B34)
www.greenislandobservatory.com
Geçitkale
North Cyprus
----- Original Message ----
From: Wolfgang Renz <w_renz at onlinehome.de>
To: Yenal Ogmen <yenalogmen at yahoo.com>
Cc: AAVSO Photometry <aavso-photometry at mira.aavso.org>
Sent: Saturday, February 14, 2009 11:46:03 PM
Subject: Re: [Aavso-photometry] Peranso vs Minima V2.3 question
Hi Yenal
You mean the "Minima" Software by Bob Nelson ?
http://members.shaw.ca/bob.nelson/software1.htm
http://members.shaw.ca/bob.nelson/Software/Minima25d.zip
the most current software version is V2.5d dated 2009-02-12.
Did you try it with the most current version too ?
If you want detail, you must probably ask the authors directly.
> ~5 min is not a small difference right?
Right.
> Kwee-van Woerden
This method is basically an algorithm for processing equidistant data
of symetric LCs. So it applies for symetric EB LCs, but not for asymetric
ones including nearly all LCs of pulsators.
See e.g.:
1982IAPPP...7...14M - A FORTRAN Subroutine for Determining Times
of Minimum Light - Mallama, A. D.
http://adsabs.harvard.edu/abs/1982IAPPP...7...14M
Using slightly non-equidistant data might not have a very large
effect.
But if the LC is asymetric, the result will depend on the choosen
margins and the ToM will move more or less strongly depending
on the grade of the asymetry.
Its the same as with the trace-paper-flip method and the polynom
method that just includes even exponents.
But it shouldn't be an issue with the Pogson's half chord method
and the polynom method that also includes uneven exponents.
> Am I doing something wrong with Peranso?
Probably not.
Is your LC asymetric ?
Then this could explain the differences of the derived ToM with
Peranso.
If your LC is asymetric (and the minima stay fixed), Minima might
ignore the margin settings. To check this, feed the program with
different subsets of the data that contains just the data between the
wanted margins.
Clear skies
Wolfgang
-- Wolfgang Renz, Karlsruhe, Germany
Rz.BAV = WRe.vsnet = RWG.AAVSO
> ----- Original Message ----- From: "Yenal Ogmen"
> To: <Aavso-photometry at mira.aavso.org>
> Sent: Saturday, February 14, 2009 3:47 PM
> Subject: [Aavso-photometry] Peranso vs Minima V2.3 question
>
> Hi all,
>
> I just want to ask why I have different (~0.003 day=~5min) results of
> minima of the LC for the same file and the same left and right margins
> used both in MinimaV2.3 and Peranso?
> Both use the same algorithm: Kwee-van Woerden.
> ~5 min is not a small difference right?
> And I also observe that when you change the positions of the margins,
> Minima always give the same result but on the other hand Peranso's
> result changes when you change margins' position and this change is
> around 0.003 days. I am surpised with this as Minima is freeware and
> Peranso is not.
> Am I doing something wrong with Peranso?
>
> Thank you in advance.
>
> Yenal Ögmen
> Green Island Observatory (B34)
> www.greenislandobservatory.com
> Geçitkale
> North Cyprus
_______________________________________________
Aavso-photometry mailing list
Aavso-photometry at mira.aavso.org
To change options or unsubscribe, goto
http://www.aavso.org/mailman/listinfo/aavso-photometry
More information about the Aavso-photometry
mailing list