stars.txt

Discussion forum for Celestia developers; topics may only be started by members of the developers group, but anyone can post replies.
Avatar
Topic author
t00fri
Developer
Posts: 8772
Joined: 29.03.2002
Age: 18
With us: 18 years 3 months
Location: Hamburg, Germany

stars.txt

Post #1by t00fri » 06.03.2007, 13:19

Chris,

my PERL script for visualbins.stc uses the ascii data/stars.txt file of the distribution. It seems, I must have misinterpreted the units of the distance values in that file!

So in what units, is the distance (4th column) in stars.txt??

I thought it should be light years, but it isn't.

[EDIT: it actually is, see further below]

Clearly:

1st column = HIP number
2nd column = RA in degrees
3rd column = Dec in degrees
4th column = distance in ???
5th column = spectral class

Here is the entry for 85 Peg, where we know that d ~ 40 ly.

171 0.54018776 27.08448905 1203.56825 5.79688 G3V

What is 1203.56825 ?? One factor that gets me into the right ball park would be LY_PER_PARSEC/100.0

But what is 100.0??

I was unable to reproduce stars.txt (--spherical) from stars.dat by means of 'startextdump.exe' , since for new stars the --spherical option is ignored...

The documentation of this stuff is rather nonexistent...

Bye Fridger
Last edited by t00fri on 06.03.2007, 16:41, edited 1 time in total.
Image

ajtribick
Developer
Posts: 1803
Joined: 11.08.2003
With us: 16 years 10 months
Location: Switzerland

Post #2by ajtribick » 06.03.2007, 13:40

From a quick check of stars.txt and other stars, it looks like the value is in light years.

However, 85 Pegasi is located close (in the sky plane) to HIP 162, which has a parallax of 2.71 milliarcseconds (1203 light years).

Apparently this tripped some code which is supposed to keep binary stars together.

It would be useful to have a list of pairs of stars for which this has happened.

Avatar
Topic author
t00fri
Developer
Posts: 8772
Joined: 29.03.2002
Age: 18
With us: 18 years 3 months
Location: Hamburg, Germany

Post #3by t00fri » 06.03.2007, 16:39

chaos syndrome wrote:From a quick check of stars.txt and other stars, it looks like the value is in light years.

However, 85 Pegasi is located close (in the sky plane) to HIP 162, which has a parallax of 2.71 milliarcseconds (1203 light years).

Apparently this tripped some code which is supposed to keep binary stars together.

It would be useful to have a list of pairs of stars for which this has happened.


Aha Chaos,

you seem to have a point here.

Simbad gives as parallaxes for HIP 171 (85 Peg) => 80.63 mas , while HIP 162 => 2.71 mas. Taking the known distance of 85 Peg as d=40.5 ly then gives

d= 40.5 * 80.63/2.71 = 1205 ly for HIP 162,

in perfect agreement with the distance entry [ly] for HIP 162 in stars.txt. Clearly then the /identical/ entry

d (HIP 171) = d(HIP 162) = 1203.5682

MUST be wrong. Your above reasoning about the origin of this wrong entry seems to make perfectly sense.
+++++++++++++++++++++
So Chris might have to revise his code for identifying close-by stars as binary systems, and attributing this way incorrect distances...
+++++++++++++++++++++

I also suspect that 85 Peg is not the only erroneous fellow!

I am happy to conclude that this also means that my incorrect entry in visualbins.stc for 85 Peg was actually a problem with stars.txt and the code that produced it, and NOT with my PERL code... ;-)

Bye Fridger
Image

chris
Site Admin
Posts: 4211
Joined: 28.01.2002
With us: 18 years 5 months
Location: Seattle, Washington, USA

Post #4by chris » 06.03.2007, 19:31

chaos syndrome wrote:Apparently this tripped some code which is supposed to keep binary stars together.

It would be useful to have a list of pairs of stars for which this has happened.

Celestia's star database builder uses records from the HIPPARCOS database to determine membership in a multiple star system, and so the stars 85 Peg and HIP 162 are apparently erroneously listed as a binary pair in the HIPPARCOS catalog. I'll have to verify this, but I can assure you that the database builder does not mark stars as binaries based on apparent proximity.

t00fri wrote:I also suspect that 85 Peg is not the only erroneous fellow!


Probably not . . . I believe that there is a published list HIPPARCOS errata that we should have a look at.

--Chris

Avatar
Topic author
t00fri
Developer
Posts: 8772
Joined: 29.03.2002
Age: 18
With us: 18 years 3 months
Location: Hamburg, Germany

Post #5by t00fri » 06.03.2007, 19:51

Here is the original output of HIP & Tycho

Full HIP RA hms DE dms Vmag RA(ICRS) DE(ICRS) Plx pmRA pmDE e B-V Notes
mag deg deg mas mas/yr mas/yr mas mag

1 171 00 02 09.65 +27 05 04.2 5.80 0.54018776 27.08448905 80.63 778.59 -918.72 3.03 0.690
2 162 00 02 03.21 +27 07 07.0 9.84 0.51336753 27.11861943 2.71 -4.89 -6.90 1.33 0.696

All data are correct unlike stars.txt. Parallax is 80.63 mas and 2.71 mas for HIP 171 and HIP 162, respectively.

The data in stars.dat are also incorrect as to the distance of HIP 171 (1203 ly).

The above version of the catalog was published in 1997 (ESA). Celestia was not yet alive in 1997 ;-) . So why was this catalog not read out?

+++++++++++++
What is the official publication reference of the star data in the global database, used for Celestia ??
+++++++++++++

Bye Fridger
Image

chris
Site Admin
Posts: 4211
Joined: 28.01.2002
With us: 18 years 5 months
Location: Seattle, Washington, USA

Post #6by chris » 06.03.2007, 20:38

t00fri wrote:All data are correct unlike stars.txt. Parallax is 80.63 mas and 2.71 mas for HIP 171 and HIP 162, respectively.

The data in stars.dat are also incorrect as to the distance of HIP 171 (1203 ly).

The above version of the catalog was published in 1997 (ESA). Celestia was not yet alive in 1997 ;-) . So why was this catalog not read out?


It was, but distances were adjusted for stars listed in the HIPPARCOS catalogs as being components of multiple star systems. I did this so that stars that were known to be gravitationally associated with each other would actually be close to each other in three dimensions. If you go to this form:

http://archive.ast.cam.ac.uk/hipp/hipparcos.html

and look up both HIP 162 and HIP 171. You'll see that these stars share the same CCDM record. The CCDM is the Catalog of Components of Double and Multiple stars. There's also a flag called CCDM History; the setting for these two stars indicates that they had been flagged as a binary pair prior to HIPPARCOS. I think that this is an error in the catalog. But it's also possible that the star database builder may be making inappropriate assumptions about CCDM record in the HIPPARCOS catalog.

--Chris

Avatar
Topic author
t00fri
Developer
Posts: 8772
Joined: 29.03.2002
Age: 18
With us: 18 years 3 months
Location: Hamburg, Germany

Post #7by t00fri » 06.03.2007, 20:40

Getting suspicious about (undocumented) stars.txt, I printed out in my visualbins.pl script those visual binay systems, where the distance from stars.txt differes more than 10% from that used in Soederhjelms original paper.

Here are the results:

Code: Select all

Distance deviation of 2944.31325763293 % for HIP 171
Distance deviation of 19.0489065566916 % for HIP 20916
Distance deviation of 14.6950288015865 % for HIP 40239
Distance deviation of 21.9588547705953 % for HIP 71914
Distance deviation of 560.778855054916 % for HIP 85582


You can see that HIP 171 is not the only certainly wrong case. Also HIP 85582 is certainly wrong! The others are possibly wrong...

Bye Fridger
Image

Avatar
selden
Developer
Posts: 10136
Joined: 04.09.2002
With us: 17 years 10 months
Location: NY, USA

Post #8by selden » 06.03.2007, 20:58

My understanding has been that stars.txt is strictly generated from the original, uncorrected Hipparcos database. No manual corrections are included in that file. Errors in the Hipparcos database are documented and corrected by the entries in the catalog file /data/revised.stc
Last edited by selden on 06.03.2007, 21:03, edited 1 time in total.
Selden

chris
Site Admin
Posts: 4211
Joined: 28.01.2002
With us: 18 years 5 months
Location: Seattle, Washington, USA

Post #9by chris » 06.03.2007, 21:03

t00fri wrote:You can see that HIP 171 is not the only certainly wrong case. Also HIP 85582 is certainly wrong! The others are possibly wrong...


I think that the distance of ~524 light years for HIP 85582 in visualbins.stc is incorrect. SIMBAD gives its parallax as 42.40 mas (76.9 ly). It also has a high proper motion, so I'd tend to favor the closer distance estimate from SIMBAD than the value from the Soederhjlems paper. I haven't had a chance to look at the others.

--Chris

Avatar
Topic author
t00fri
Developer
Posts: 8772
Joined: 29.03.2002
Age: 18
With us: 18 years 3 months
Location: Hamburg, Germany

Post #10by t00fri » 06.03.2007, 21:07

chris wrote:
t00fri wrote:All data are correct unlike stars.txt. Parallax is 80.63 mas and 2.71 mas for HIP 171 and HIP 162, respectively.

The data in stars.dat are also incorrect as to the distance of HIP 171 (1203 ly).

The above version of the catalog was published in 1997 (ESA). Celestia was not yet alive in 1997 ;-) . So why was this catalog not read out?

It was, but distances were adjusted for stars listed in the HIPPARCOS catalogs as being components of multiple star systems. I did this so that stars that were known to be gravitationally associated with each other would actually be close to each other in three dimensions. If you go to this form:

http://archive.ast.cam.ac.uk/hipp/hipparcos.html

and look up both HIP 162 and HIP 171. You'll see that these stars share the same CCDM record.


Yes, I know, that's also why the two were printed next to each other in the query-output in my post above yours. I used just another query form (CDS Strasbourg data base) , but it's the same catalog. From the CCDM dat it is obvious that the misidentification as component of a multiple star system is partly from 1875 and older than ~1935 for each component.

From glancing through your relevant routines I have the suspicion that after you read in and evaluated multiple star info, the correct HIP parallaxes were overwritten by the incorrect 1203 ly distance value due to the presence of the multiple star flag. Right?

In any case the HIP catalog has the correct parallax for the two stars considered. The code should be adapted to take preference of the parallax values if they exist with decent error and are different.


Bye Fridger
Image

symaski62
Posts: 602
Joined: 01.05.2004
Age: 37
With us: 16 years 2 months
Location: france, divion

Post #11by symaski62 » 06.03.2007, 21:07

hm ? http://www.celestiaproject.net/~claurel/celest ... b/?C=D;O=D

startextdump.exe 09-Sep-2004 23:27 188K
vista home basic
intel(R) Pentium(R) Dual CPU E2160 @ 1,80GHz 1,79GHz
1Go RAM
NVIDIA GeForce 8400 GS
directX 11
celestia 1.6
with a general handicap of 80% and it makes much d' efforts for the community and s' expimer, thank you d' to be understanding.

chris
Site Admin
Posts: 4211
Joined: 28.01.2002
With us: 18 years 5 months
Location: Seattle, Washington, USA

Post #12by chris » 06.03.2007, 21:10

selden wrote:My understanding has been that stars.txt is strictly generated from the original, uncorrected Hipparcos database. No manual corrections are included in that file. Errors in the Hipparcos database are corrected by the entries in the catalog file /data/revised.stc


There are no manual corrections applied to stars.txt, nor any corrections that rely on data from sources other than the HIPPARCOS catalog. The sole adjustment is aligning the distances of stars in the HIPPARCOS dataset that are marked as components of multiple systems. An argument could be made that any stars so modified should appear in an auxilliary catalog.

Note that the distance to 85 Peg is already fixed in revised.stc; the parallaxes of a handful of other stars have also been restored to their original HIPPARCOS values.

--Chris

Avatar
Topic author
t00fri
Developer
Posts: 8772
Joined: 29.03.2002
Age: 18
With us: 18 years 3 months
Location: Hamburg, Germany

Post #13by t00fri » 06.03.2007, 21:11

selden wrote:My understanding has been that stars.txt is strictly generated from the original, uncorrected Hipparcos database. No manual corrections are included in that file. Errors in the Hipparcos database are documented and corrected by the entries in the catalog file /data/revised.stc


Selden,

that is certainly NOT correct, since the 1997 ESA edition of HIPPARCOS as available in the global archives does have the correct parallax for both HIP 162 and HIP 171, unlike stars.txt. I have downloaded the 1997 catalog. You may also query it.

There is an early vaersion of the HIP catalog from 1992/3, but the official one is from 1997.
Image

Avatar
selden
Developer
Posts: 10136
Joined: 04.09.2002
With us: 17 years 10 months
Location: NY, USA

Post #14by selden » 06.03.2007, 21:20

I do remember discussions with Grant about the issue of undocumented "corrections" in old versions of stars.dat and that he was intending to undo them. Perhaps that never happened?
Selden

Avatar
Topic author
t00fri
Developer
Posts: 8772
Joined: 29.03.2002
Age: 18
With us: 18 years 3 months
Location: Hamburg, Germany

Post #15by t00fri » 06.03.2007, 21:21

chris wrote:
t00fri wrote:You can see that HIP 171 is not the only certainly wrong case. Also HIP 85582 is certainly wrong! The others are possibly wrong...

I think that the distance of ~524 light years for HIP 85582 in visualbins.stc is incorrect. SIMBAD gives its parallax as 42.40 mas (76.9 ly). It also has a high proper motion, so I'd tend to favor the closer distance estimate from SIMBAD than the value from the Soederhjlems paper. I haven't had a chance to look at the others.

--Chris


I disagree! This is the same type of stars.txt error as 85 Peg:

Here is the output of stars.txt:

85582 262.3342 29.392614 524.38426 8.98828 K2

You see that I got the incorrect value of d= 524.38426 ly once more by reading out stars.txt with visualbins.pl!

Soederhjelm's value is the correct one: d= 79.3585 ly and differs from the one in stars.txt by 561 % ;-)

Bye Fridger
Last edited by t00fri on 06.03.2007, 21:40, edited 4 times in total.
Image

Avatar
Topic author
t00fri
Developer
Posts: 8772
Joined: 29.03.2002
Age: 18
With us: 18 years 3 months
Location: Hamburg, Germany

Post #16by t00fri » 06.03.2007, 21:28

selden wrote:I do remember discussions with Grant about the issue of undocumented "corrections" in old versions of stars.dat and that he was intending to undo them. Perhaps that never happened?


Selden,

I think Chris and I agree here qualitatively. Stars.txt is basically the unmodified HIP catalog, EXCEPT if in the CCDM addition file, for a particular star, the multiple star flag was set. This is exploited in Chris code! Then the correct parallax-based distance is replaced via the starbuilder code by an incorrect common value. In some cases the code error has been noted and corrected (by hand) in revised.stc. That's the state of the "art" as I see it.

I think Chris code can rather be modified such that these mis-identifications are eliminated from the start.

Bye Fridger
Last edited by t00fri on 06.03.2007, 22:08, edited 2 times in total.
Image

symaski62
Posts: 602
Joined: 01.05.2004
Age: 37
With us: 16 years 2 months
Location: france, divion

Post #17by symaski62 » 06.03.2007, 21:29

http://celestia.cvs.sourceforge.net/cel ... e&view=log

Sun Sep 4 21:23:04 2005 UTC (18 months ago) by granthutchison

162 0.51336754 27.11862 1203.5682 9.83984 G0
171 0.54018776 27.08448905 1203.56825 5.79688 G3V
vista home basic
intel(R) Pentium(R) Dual CPU E2160 @ 1,80GHz 1,79GHz
1Go RAM
NVIDIA GeForce 8400 GS
directX 11
celestia 1.6
with a general handicap of 80% and it makes much d' efforts for the community and s' expimer, thank you d' to be understanding.

Avatar
Topic author
t00fri
Developer
Posts: 8772
Joined: 29.03.2002
Age: 18
With us: 18 years 3 months
Location: Hamburg, Germany

Post #18by t00fri » 06.03.2007, 21:55

Chris wrote:There are no manual corrections applied to stars.txt, nor any corrections that rely on data from sources other than the HIPPARCOS catalog.


Chris, this is not correct.

Go and look to Grant's respective logs in CVS along with his by-hand changes of stars.txt in the course of time...Symaski62 has obviously localized Grant as the origin of these poorly documented by hand modifications.

http://celestia.cvs.sourceforge.net/cel ... e&view=log

It's certainly NOT the kind of scientific documentation of catalog data that I am dreaming about...

Bye Fridger
Image

chris
Site Admin
Posts: 4211
Joined: 28.01.2002
With us: 18 years 5 months
Location: Seattle, Washington, USA

Post #19by chris » 06.03.2007, 22:46

t00fri wrote:
Chris wrote:There are no manual corrections applied to stars.txt, nor any corrections that rely on data from sources other than the HIPPARCOS catalog.

Chris, this is not correct.

Go and look to Grant's respective logs in CVS along with his by-hand changes of stars.txt in the course of time...Symaski62 has obviously localized Grant as the origin of these poorly documented by hand modifications.

http://celestia.cvs.sourceforge.net/cel ... e&view=log

It's certainly NOT the kind of scientific documentation of catalog data that I am dreaming about...


Note the comment by the most recent change:

Regressed to contain only Hipparcos-derived data

There should be no remaining hand modifications to stars.txt.

--Chris

chris
Site Admin
Posts: 4211
Joined: 28.01.2002
With us: 18 years 5 months
Location: Seattle, Washington, USA

Post #20by chris » 06.03.2007, 22:49

t00fri wrote:
selden wrote:I do remember discussions with Grant about the issue of undocumented "corrections" in old versions of stars.dat and that he was intending to undo them. Perhaps that never happened?

Selden,

I think Chris and I agree here qualitatively. Stars.txt is basically the unmodified HIP catalog, EXCEPT if in the CCDM addition file, for a particular star, the multiple star flag was set. This is exploited in Chris code! Then the correct parallax-based distance is replaced via the starbuilder code by an incorrect common value. In some cases the code error has been noted and corrected (by hand) in revised.stc. That's the state of the "art" as I see it.

I think Chris code can rather be modified such that these mis-identifications are eliminated from the start.


It would be easy to skip distance modification for stars that are listed as being part of a binary system, but have wildly differing parallaxes. But instead of doing that, I'd like to see if there's an official list of errata in the CCDM records of the data set.

--Chris


Return to “Ideas & News”

Who is online