Stellar Proper motions are not right

[Andrew James]

I noticed in Stellarium 14.1beta (and back to Version 12) there is a problem with the proper motions.

If you look at Alpha Centauri, the future motion is moving northwards at a PA of about 345 degrees, however that is not possible. If the HIP2 parallax is -3679.25 and +473.67 mas/year, the position angle should be about 277.3 degrees, calculated as tan-1(-3679.25/473.67) in Quadrant 4. I.e. It should be heading just little north of Beta Centauri, not northwards!.

Yet if you select the star HIP 57367 (11.5 magnitude) in Musca, which has an equally large proper motion but is in the opposite direction. I.e. +2663.91 and -347.38 mas/ yr, the position angle is +97.4 degrees.

The motion in Stellarium is seemingly wrong for Alpha Centauri but appears roughly correct for HIP 57367.

I also checked , being roughly correct.

Arcturus, pm. -1093.99, -2000.16 mas/yr, moving in PA 208.7 deg.
Sirius , pm. -546.01, -1223.07, PA 204.8 deg.
Barnard's Star -798.71, 10337.2 mas/yr, PA 353.6 deg.

1) Please check the calculations when adding proper motions.
2) Could someone point out the C++ file that does these proper motions calculation?
3) Are the proper motion calculated independently of the precession, as some stars move as arcs against the background stars in Stellarium rather than straight lines? (I.e. Follow the motion of HIP 57367.)

Alpha Centauri is a critical star because of its brightness and being the closest star, and educationally is important to many who us Stellarium. It is critical that something is done to fix these apparent errors.

Kind regards and appreciation to its solution,
And hope this is useful.

Andrew

I use Mac OS X 10.7.5, ATI Radeon HD 4670 256MB graphics

Note 1 : it is possible there is something wrong in solving the tan ambiguity here, where the PA is being calculated incorrectly.

Note 2 : As previously noted, SIMBAD gives the following proper motions;

Alpha Centauri : -3608, 686 mas/yr [PA 280.8 deg]
http://simbad.u-strasbg.fr/simbad/sim-id?Ident=alpha+cen&NbIdent=1&Radius=2&Radius.unit=arcmin&submit=submit+id

Alpha Centauri A : -3679.85, 473.67 mas/yr [PA 277.3 deg]
http://simbad.u-strasbg.fr/simbad/sim-id?Ident=%403396054&Name=*%20alf%20Cen%20A&submit=submit

Alpha Centauri B : -3614.39, 802.98 mas/yr [PA 282.5 deg]
http://simbad.u-strasbg.fr/simbad/sim-id?Ident=%403396297&Name=*%20alf%20Cen%20B&submit=submit

As the stars are labelled individually, the two components appear to diverge and are moving wider apart, rather than towards their common mutual shared motion.

The HIP2 file should be modified to either Alpha Centauri's mean motion of -3608, 686 mas/yr [PA 280.8 deg], or the mean motion of AB -3648.62, 638.33 mas.yr (HIP2) [PA 279.9 deg]. This number should be assigned to both the A and B stars in Stellarium's HIP2 star file.

[Alexander Wolf (alexwolf)]

The bug is partially crossed with: https://bugs.launchpad.net/stellarium/+bug/1271270

[Alexander Wolf (alexwolf)]

>2) Could someone point out the C++ file that does these proper motions calculation?

You need StarWrapper.hpp and Star.hpp

[gzotti (georg-zotti)]

> 3) Are the proper motion calculated independently of the precession, ...

Yes. Precession is only computed as motion of earth's axis which is reflected in motion of coordinate systems "Ecliptical of date" and "Equatorial of date".

[Andrew James (andjames)]

Thanks for the replies here, and apologies for the delay in response. Correcting code that someone else has written is hard work, and I've spent much time looking how tings work.

To gzotti, I ask this obvious question "Are the proper motion calculated independently of the precession,..." just to confirm that there is no other routine affecting this.

(I thought these proper motions seem to curve against the background sky, suggesting another motion added into this proper motion. Any motion should be almost linear. If this is not true, then looking at the calculated magnitudes, there should be a problem too. I note that some other User comments on apparent star disappearing and reappearing when accelerating overtime, may also be the cause of these issues too.)

Thanks for the response.

To Alex, I couldn't find the routine in StarWrapper.hpp or Star.hpp, but a routine does exist in ParseHip.C and ParseHip.v2.c. I.e.

==============
PM VECTOR ROUTINE
==============
     void ChangeEpoch(double delta_years,
                    double &ra,double &dec, // degrees
                    double pm_ra,double pm_dec // mas/yr
                   ) {
     ra *= (M_PI/180);
     dec *= (M_PI/180);

     const double cdec = cos(dec);
     Vector x = {cos(ra)*cdec,sin(ra)*cdec,sin(dec)};
     const Vector north = {0.0,0.0,1.0};

     Vector axis0 = north ^ x;
     axis0.normalize();
      const Vector axis1 = x ^ axis0;

      const double f = delta_years*(0.001/3600)*(M_PI/180);

      x += pm_ra*f*axis1 + pm_dec*f*axis0;

     ra = atan2(x.x[1],x.x[0]);
     if (ra < 0.0) ra += 2*M_PI;
     dec = atan2(x.x[2],sqrt(x.x[0]*x.x[0]+x.x[1]*x.x[1]));

     ra *= (180/M_PI);
     dec *= (180/M_PI);

==========

The tan ambiguity (atan2) seems absolutely fine here.

I suspect that two possible problems are here;

1) That for some reason the HIP2 data is in error, either switched around or rounded in correctly.
2) Or somethings wrong with the pm_ra or pm_dec equation measured in mas per year.

As for the two stars flying apart there is a routine to correct for this ParseHip.C (lines 171-240) under mergeSameComponents (line 213), suggest the flag in the .dat file has not been set. [I'd suggest writing an additional routine for special cases of double stars that have not been flagged.[

=======================
Past and Future Magnitudes
=======================

A further problem, that might be related to this is that 29,240+/-1370 AD, Alpha Centauri makes its closest approach to the Sun at 2.970+/-0.012 light years when the parallax is 1.098 mas (See Mattews (2004)). The apparent total visual magnitude -1.05.

In Stellarium 14 here, on that date, finds the result is still 4.39 light years and the magnitude of both components is +0.10 and +1.20. (Total about -0.2V magnitude.) Also Sirius remains at -1.45, even though it significantly changes too. In ParsHip.C (lines 154-253) there seems no formal correction for the changes in the parallax affecting these apparent magnitudes.

If this is not correct, then I'd suspect the proper motions changing values (specifically accelerating or deaccelerating) over time are not being account for either. I.e. Line 139 of...

[Alexander Wolf (alexwolf)]

>To Alex, I couldn't find the routine in StarWrapper.hpp or Star.hpp, but a routine does exist in ParseHip.C and ParseHip.v2.c. I.e.......

See here for example: http://bazaar.launchpad.net/~stellarium/stellarium/trunk/view/head:/src/core/modules/StarWrapper.hpp#L76
http://bazaar.launchpad.net/~stellarium/stellarium/trunk/view/head:/src/core/modules/Star.hpp#L109

I guess that reducing of the epoch in the tool for convert original catalog into Stellarium's format has no sense here.

> A further problem, that might be related to this is that 29,240+/-1370 AD, Alpha Centauri makes its closest approach to the Sun at 2.970+/-0.012 light years when the parallax is 1.098 mas (See Mattews (2004)). The apparent total visual magnitude -1.05........

How exactly we can change magnitudes for our case: Stellarium's catalogues has no 3D data for proper motion and parallax? It can be solved only through creating a new catalog with all vectors and absolute magnitudes. But in any case tool for convert catalogs have no sense for main application.

> Routine in Lines 304-346 makes corrections for B_T and V_T HIP magnitudes for spectral class then star colours, however, they do not convert these two magnitudes into B and V Johnston equivalents.

Thanks for link, I think it has sense for prepare updates for the star catalogues.

[Jerome Colburn (jscolburn)]

The root of the problem is that Hipparcos didn't measure radial velocities, so they aren't stored in its catalogue. For most stars, that's not a problem, but it is for nearby stars such as Alpha Cen and Sirius. A new column would have to be added to the catalogue file, the values would have to be added manually to that, and the routine would have to be written to combine the proper motion and the radial velocity into a space velocity vector, project the star's motion along that, and reconvert the result to a sky position.

[Andrew James (andjames)]

Jerome
I think you might be understating the whole issue here. The lack of a radial velocity does not account for the apparent direction, which only changes pmRA and pmDec by small amounts.

Alpha Cen A (Accounting for pm and radial velocity at Epoch 2000.0)
14 39 34.3 -60 49 39.4 (2000)
15 54 21.7 -64 23 00.8 (3000)

Without added pm this is:
16 04 05.3 -64 22 14.5 (3000)

Even accounting for epoch of date at 3000.0AD, this position is:
15 54 21.9 -64 23 01.1 (3000)

Yet, Stellarium 13.3 gives
14 39 36.5 -60 50 02.3 (2000) Epoch J2000
15 56 44.6 -63 36 26.0 (3000) Epoch 3000 (*)

14 34 17.2 -59 57 06.8 (3000) Epoch J2000 (Which also seems wrong.)

Here, the declination (*) is moving too far north by 1 deg 14' 13.5" in 1000 years, as explained be me above. Hence, there is something wrong with the pm calculation and NOT just merely a lack of a radial velocity. (This explains Stellarium's apparent incorrect observed motion, and my actual point.)

Furthermore, Stellarium 15.0 now uses XHIP, which does have the radial velocity included.

Regards,
Andrew James

Note 1: It should be noted that the epoch for proper motions is set at 2000.0, but as time progresses, the values of the proper motions change slightly. Also , especially for alpha Centauri, mean radial velocity of AB system will change over the centuries. It will became smaller until c.36,000AD, zero, the move away with its opposite sign.

Note 2: With Stellarium using planetary positions from JPL DE430 and DE431 ephemerides, the but the actual stellar positions being incorrect at any date, means the planet displayed places are not accurate by the stars. A really BIG problem methinks!!!

[Alexander Wolf (alexwolf)]

We waiting the first release Gaia's data for improvement star catalogs.

[Andrew James (andjames)]

Alexander

Gaia is supposed to be able to measure stars above 12th magnitude, and those brighter will have significantly lower accuracy. None of the stars above 2nd to 3rd will be able to be measured at at all, leaving the exact problem so far expressed above. Likely, for all the naked-eye stars, Hipparcos will remain the primary source, and the spectra for radial velocities will rely on XHiP or other sources. Of course, also dealing with G-magnitudes to visual magnitudes will require some thought.

[Andrew James (andjames)]

Oops!!

The previous post should read:

"Gaia is supposed to be able to measure stars BELOW 12th magnitude,"

Sorry 'bout that!!

[Alexander Wolf (alexwolf)]

Copy on Github: https://github.com/Stellarium/stellarium/issues/85

[gzotti (georg-zotti)]

Set to invalid here because it will be handled now via Github.