Re: pgsphere

From: Jan-Benedict Glaw <jbglaw(at)lug-owl(dot)de>
To: Oleg Bartunov <oleg(at)sai(dot)msu(dot)su>
Cc: Andrew Dunstan <andrew(at)dunslane(dot)net>, Dave Cramer <pg(at)fastcrypt(dot)com>, karpov(dot)sv(at)gmail(dot)com, Pgsql Hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: pgsphere
Date: 2012-01-10 21:53:22
Message-ID: 20120110215322.GD21941@lug-owl.de
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On Wed, 2012-01-11 00:43:25 +0400, Oleg Bartunov <oleg(at)sai(dot)msu(dot)su> wrote:
> On Tue, 10 Jan 2012, Jan-Benedict Glaw wrote:
> > On Sun, 2012-01-08 22:19:53 +0400, Oleg Bartunov <oleg(at)sai(dot)msu(dot)su> wrote:
> > > pgsphere now could benefit very much from our KNNGiST feature.
> > > This is number one development from my point of view. I and
> > > Teodor have no time to work on pgsphere, sorry. But, there are
> > > some astronomers I'm working with, who can take part in this.
> > > Sergey Karpov has done extensive benchmarks of q3c, rtree and
> > > pgsphere and found the latter still has some benefits in some
> > > workload, so we are interesting in development.
> > Could the PostGIS stuff be abused for stellar coordinates?
>
> There is no principal difference between celestial sphere and earth,
> it's a matter of conversion between coordinates.

I'm a hobby astronomer myself--so I'm asking myself what is actually
needed. I had a look at pgsphere some weeks ago, but didn't use it,
because it seemed to be somewhat dead.

My next approach was to load a dataset as-is (with floting point RA
and dec) and then created a new table selecting all data and converted
the coordinates to POINTs. Even with a combined index on (RA,Dec) (as
well as one on the POINT column of the new PostGIS enabled table), the
later was quite faster when searching for specific areas etc.

One important thing that's needed is transformation between
equatorial, ecliptic, galactic and probably (local, incorporating
current local longitude/latitude and time) horizontal
coordinates. What might be important, too, is to be able to change
between J2000.0 and B1950.0 etc.

The probably easiest thing is to change the (printable) representation
of coordinates, because (depending on the people you talk to),
especially for RA (equivalent to the earth's longitude), there are at
least two totally different notations used:

2h 30min 4.3sec = 2.5011944444h = 37.5179°

The other axis is usually simply written in degree.

MfG, JBG

--
Jan-Benedict Glaw jbglaw(at)lug-owl(dot)de +49-172-7608481
Signature of: Wenn ich wach bin, träume ich.
the second :

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Joachim Wieland 2012-01-11 01:16:22 Re: Sending notifications from the master to the standby
Previous Message Greg Smith 2012-01-10 21:46:49 Re: pgsphere