[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[gnuastro-devel] [task #14689] faster ageofuniverse in cosmology library

From: Boud Roukema
Subject: [gnuastro-devel] [task #14689] faster ageofuniverse in cosmology library + other cosmo extensions
Date: Tue, 3 Oct 2017 10:53:26 -0400 (EDT)
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Firefox/52.0


                 Summary: faster ageofuniverse in cosmology library + other
cosmo extensions
                 Project: GNU Astronomy Utilities
            Submitted by: boud
            Submitted on: Tue 03 Oct 2017 02:53:24 PM UTC
         Should Start On: Tue 03 Oct 2017 12:00:00 AM UTC
   Should be Finished on: Tue 03 Oct 2017 12:00:00 AM UTC
                Category: Libraries
                Priority: 5 - Normal
              Item Group: Enhancement
                  Status: In Progress
                 Privacy: Public
        Percent Complete: 20%
             Assigned to: boud
             Open/Closed: Open
         Discussion Lock: Any
                  Effort: 0.00



for a proof-of-concept of a 100x speed up for Universe ages.
It's a bit hacked but enough to get started on
improving the cosmo functions, I think.

I get the output from commit dd4443a :

ageofuniverse benchmark: old 1.81705  new 0.013959, old/new 130.17

I don't think we'll get much of a speed-up in proper distances
compared to (ignore the comments
there about luminosity distance being the main distance; it's
main in terms of observational analysis, but proper distance
is simpler in terms of the FLRW models.)

We would need an analytical integral of (sinh x)^(-2/3), because
d_{radial_comoving} = \int c dt/a(t) - see

The best that
would be to rewrite it in terms of \int (sinh x)^(4/3) , which
wouldn't help much. :P

We can implement what looks best for proper distances after re-reading

The non-flat cases should be added too. 


Reply to this item at:


  Message sent via/by Savannah

reply via email to

[Prev in Thread] Current Thread [Next in Thread]