Re: ITL Prayer Time

[abdulhaq <al-arabeyes at alinsyria dot fsnet dot co dot uk>]

Thamer, the new changes sound great, but I still have some concerns (please 
be patient with me).

I know that I don't know anything about astronomy but I did do a physics 
degree so I can talk mathematics (but I really DON'T know anything about 
astronomy).

I take your point about libnova not being a VSOP87 implementation in respect 
of these calculations (it does however provide VSOP87 for geometric solar 
coordinates etc).

However, I'd still like to put my mind to rest in this important matter. The 
yanabi website claims to have times accurate to +- 2 seconds, and also seems 
to very closely conform to the libnova generated data. I can't find any 
mention of marsbase.net claiming to use VSOP87 in its calculation. My point 
is that our times (ITL, abdali etc) are varying, as you point out, by 
minutes and yet it seems as if it is possible to calculate the time to the 
nearest two seconds.

The yanabi website only gives its answers to the nearest minute, which is 
rounded up for maghrib and down for shuruq. Also, dhuhr time is adjusted for 
the full disc of the sun to move off the highest point. This explains the 
discrepancy you found between libnova and yanabi for instance.

Anyway, I'll wait to see your changes and how the calculations compare after 
that. If you stick with the current general technique then I'd like to know 
the maximum error so that I can account for it, inshaa'allah. I can perhaps 
work this out experimentally after you finish. Have you found any data 
source which gives real actual times that we can use for comparison 
purposes? Libnova has a test suite but it doesn't seem to cover this issue.

Of course no solution is perfect, allahu a`lam, but I'd like it to be as good 
as is reasonably possible. If there is anything I can do to help out then 
let me know (although I appreciate that it is unlikely). In the mean time I 
will try to put a script together that does a more wholesale comparison of 
the different programs.

I apologise if my coming back to these issues is annoying. I realise that 
there are the same problems with the algorithm I am using at the moment, but 
you've spurred me on to try and get the best solution to this issue.

wassalaam
abdulhaq



On Friday 04 June 2004 20:37, Thamer Mahmoud wrote:
> Salaam,
>
> Actually, implementing a more standard (i.e. a simplified Sun-only
> VSOP87) is one of the two significant changes that will happen before
> 0.6 release that I've talked about in another post. The other change
> is not as significant though ;)
>
> I would say that I'm 80% done so far. What's left is that I need to
> implement a few remaingin formulas and a more accurate atmospheric
> refraction code (so that we take advantage of pressure, temperature,
> among other things). After that I'm going to add a new structure to
> hold the data as well as convert the old Get[Salatname] functions to
> use the new structure (and formulas).
>
> Finally, If all went well with testing, I will be checking in the new
> code. Just a note though, the new code will be, Inshallah, more
> accurate than the formulas used currently on ITL-CVS, but please do
> not expect that we will actually get the same exact results as
> libnova, marsbase, or any of the other VSOP87 implementers. They both
> claim to be using VSOP87 but their results differ by minutes as shown
> here.
>
>
> http://lists.arabeyes.org/archives/developer/2004/May/msg00000.html
>
>
> By the way, Libnova does not use the full VSOP87, hence the small
> size, but they use well known simplified formulas. These are the same
> formulas that I'm currently implementing.
>
>
> Regards,
> Thamer Mahmoud
>
> _______________________________________________
> Developer mailing list
> Developer at arabeyes dot org
> http://lists.arabeyes.org/mailman/listinfo/developer