[time-nuts] Re: Perl program for stability analysis available

John Ackermann N8UR jra at febo.com
Sun Jan 2 20:34:03 EST 2005


After a fairly massive hacking session, I now have the "stable-stats.pl 
program I mentioned the other day whipped into shape to do what I 
wanted, which is to automagically turn a file of phase data into a web 
page with statistics and plots.  Here's an example of what it will 
generate from a single command line invocation:  
http://www.febo.com/time-freq/tools/test.html.  It does require that you 
have the "Grace" plotting program installed. 

Though this program was written under and for Linux, the perl code 
itself should work under any OS that has perl available.  However, I'm 
not sure if the Win32 port of "Grace" will work with this application.

The program is available from http://www.febo.com/time-freq/tools/.  As 
noted below, I'm sure it still has lots of bad code that needs to be 
improved -- I spent my time this weekend concentrating on functionality, 
and the next version will concentrate on de-crapifying the code.

John

---------------
John Ackermann N8UR wrote:

> Hi --
>
> I've put a very preliminary alpha version of a perl program called 
> "stable-stats.pl" up on my website.  You can download it from 
> http://www.febo.com/time-freq/tools/index.html.  It should run on any 
> system capable of running perl, but it does require one external 
> statistics module; I have a link to that module on the web page.
>
> stable-stats.pl will read (from standard input) a file containing 
> MJD/phase pairs, one pair per line with values separated by spaces or 
> tabs.  It will output the drift, the frequency offset based on both 
> linear regression and endpoints, and Allen Variance for a reasonable 
> set of tau values.
>
> I'll repeat that this is an alpha program with little error checking 
> or flexibility.  It also truncates where it ought to round.  I hope to 
> fancy it up once I'm sure the basic algorithms are working properly.*  
> Based on a limited set of test data, its results seem to reasonably 
> match those of Stable32 (though it doesn't do nearly as much as that 
> program).
>
> Please feel free to try this out, and particularly to scrutinize the 
> code for any errors -- there are plenty of opportunities for 
> off-by-one and other indexing errors.  I welcome problem reports, 
> suggestions, and bug fixes.
>
> Happy New Year!
>
> John
>
> * On the to-do list are the automatic generation of phase and AVAR 
> plots, as well as general robustness and flexibility improvements.
>





More information about the time-nuts mailing list