st: Mata --- polyroots() accuracy

[Adrian Mander <adrian.mander@mrc-bsu.cam.ac.uk>]

I am trying to write a function with as much precision as Stata will allow.

Part of the function requires finding the roots of a polynomial and I 
realise that I am dealing with some very small numbers

coef12 =(1,-12,33,-110/3,165/8,-33/5,77/60,-11/70,11/896,- 
11/18144,11/604800,-1/3326400,1/479001600)
polyroots( coef12 )

Stata finds one of the roots as 37.09912104447676739

 polyeval( coef12, 37.09912104447676739)

But my function is better 37.09912104446606662 gives a value that is a 
little closer to 0 (BUT still not great!)
 polyeval( coef12, 37.09912104446606662)

I wonder if Stata Corps thinks it worth doing polyroots to quadprecision?

Thanks
Adrian

--
Dr Adrian Mander
Leader of Cambridge Hub in Trials Methodology Research
MRC Biostatistics Unit    Institute of Public Health 
University Forvie Site    Cambridge CB2 0SR

Tel: 01223 760728       Fax: 01223 760727

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/