MathGroup Archive 2007

[Date Index] [Thread Index] [Author Index]

Search the Archive

Approximate/asymptotic factorization

  • To: mathgroup at
  • Subject: [mg73509] Approximate/asymptotic factorization
  • From: Paul Abbott <paul at>
  • Date: Wed, 21 Feb 2007 01:35:21 -0500 (EST)
  • Organization: The University of Western Australia

Consider the monomial H[4,4]

  256*(23625 + 126000*z^4 - 7200*z^8 + 3840*z^12 + 256*z^16)

where H[m,n] is a Generalized Hermite Polynomial (Noumi & Yamada 1998), 
a rational solution to the Painlevé equation P_IV. A plot of the roots 
of H[4,4] displays an interesting quasi-rectangular structure.

Then consider the monomial Q[4,4]

  1528899609315374375625 + 6795109374734997225000*z^4 -  
  560866170613047390000*z^8 + 153399294526645440000*z^12 +  
  2734590598399296000*z^16 - 167891551278796800*z^20 +  
  2948686820352000*z^24 - 40649991782400*z^28 +  277762867200*z^32 - 
  920125440*z^36 + 1048576*z^40

where Q[m,n] is a Generalized Okamoto Polynomial, another rational 
solution to P_IV. A plot of the roots displays an even more interesting 
structure, with a quasi-rectangular core and 4 triangular "lobes" 
(related, in some way, to the the Yablonskii­Vorob¹ev Polynomials, Q[n], 
which are Rational Solutions of P_II)

If you overlay plots of the roots of H[4,4] and Q[4,4] what is most 
striking is that the "common" roots are very similar. In other words, 
H[4,4] is "approximately" a factor of Q[4,4]. 

Is there a "standard" way of quantifying this approximate factorization?

This approximate factorization holds for any m and n and appears to 
improve as m and n increases. Perhaps there is a limit in which the 
factorization holds asymptotically?


Paul Abbott                                      Phone:  61 8 6488 2734
School of Physics, M013                            Fax: +61 8 6488 1014
The University of Western Australia         (CRICOS Provider No 00126G)    

  • Prev by Date: Re: FullGraphics question
  • Next by Date: Re: Precision of a number raised to a power
  • Previous by thread: LogPlot and Epilog
  • Next by thread: Re: Approximate/asymptotic factorization