ENG  RUSTimus Online Judge
Online Judge
Online contests
About Online Judge
Frequently asked questions
Site news
Problem set
Submit solution
Judge status
Update your info
Authors ranklist
Current contest
Scheduled contests
Past contests
back to board

Discussion of Problem 1503. Polynomial

Posted by svr 31 Oct 2006 00:22
Number of solutions depend on untold rounding rules.
For exaple, let our algorithm gave comlex root
1+0.0000001*i. What descision, include 1 to real roots or not. For me right answer may vary in root numbers but contain in it's 0.000001 - neighbornhood ideal mathamatical set of roots.
Re: Question
Posted by svr 31 Oct 2006 14:07
It seems that with type double and 16 righs digits in results the problem impossible to solve. For examle if we try compair solutions x1=0.999988 and x2=0.999982 for equation f(x)=(x-1)^4=x^4-4*x^3+6*x^2-4*x+1=0 computer
says that x2 is better. If we have 30-digits floats we would correct answer by optimization near initial approximation. If somebody knows internet resources with such module I shall be grateful.