(PC-12xx~14xx) Laguerre method
|
08-21-2024, 05:51 PM
Post: #3
|
|||
|
|||
RE: (PC-12xx~14xx) Laguerre method
After the first derivative, factorial constants should come in:
p'(z) = p1 p"(z) = 2 * p2 When these constants are taken into account, the number of iterations to converge will be significantly reduced. In the implementations which use separate real and imaginary variables, vr and vi (or V and Y in line 310) are not updated when hi (or K in line 270) equals zero and so the previous values are used. A better way to compute the real part of G^2 is ur = (gr-gi) * (gr+gi) (or U = (F-G) * (F+G) in line 260). The maximal denominator magnitude can be simply computed as: if gr*vr < -gi*vi {ur = gr-vr; ui = gi-vi;} else {ur = gr+vr; ui = gi+vi;} (or in line 310 IF F*V<-G*Y LET U=F-V,X=G-Y: GOTO 320 newline 315 U=F+V,X=G+Y). |
|||
« Next Oldest | Next Newest »
|
Messages In This Thread |
(PC-12xx~14xx) Laguerre method - robve - 03-11-2022, 01:48 AM
RE: (PC-12xx~14xx) Laguerre method - robve - 03-11-2022, 02:11 AM
RE: (PC-12xx~14xx) Laguerre method - Quattro - 08-21-2024 05:51 PM
|
User(s) browsing this thread: 2 Guest(s)