Newton-Hines fractals

  • 38 Replies
  • 1975 Views

0 Members and 1 Guest are viewing this topic.

Offline superheal

  • *
  • Fractal Fruit Salad
  • *****
  • Posts: 64
« Reply #30 on: May 13, 2019, 11:34:42 AM »

Offline gerrit

  • *
  • 3f
  • ******
  • Posts: 1973
« Reply #31 on: May 15, 2019, 08:53:12 AM »
Pure Newton applied to f(c) = "11 iterations of z^2+c". Looks like external rays, wonder why or maybe it's just a coincidence.

Offline claude

  • *
  • 3f
  • ******
  • Posts: 1369
    • mathr.co.uk
« Reply #32 on: May 15, 2019, 09:15:31 AM »
the basins are ray-like (approach infinity in a straight line)
I suspect this is true for any polynomial, M-set-related or not

but I don't think the angles match what you would hope for
If the basins for a root coincided exactly with corresponding external rays, you wouldn't need to trace external rays step by step
you could just use Newton's method starting from a sufficiently large circle at a sufficiently accurate angle

Offline gerrit

  • *
  • 3f
  • ******
  • Posts: 1973
« Reply #33 on: May 16, 2019, 08:16:35 PM »
Halley method update is \( -1/(\frac{f'}{f}- q \frac{f''}{2f'}     ) \) with q=1, and q=0 gives you Newton method back.
Image is based on a (4,3) rational function with q=0.9.
Solid red = "escaping" = "diverging". The divergent regions seem to be filled with some Cantor dust of converging points.

Offline gerrit

  • *
  • 3f
  • ******
  • Posts: 1973
« Reply #34 on: May 18, 2019, 07:32:08 AM »
Colorful Halley as above with "wrong sign" q=-1, applied to 11th degree polynomial.

Offline marcm200

  • *
  • Fractal Frankfurter
  • *
  • Posts: 620
« Reply #35 on: July 15, 2019, 10:18:26 AM »
I stumbled upon a very interesting paper (I searched the forum for this but couldn't find a reference to it):

Quote
How to find all roots of complex polynomials by Newton’s method
John Hubbard, Dierk Schleicher, Scott Sutherland
Digital Object Identifier
Invent. math. 146, 1–33 (2001)

The result being, that there is a constructible set of finite many complex points so that when starting Newton's method from those points all roots of a polynomial will be found - and that set is only dependent on the degree of the polynomial and not the polynomial itself. So instead of hoping to have found a good enough starting point that will converge to a root (my approach), one could use a precomputed list and scale those accordingly (they work with normalized polynomials, where all the roots lie in the unit disk).

I find those general and constructible results very prize-worthy.


Offline gerrit

  • *
  • 3f
  • ******
  • Posts: 1973
« Reply #36 on: July 15, 2019, 07:52:45 PM »
Thanks, not sure if it's comprehensible but I'll try to read it: https://pi.math.cornell.edu/~hubbard/NewtonInventiones.pdf

Offline gerrit

  • *
  • 3f
  • ******
  • Posts: 1973
« Reply #37 on: July 23, 2019, 04:33:34 AM »
Secant method for rational function with 791 zeros and 791 poles.

Offline gerrit

  • *
  • 3f
  • ******
  • Posts: 1973
« Reply #38 on: October 01, 2019, 08:28:25 PM »
Adding a tiny Hines factor of k=1E-4 to https://fractalforums.org/image-threads/25/gerrit-images/565/msg17007#msg17007 shatters the green non convergent areas into something else. Image is a zoom on biggest green blob from the referenced post.


xx
Slightly extended Newton method fractals

Started by gannjondal on Fractal Mathematics And New Theories

6 Replies
330 Views
Last post September 29, 2019, 10:18:01 PM
by gerrit
exclamation
Newton Fractal but different

Started by chrizbee on Downloads

15 Replies
792 Views
Last post August 29, 2019, 06:43:53 PM
by chrizbee
xx
Newton's Garden

Started by gannjondal on Fractal Image Gallery

4 Replies
406 Views
Last post April 17, 2018, 11:22:45 PM
by spain2points
clip
Mandelbrot/Newton - Has it already been done?

Started by mrrudewords on Share a fractal

2 Replies
453 Views
Last post February 09, 2019, 09:35:17 PM
by mrrudewords
clip
Revisiting the 3D Newton

Started by gannjondal on Fractal Mathematics And New Theories

26 Replies
2310 Views
Last post October 01, 2018, 10:24:43 PM
by FractalDave