home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
ftp.xmission.com
/
2014.06.ftp.xmission.com.tar
/
ftp.xmission.com
/
pub
/
lists
/
fractint
/
archive
/
v01.n570
< prev
next >
Wrap
Internet Message Format
|
2001-06-08
|
61KB
From: owner-fractint-digest@lists.xmission.com (fractint-digest)
To: fractint-digest@lists.xmission.com
Subject: fractint-digest V1 #570
Reply-To: fractint-digest
Sender: owner-fractint-digest@lists.xmission.com
Errors-To: owner-fractint-digest@lists.xmission.com
Precedence: bulk
fractint-digest Friday, June 8 2001 Volume 01 : Number 570
----------------------------------------------------------------------
Date: Fri, 08 Jun 2001 07:25:49 +0000
From: "Thierry B." <oulala@chez.com>
Subject: Re: (fractint) Some intriguing stuff.
> These formulas make it possible to investigate the escape behavior of t=
he Henon map:
=20
> x -> a - by + x^2
> y -> x
I've also a few research on the mappin of H=E9non diagram.
http://la.buvette.org/fractales/map_henon.euh
Sorry, this is only a Fortran source, but I can write some
explanation in english this weekend.
=20
- --=20
Thierry, 42++
- --------------------------------------------------------------
Thanks for using Fractint, The Fractals and Fractint Discussion List
Post Message: fractint@lists.xmission.com
Get Commands: majordomo@lists.xmission.com "help"
Administrator: twegner@fractint.org
Unsubscribe: majordomo@lists.xmission.com "unsubscribe fractint"
------------------------------
Date: Fri, 08 Jun 2001 09:57:44 +0200
From: Guy Marson <guy.marson@mnhn.lu>
Subject: Re: (fractint) Some intriguing stuff.
At 07:25 08/06/01 +0000, you wrote:
>> These formulas make it possible to investigate the escape behavior of
the Henon map:
>=20
>> x -> a - by + x^2
>> y -> x
>
> I've also a few research on the mappin of H=E9non diagram.
> http://la.buvette.org/fractales/map_henon.euh
> Sorry, this is only a Fortran source, but I can write some
> explanation in english this weekend.
mais pas dans la buvette, avec des Jupiler s.v.p. (hickkk..)=20
>
>=20
>--=20
>Thierry, 42++
>
cheers,=20
guy 47+
- --------------------------------------------------------------
Thanks for using Fractint, The Fractals and Fractint Discussion List
Post Message: fractint@lists.xmission.com
Get Commands: majordomo@lists.xmission.com "help"
Administrator: twegner@fractint.org
Unsubscribe: majordomo@lists.xmission.com "unsubscribe fractint"
------------------------------
Date: Fri, 8 Jun 2001 09:15:51 EDT
From: JimMuth@aol.com
Subject: (fractint) C-FOTD 08-06-01 (Too Much Fractal [7])
Classic FOTD -- June 08, 2001 (Rating 7)
Fractal visionaries and enthusiasts:
It seems strange to denigrate a fractal with a rating two points
above average, but that's what I feel I must do with today's
image.
Sometimes, even in the world of fractals, it's possible to have
too much of a good thing. Today's image is an example of such
fractal excess. The image simply goes too far with too little.
The midget at the center is too small to act as a center of
attention, leaving the surrounding decorations with nothing to
decorate. And the color is excessive. Vibrant color can be
spectacular when it is done properly. When it's not done right,
the result is boring gaudiness. Today's image just doesn't give
me that "this color is right" feeling.
Oh, the scene has a surface glitter that brings its rating up to
a 7, but it lacks the depth that could bring a rating of 8 or 9.
I named the image "Too Much Fractal" in response to my feelings
about it.
Today's scene lies in another valley, directly across the bay
from yesterday's, though it is some distance back from the
shoreline, and at a considerable greater depth. It is actually
located at the center of a figure-8 ring, which the area is
filled with.
The 4-1/2 minute render time is marginal, making it the viewers
choice whether to go online and download the GIF image from:
<http://home.att.net/~Paul.N.Lee/FotD/FotD.html>
or from:
<http://sdboyd.dyndns.org/~sdboyd/fotd/index.html>
The fractal weather today started with rain, but the rain ended
in mid-morning, and the sun returned in mid-afternoon, sending
the temperature up to 75F (24C). The fractal cats celebrated by
venturing cautiously into the still-wet grass.
It's now time to get busy on other things, so until next time,
take care and check the fractal on the cover of the latest issue
of "Skeptic" magazine.
Jim Muth
jamth@mindspring.com
START 20.0 PAR-FORMULA FILE================================
Too_Much_Fractal { ; time=0:04:25.66--SF5 on a P200
reset=2001 type=formula formulafile=allinone.frm
formulaname=MandelbrotMix4 function=recip passes=b
center-mag=+10.09705566264443000/-15.5438271495980\
4000/1.007928e+011/1/166.953/-0.502
params=-0.6/-1.3/-0.006/-3/0/1000 float=y
maxiter=1500 inside=255 logmap=133 periodicity=10
colors=0800DL0DL0DO0DR0DU0DX0D_0Db0De0Dh0Dk0Dn0Dq6\
FtGGwPDzZIzhQzrXzzdzzlzztzzwzwzwYjlnZrzQyzGzZIf0IN\
0QW0V`0bf0hl0pt0vz0wz1wz1wz8wzGwzNrzWjzbbzjVzrOzzG\
zzDzzDzzDzzDzzDzzDvzDnzDfyD`tDUpGLlIGhM8dO1`R0YT0W\
V0ZX0bX0dX0hX0jX0nX1rX4tX8yXAzXEzXHzXJzwJzwPywUvwY\
tw`vwYywWywRzwPzwLzwJzwGzwEzwCzwGzwHztJzpNzjPzdRz`\
WzVYzQZzM`zOdzQfzRjzTlzVnzXrzZtz`yzbzzbzzfzzhzzlzz\
nzzrzztzzwzzwzlnzWbnETZ0IL0D80F40I14M0EQ0LT0UX0bb0\
jf0rj0zn0zr0zv0zt0zr0zp0zp0z`0jO0PD06D0ED0JD0RD2YG\
6dMAjQErVHyZLzbNvfPphPhlRbnRYrRPtUJwUEwU6wW1wW0wW0\
rN0bG0O80T00Z00d00j00n00r00v06w0Cw0Jw0Rw0Yw0dw0lw0\
rw0fw4YwUNwrEwzCwzCwzCvzCtzArzApzAnzAlzAlzHvzNwzWw\
z`wzfwzbwz`wzZwzWnzUdzRVzYRzbOzfKzlGzpDtvDpzDjzDdz\
DZzDUzDNzDJzDLzDNrIPfTRWbULlWAvY0wZ0w`0wb0wb0wW2wN\
JfH`OApD4rD4tD6vD6CwfGwdHwbLwZNwYRwWUwUZwWdwYhwZnw\
`rwbywdzwflwnYwtHwz2wz0D0
}
frm:MandelbrotMix4 {; Jim Muth
a=real(p1), b=imag(p1), d=real(p2), f=imag(p2),
g=1/f, h=1/d, j=1/(f-b), z=(-a*b*g*h)^j,
k=real(p3)+1, l=imag(p3)+100, c=fn1(pixel):
z=k*((a*(z^b))+(d*(z^f)))+c,
|z| < l
}
END 20.0 PAR-FORMULA FILE==================================
- --------------------------------------------------------------
Thanks for using Fractint, The Fractals and Fractint Discussion List
Post Message: fractint@lists.xmission.com
Get Commands: majordomo@lists.xmission.com "help"
Administrator: twegner@fractint.org
Unsubscribe: majordomo@lists.xmission.com "unsubscribe fractint"
------------------------------
Date: Fri, 8 Jun 2001 21:29:56 -0400
From: "Multiple Bogeys" <neo_1061@hotmail.com>
Subject: (fractint) Hairy Newton
- ------=_NextPart_001_0000_01C0F062.2AD5F4A0
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Yesterday evening I set out to find an interesting family of Newton-based=
Mandelbrot mappings. The result was the collection of formulae below.
You'll note the lack of 3D disease -- false alarm. All my posts to anothe=
r listserv had it, but I now think it must be that listserv rather than M=
SN Exploder. Or rather, some interaction between the two (since some post=
s from that listserv don't have it), like the weird "email laser" that ha=
ppened with this listserv last week (lots of peoples' messages were being=
duplicated, but for some reason mine were showing up in sets of five or =
six!)...Anyone who's an expert on listservs care to speculate further on =
what might be going on? One thing is clear: recent mail software and list=
servs have unnecessary complexity, and while we like complexity in our fr=
actals, we could do without chaos erupting in the mail system we depend o=
n to communicate here...
The hrynewt_j and hrynewt_m formulae iterate Newton's method for p(z) =3D=
(z^n - 1)(z^3 - az - 1). Both n and a are parameters, as is the toleranc=
e, an inverse bailout radius about the roots of p. The Mandelbrot variant=
has a vary over the screen while initial z is zero; this is a critical p=
oint but not a root of p for n real and greater than 2. You can plug in o=
ther values of n -- arbitrary negative or even complex values -- but won'=
t generally be able to find minibrots unless n has a positive real part g=
reater than two. If n is not an integer, there will be branch cuts in bot=
h the Mandelbrot and the Julia variants.
The hrynewtnnn_j and _m formulae are optimized versions with specific val=
ues for n, mostly small positive integers. They avoid a slow arbitrary ex=
ponentiation, and for the smaller values of n re-use powers that are used=
on both sides of the polynomial or its derivative. The hrynewt2_m formul=
a also has the feature of using a critical point for initial z, instead o=
f zero (which is *not* a critical point for n =3D 2). The result is a pro=
per Mandelbrot view, but it has a branch cut due to a square root in the =
calculation of the critical point, which is a-dependent. The branch cut h=
as been intentionally manipulated to put it in a fairly unobtrusive place=
, but can't be eliminated; the full Mandelbrot for this one lives on a tw=
o-layer Riemann sheet like that of the square root function.
The hrynewtnnn_m formulae also use an (XAXIS) symmetry declaration. (The =
generic hrynewt_m can't use this without trashing the output for non-real=
values of n.)
Observations:
* Certain choices of n produce three-fold-symmetric Mandelbrot sets. Find=
out which!
* Mangled and occasionally also intact Mandelbrots can be extracted when =
n is "strange" but has a real part greater than 2.
* You get radial petals with n real, concentric patterns with n imaginary=
, and logarithmic spirals with complex n; the ratio of
real to imaginary parts determines whether the spiral is steep (n clos=
e to real) or shallow (n close to imaginary).
* The Mandelbrots are always quadratic -- for real n > 2, the critical po=
int at zero is nondegenerate, and the critical point
pair for hrynewt2_m is degenerate only at one specific value of a.
The formula file begins with an extensive comment that details the mathem=
atical constructions that informed their design.
comment {
We want a Newton's method with a large number of basins, most of which =
are fixed and predictable.
This is accomplished by choosing a polynomial function to solve compose=
d of two factors, one with many fixed roots,
the other with a few mobile ones:
p(z) =3D (z^n-1)(z^3-az-1).
The Newton iteration is:
z -> r(z)
where
r(z) =3D z - p(z)/p'(z)
=3D (zp'(z) - p(z))/p'(z)
We easily discover p'(z) to be
p'(z) =3D (z^n-1)(3z^2-a) + (nz^(n-1))(z^3-az-1)
=3D (3+n)z^(n+2) - a(n+1)z^n - nz^(n-1) - 3z^2 + a
so
(zp'(z) - p(z)) =3D (z^n-1)(3z^3-az)+(nz^n)(z^3-az-1) - (z^n-1)(z^3-az-=
1)
=3D (z^n-1)(2z^3+1)+(nz^n)(z^3-az-1)
=3D (2+n)z^(n+3) - anz^(n+1) - (n-1)z^n - 2z^3 - 1
and
r(z) =3D ((2+n)z^(n+3) - anz^(n+1) - (n-1)z^n - 2z^3 - 1)/((3+n)z^(n+2)=
- a(n+1)z^n - nz^(n-1) - 3z^2 + a)
Using the quotient rule the numerator of r'(z) is
((3+n)z^(n+2) - a(n+1)z^n - nz^(n-1) - 3z^2 + a)((n+2)(n+3)z^(n+2) - an=
(n+1)z^n - n(n-1)z^(n-1) - 6z^2) -
((2+n)z^(n+3) - anz^(n+1) - (n-1)z^n - 2z^3 - 1)((n+2)(n+3)z^(n+1) - =
an(n+1)z^(n-1) - n(n-1)z^(n-2) - 6z)
which factors into
((n+2)(n+3)z^(n+1) - an(n+1)z^(n-1) - n(n-1)z^(n-2) - 6z)
and
((3+n)z^(n+3) - a(n+1)z^(n+1) - nz^n - 3z^3 + az) - ((2+n)z^(n+3) - anz=
^(n+1) - (n-1)z^n - 2z^3 - 1)
which simplifies to
z^(n+3) - az^(n+1) - z^n - z^3 + az + 1
Note that p(z) =3D z^(n+3) - az^(n+1) - z^n - z^3 + az + 1.
Thus the critical points of r(z) are the roots of p(z) and the roots of
q(z) :=3D ((n+2)(n+3)z^(n+1) - an(n+1)z^(n-1) - n(n-1)z^(n-2) - 6z)
These latter are the "interesting" critical points, as the other critic=
al points of r(z) are all superattracting.
Note that q(z) is divisible by z, so 0 is an "interesting" critical poi=
nt of r(z), for n not one of 2, 1, or -1.
This is the critical point used in the below hrynewt_m formulas except =
for hrynewt2_m. For n =3D 2,
q(z)/2 =3D 10z^3 - 3(a+1)z - 1
Put z =3D y + (a+1)/10y to get
q(z)/2 =3D y^6 - y^3/10 + (a+1)^3/1000
so
2y^3 =3D 1/10 +/-sqrt(1/100 - 4(a+1)^3/1000),
y =3D ((1/10 +/-sqrt(1/100 - 4(a+1)^3/1000))/2)^(1/3)
and
z =3D ((1/10 +/-sqrt(1/100 - 4(a+1)^3/1000))/2)^(1/3) + (a+1)/((1/10 +/=
- -sqrt(1/100 - 4(a+1)^3)/1000)/2)^(1/3)
}
hrynewt_j { ; p1 is Julia parameter, p2 is exponent n, p3 is tolerance (i=
f 0, will act like 0.001).
; SLOW. Use predefined hrynewtnnn_j where possible.
z =3D pixel, a =3D p1, n =3D p2, n1 =3D n - 1, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zn1 =3D z^n1
zno =3D (z*zn1 - 1)
zzz =3D (z3 - a*z - 1)
pz =3D zno*zzz
ppz =3D zno*(3*z2 - a) + n*zn1*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt_m { ; p2 is exponent n, p3 is tolerance (if 0, will act like 0.00=
1).
; SLOW. Use predefined hrynewtnnn_m where possible.
z =3D 0, a =3D pixel, n =3D p2, n1 =3D n - 1, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zn1 =3D z^n1
zn =3D z*zn1
zno =3D (zn - 1)
zzz =3D (z3 - a*z - 1)
pz =3D zno*zzz
ppz =3D zno*(3*z2 - a) + n*zn1*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt2_j { ; p1 is Julia parameter, p3 is tolerance (if 0, will act lik=
e 0.001).
; n =3D 2.
z =3D pixel, a =3D p1, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zno =3D (z2 - 1)
zzz =3D (z3 - a*z - 1)
pz =3D zno*zzz
ppz =3D zno*(3*z2 - a) + 2*z*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt2_m (XAXIS) { ; p3 is tolerance (if 0, will act like 0.001).
; n =3D 2.
a =3D pixel,
ap1 =3D a + 1,
IF((real(ap1) >=3D 0) || ((abs(real(ap1))*(3^(0.5))) < abs(imag(ap1))))
t =3D ((0.1 + (0.01 - 0.004*sqr(ap1)*ap1)^(0.5))/2)^(1/3),
ELSE
t =3D ((0.1 - (0.01 - 0.004*sqr(ap1)*ap1)^(0.5))/2)^(1/3),
ENDIF
z =3D t + 0.1*ap1/t, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zno =3D (z2 - 1)
zzz =3D (z3 - a*z - 1)
pz =3D zno*zzz
ppz =3D zno*(3*z2 - a) + 2*z*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt3_j { ; p1 is Julia parameter, p3 is tolerance (if 0, will act lik=
e 0.001).
; n =3D 3.
z =3D pixel, a =3D p1, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zno =3D (z3 - 1)
zzz =3D zno - a*z
tz2 =3D 3*z2
pz =3D zno*zzz
ppz =3D zno*(tz2 - a) + tz2*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt3_m (XAXIS) { ; p3 is tolerance (if 0, will act like 0.001).
; n =3D 3.
z =3D 0, a =3D pixel, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zno =3D (z3 - 1)
zzz =3D zno - a*z
tz2 =3D 3*z2
pz =3D zno*zzz
ppz =3D zno*(tz2 - a) + tz2*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt4_j { ; p1 is Julia parameter, p3 is tolerance (if 0, will act lik=
e 0.001).
; n =3D 4.
z =3D pixel, a =3D p1, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zno =3D (z*z3 - 1)
zzz =3D (z3 - a*z - 1)
pz =3D zno*zzz
ppz =3D zno*(3*z2 - a) + 4*z3*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt4_m (XAXIS) { ; p3 is tolerance (if 0, will act like 0.001).
; n =3D 4.
z =3D 0, a =3D pixel, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zno =3D (z*z3 - 1)
zzz =3D (z3 - a*z - 1)
pz =3D zno*zzz
ppz =3D zno*(3*z2 - a) + 4*z3*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt5_j { ; p1 is Julia parameter, p3 is tolerance (if 0, will act lik=
e 0.001).
; n =3D 5.
z =3D pixel, a =3D p1, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zn1 =3D sqr(z2)
zno =3D (z*zn1 - 1)
zzz =3D (z3 - a*z - 1)
pz =3D zno*zzz
ppz =3D zno*(3*z2 - a) + 5*zn1*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt5_m (XAXIS) { ; p3 is tolerance (if 0, will act like 0.001).
; n =3D 5.
z =3D 0, a =3D pixel, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zn1 =3D sqr(z2)
zno =3D (z*zn1 - 1)
zzz =3D (z3 - a*z - 1)
pz =3D zno*zzz
ppz =3D zno*(3*z2 - a) + 5*zn1*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt17_j { ; p1 is Julia parameter, p3 is tolerance (if 0, will act li=
ke 0.001).
; n =3D 17.
z =3D pixel, a =3D p1, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zn1 =3D sqr(sqr(sqr(z2)))
zno =3D (z*zn1 - 1)
zzz =3D (z3 - a*z - 1)
pz =3D zno*zzz
ppz =3D zno*(3*z2 - a) + 17*zn1*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt17_m (XAXIS) { ; p3 is tolerance (if 0, will act like 0.001).
; n =3D 17.
z =3D 0, a =3D pixel, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zn1 =3D sqr(sqr(sqr(z2)))
zno =3D (z*zn1 - 1)
zzz =3D (z3 - a*z - 1)
pz =3D zno*zzz
ppz =3D zno*(3*z2 - a) + 17*zn1*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt33_j { ; p1 is Julia parameter, p3 is tolerance (if 0, will act li=
ke 0.001).
; n =3D 33.
z =3D pixel, a =3D p1, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zn1 =3D sqr(sqr(sqr(sqr(z2))))
zno =3D (z*zn1 - 1)
zzz =3D (z3 - a*z - 1)
pz =3D zno*zzz
ppz =3D zno*(3*z2 - a) + 33*zn1*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt33_m (XAXIS) { ; p3 is tolerance (if 0, will act like 0.001).
; n =3D 33.
z =3D 0, a =3D pixel, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zn1 =3D sqr(sqr(sqr(sqr(z2))))
zno =3D (z*zn1 - 1)
zzz =3D (z3 - a*z - 1)
pz =3D zno*zzz
ppz =3D zno*(3*z2 - a) + 33*zn1*zzz
z =3D z - pz/ppz,
|pz| > r
}<br clear=3Dall><hr>Get Your Private, Free E-mail from MSN Hotmail at <a=
href=3D"http://www.hotmail.com">http://www.hotmail.com</a>.<br></p>
- ------=_NextPart_001_0000_01C0F062.2AD5F4A0
Content-Type: text/html; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
<HTML><BODY STYLE=3D"font:10pt verdana; border:none;"><DIV>Yesterday even=
ing I set out to find an interesting family of Newton-based Mandelbrot ma=
ppings. The result was the collection of formulae below.</DIV> <DIV> =
;</DIV> <DIV>You'll note the lack of 3D disease -- false alarm. All my po=
sts to another listserv had it, but I now think it must be that listserv =
rather than MSN Exploder. Or rather, some interaction between the two (si=
nce some posts from that listserv don't have it), like the weird "em=
ail laser" that happened with this listserv last week (lots of peoples' m=
essages were being duplicated, but for some reason mine were showing up i=
n sets of five or six!)...Anyone who's an expert on listservs care to spe=
culate further on what might be going on? One thing is clear: recent mail=
software and listservs have unnecessary complexity, and while we like co=
mplexity in our fractals, we could do without chaos erupting in the mail =
system we depend on to communicate here...</DIV> <DIV> </DIV> <DIV>T=
he hrynewt_j and hrynewt_m formulae iterate Newton's method for p(z) =3D =
(z^n - 1)(z^3 - az - 1). Both n and a are parameters, as is the tolerance=
, an inverse bailout radius about the roots of p. The Mandelbrot variant =
has a vary over the screen while initial z is zero; this is a critical po=
int but not a root of p for n real and greater than 2. You can =
plug in other values of n -- arbitrary negative or even complex values --=
but won't generally be able to find minibrots unless n has a positive re=
al part greater than two. If n is not an integer, there will be branch cu=
ts in both the Mandelbrot and the Julia variants.</DIV> <DIV> </DIV>=
<DIV>The hrynewtnnn_j and _m formulae are optimized versions with specif=
ic values for n, mostly small positive integers. They avoid a slow arbitr=
ary exponentiation, and for the smaller values of n re-use powers that ar=
e used on both sides of the polynomial or its derivative. The hrynewt2_m =
formula also has the feature of using a critical point for initial z, ins=
tead of zero (which is *not* a critical point for n =3D 2). The result is=
a proper Mandelbrot view, but it has a branch cut due to a square root i=
n the calculation of the critical point, which is a-dependent. The branch=
cut has been intentionally manipulated to put it in a fairly unobtrusive=
place, but can't be eliminated; the full Mandelbrot for this one lives o=
n a two-layer Riemann sheet like that of the square root function.</DIV> =
<DIV> </DIV> <DIV>The hrynewtnnn_m formulae also use an (XAXIS) symm=
etry declaration. (The generic hrynewt_m can't use this without trashing =
the output for non-real values of n.)</DIV> <DIV> </DIV> <DIV>Observ=
ations:</DIV> <DIV>* Certain choices of n produce three-fold-symmetric Ma=
ndelbrot sets. Find out which!</DIV> <DIV>* Mangled and occasionally also=
intact Mandelbrots can be extracted when n is "strange" but has a real p=
art greater than 2.</DIV> <DIV>* You get radial petals with n real, conce=
ntric patterns with n imaginary, and logarithmic spirals with complex n; =
the ratio of</DIV> <DIV> real to imaginary parts determines w=
hether the spiral is steep (n close to real) or shallow (n close to imagi=
nary).</DIV> <DIV>* The Mandelbrots are always quadratic -- for real n &g=
t; 2, the critical point at zero is nondegenerate, and the critical point=
</DIV> <DIV> pair for hrynewt2_m is degenerate only at o=
ne specific value of a.</DIV> <DIV> </DIV> <DIV>The formula file beg=
ins with an extensive comment that details the mathematical constructions=
that informed their design.</DIV> <DIV> </DIV> <DIV>comment {<BR>&n=
bsp; We want a Newton's method with a large number of basins, most of whi=
ch are fixed and predictable.<BR> This is accomplished by choosing =
a polynomial function to solve composed of two factors, one with many fix=
ed roots,<BR> the other with a few mobile ones:<BR> p(z) =3D =
(z^n-1)(z^3-az-1).<BR> The Newton iteration is:<BR> z -> r=
(z)<BR> where<BR> r(z) =3D z - p(z)/p'(z)<BR> &nbs=
p; =3D (zp'(z) - p(z))/p'(z)<BR> We easily discov=
er p'(z) to be<BR> p'(z) =3D (z^n-1)(3z^2-a) + (nz^(n-1))(z^3-az-1)=
<BR> =3D (3+n)z^(n+2) - a(n+1)z=
^n - nz^(n-1) - 3z^2 + a<BR> so<BR> (zp'(z) - p(z)) =3D (z^n-=
1)(3z^3-az)+(nz^n)(z^3-az-1) - (z^n-1)(z^3-az-1)<BR> &nb=
sp; &nbs=
p; =3D (z^n-1)(2z^3+1)+(nz^n)(z^3-az-1)<BR> =
&=
nbsp; =3D (2+n)z^(n+3) - anz^(n+1) - (n-1)z^n - 2z^3 - 1<BR> and<BR=
> r(z) =3D ((2+n)z^(n+3) - anz^(n+1) - (n-1)z^n - 2z^3 - 1)/((3+n)z=
^(n+2) - a(n+1)z^n - nz^(n-1) - 3z^2 + a)<BR> Using the quotient ru=
le the numerator of r'(z) is<BR> ((3+n)z^(n+2) - a(n+1)z^n - nz^(n-=
1) - 3z^2 + a)((n+2)(n+3)z^(n+2) - an(n+1)z^n - n(n-1)z^(n-1) - 6z^2) -<B=
R> ((2+n)z^(n+3) - anz^(n+1) - (n-1)z^n - 2z^3 - 1)((n+=
2)(n+3)z^(n+1) - an(n+1)z^(n-1) - n(n-1)z^(n-2) - 6z)<BR> which fac=
tors into<BR> ((n+2)(n+3)z^(n+1) - an(n+1)z^(n-1) - n(n-1)z^(n-2) -=
6z)<BR> and<BR> ((3+n)z^(n+3) - a(n+1)z^(n+1) - nz^n - 3z^3 =
+ az) - ((2+n)z^(n+3) - anz^(n+1) - (n-1)z^n - 2z^3 - 1)<BR> which =
simplifies to<BR> z^(n+3) - az^(n+1) - z^n - z^3 + az + 1<BR> =
Note that p(z) =3D z^(n+3) - az^(n+1) - z^n - z^3 + az + 1.<BR> Th=
us the critical points of r(z) are the roots of p(z) and the roots of<BR>=
q(z) :=3D ((n+2)(n+3)z^(n+1) - an(n+1)z^(n-1) - n(n-1)z^(n-2) - 6z=
)<BR> These latter are the "interesting" critical points, as the ot=
her critical points of r(z) are all superattracting.<BR> Note that =
q(z) is divisible by z, so 0 is an "interesting" critical point of r(z), =
for n not one of 2, 1, or -1.<BR> This is the critical point used i=
n the below hrynewt_m formulas except for hrynewt2_m. For n =3D 2,<BR>&nb=
sp; q(z)/2 =3D 10z^3 - 3(a+1)z - 1<BR> Put z =3D y + (a+1)/10y to g=
et<BR> q(z)/2 =3D y^6 - y^3/10 + (a+1)^3/1000<BR> so<BR> =
; 2y^3 =3D 1/10 +/-sqrt(1/100 - 4(a+1)^3/1000),<BR> y =3D ((1/10 +/=
- -sqrt(1/100 - 4(a+1)^3/1000))/2)^(1/3)<BR> and<BR> z =3D ((1/=
10 +/-sqrt(1/100 - 4(a+1)^3/1000))/2)^(1/3) + (a+1)/((1/10 +/-sqrt(1/100 =
- - 4(a+1)^3)/1000)/2)^(1/3)<BR>}</DIV> <DIV> </DIV> <DIV>hrynewt_j { =
; p1 is Julia parameter, p2 is exponent n, p3 is tolerance (if 0, will ac=
t like 0.001).<BR> &=
nbsp; ; SLOW. Use predefined hrynewtnnn_j where possible.<BR> =
z =3D pixel, a =3D p1, n =3D p2, n1 =3D n - 1, r =3D p3<BR> IF(r =3D=
=3D 0)<BR> r =3D 0.001<BR> ENDIF<BR> :<BR>&=
nbsp; z2 =3D sqr(z)<BR> z3 =3D z*z2<BR> zn1 =3D z^n1<BR> =
; zno =3D (z*zn1 - 1)<BR> zzz =3D (z3 - a*z - 1)<BR> pz =3D z=
no*zzz<BR> ppz =3D zno*(3*z2 - a) + n*zn1*zzz<BR> z =3D z - p=
z/ppz,<BR> |pz| > r<BR>}</DIV> <DIV> </DIV> <DIV>hrynewt_m =
{ ; p2 is exponent n, p3 is tolerance (if 0, will act like 0.001).<BR>&nb=
sp; ; SLOW. U=
se predefined hrynewtnnn_m where possible.<BR> z =3D 0, a =3D pixel=
, n =3D p2, n1 =3D n - 1, r =3D p3<BR> IF(r =3D=3D 0)<BR> &nbs=
p; r =3D 0.001<BR> ENDIF<BR> :<BR> z2 =3D sqr(z)<=
BR> z3 =3D z*z2<BR> zn1 =3D z^n1<BR> zn =3D z*zn1<BR>&n=
bsp; zno =3D (zn - 1)<BR> zzz =3D (z3 - a*z - 1)<BR> pz =3D z=
no*zzz<BR> ppz =3D zno*(3*z2 - a) + n*zn1*zzz<BR> z =3D z - p=
z/ppz,<BR> |pz| > r<BR>}</DIV> <DIV> </DIV> <DIV>hrynewt2_j=
{ ; p1 is Julia parameter, p3 is tolerance (if 0, will act like 0.001).<=
BR> &nbs=
p; ; n =3D 2.<BR> z =3D pixel, a =3D p1, r =3D p3<BR> IF(r =3D=
=3D 0)<BR> r =3D 0.001<BR> ENDIF<BR> :<BR>&=
nbsp; z2 =3D sqr(z)<BR> z3 =3D z*z2<BR> zno =3D (z2 - 1)<BR>&=
nbsp; zzz =3D (z3 - a*z - 1)<BR> pz =3D zno*zzz<BR> ppz =3D z=
no*(3*z2 - a) + 2*z*zzz<BR> z =3D z - pz/ppz,<BR> |pz| > r=
<BR>}</DIV> <DIV> </DIV> <DIV>hrynewt2_m (XAXIS) { ; p3 is tolerance=
(if 0, will act like 0.001).<BR> &nbs=
p; ; n =3D 2.<BR> a =3D pixel,<BR>&nb=
sp; ap1 =3D a + 1,<BR> IF((real(ap1) >=3D 0) || ((abs(real(ap1))=
*(3^(0.5))) < abs(imag(ap1))))<BR> t =3D ((0.1 + (0.=
01 - 0.004*sqr(ap1)*ap1)^(0.5))/2)^(1/3),<BR> ELSE<BR> &=
nbsp; t =3D ((0.1 - (0.01 - 0.004*sqr(ap1)*ap1)^(0.5))/2)^(1/3),<BR> =
; ENDIF<BR> z =3D t + 0.1*ap1/t, r =3D p3<BR> IF(r =3D=3D 0)<=
BR> r =3D 0.001<BR> ENDIF<BR> :<BR> z=
2 =3D sqr(z)<BR> z3 =3D z*z2<BR> zno =3D (z2 - 1)<BR> z=
zz =3D (z3 - a*z - 1)<BR> pz =3D zno*zzz<BR> ppz =3D zno*(3*z=
2 - a) + 2*z*zzz<BR> z =3D z - pz/ppz,<BR> |pz| > r<BR>}</=
DIV> <DIV> </DIV> <DIV>hrynewt3_j { ; p1 is Julia parameter, p3 is t=
olerance (if 0, will act like 0.001).<BR> &n=
bsp; ; n =3D 3.<BR> z =3D pixel=
, a =3D p1, r =3D p3<BR> IF(r =3D=3D 0)<BR> r =3D=
0.001<BR> ENDIF<BR> :<BR> z2 =3D sqr(z)<BR> z3 =3D=
z*z2<BR> zno =3D (z3 - 1)<BR> zzz =3D zno - a*z<BR> tz=
2 =3D 3*z2<BR> pz =3D zno*zzz<BR> ppz =3D zno*(tz2 - a) + tz2=
*zzz<BR> z =3D z - pz/ppz,<BR> |pz| > r<BR>}</DIV> <DIV>&n=
bsp;</DIV> <DIV>hrynewt3_m (XAXIS) { ; p3 is tolerance (if 0, will act li=
ke 0.001).<BR>  =
; ; n =3D 3.<BR> z =3D 0, a =3D pixel, r =3D p3<BR>&nbs=
p; IF(r =3D=3D 0)<BR> r =3D 0.001<BR> ENDIF<BR>&n=
bsp; :<BR> z2 =3D sqr(z)<BR> z3 =3D z*z2<BR> zno =3D (z=
3 - 1)<BR> zzz =3D zno - a*z<BR> tz2 =3D 3*z2<BR> pz =3D=
zno*zzz<BR> ppz =3D zno*(tz2 - a) + tz2*zzz<BR> z =3D z - pz=
/ppz,<BR> |pz| > r<BR>}</DIV> <DIV> </DIV> <DIV>hrynewt4_j =
{ ; p1 is Julia parameter, p3 is tolerance (if 0, will act like 0.001).<B=
R>  =
; ; n =3D 4.<BR> z =3D pixel, a =3D p1, r =3D p3<BR> IF(r =3D=
=3D 0)<BR> r =3D 0.001<BR> ENDIF<BR> :<BR>&=
nbsp; z2 =3D sqr(z)<BR> z3 =3D z*z2<BR> zno =3D (z*z3 - 1)<BR=
> zzz =3D (z3 - a*z - 1)<BR> pz =3D zno*zzz<BR> ppz =3D=
zno*(3*z2 - a) + 4*z3*zzz<BR> z =3D z - pz/ppz,<BR> |pz| >=
; r<BR>}</DIV> <DIV> </DIV> <DIV>hrynewt4_m (XAXIS) { ; p3 is tolera=
nce (if 0, will act like 0.001).<BR> &=
nbsp; ; n =3D 4.<BR> z =3D 0, a =3D p=
ixel, r =3D p3<BR> IF(r =3D=3D 0)<BR> r =3D 0.001=
<BR> ENDIF<BR> :<BR> z2 =3D sqr(z)<BR> z3 =3D z*z=
2<BR> zno =3D (z*z3 - 1)<BR> zzz =3D (z3 - a*z - 1)<BR> =
pz =3D zno*zzz<BR> ppz =3D zno*(3*z2 - a) + 4*z3*zzz<BR> z =3D=
z - pz/ppz,<BR> |pz| > r<BR>}</DIV> <DIV> </DIV> <DIV>hryn=
ewt5_j { ; p1 is Julia parameter, p3 is tolerance (if 0, will act like 0.=
001).<BR> &nbs=
p; ; n =3D 5.<BR> z =3D pixel, a =3D p1, r =3D p3<BR> I=
F(r =3D=3D 0)<BR> r =3D 0.001<BR> ENDIF<BR> =
:<BR> z2 =3D sqr(z)<BR> z3 =3D z*z2<BR> zn1 =3D sqr(z2=
)<BR> zno =3D (z*zn1 - 1)<BR> zzz =3D (z3 - a*z - 1)<BR> =
; pz =3D zno*zzz<BR> ppz =3D zno*(3*z2 - a) + 5*zn1*zzz<BR> z=
=3D z - pz/ppz,<BR> |pz| > r<BR>}</DIV> <DIV> </DIV> <DIV>=
hrynewt5_m (XAXIS) { ; p3 is tolerance (if 0, will act like 0.001).<BR>&n=
bsp; ; =
n =3D 5.<BR> z =3D 0, a =3D pixel, r =3D p3<BR> IF(r =3D=3D 0=
)<BR> r =3D 0.001<BR> ENDIF<BR> :<BR> =
z2 =3D sqr(z)<BR> z3 =3D z*z2<BR> zn1 =3D sqr(z2)<BR> =
zno =3D (z*zn1 - 1)<BR> zzz =3D (z3 - a*z - 1)<BR> pz =3D zno=
*zzz<BR> ppz =3D zno*(3*z2 - a) + 5*zn1*zzz<BR> z =3D z - pz/=
ppz,<BR> |pz| > r<BR>}</DIV> <DIV> </DIV> <DIV>hrynewt17_j =
{ ; p1 is Julia parameter, p3 is tolerance (if 0, will act like 0.001).<B=
R>  =
; ; n =3D 17.<BR> z =3D pixel, a =3D p1, r =3D p3<BR> I=
F(r =3D=3D 0)<BR> r =3D 0.001<BR> ENDIF<BR> =
:<BR> z2 =3D sqr(z)<BR> z3 =3D z*z2<BR> zn1 =3D sqr(sq=
r(sqr(z2)))<BR> zno =3D (z*zn1 - 1)<BR> zzz =3D (z3 - a*z - 1=
)<BR> pz =3D zno*zzz<BR> ppz =3D zno*(3*z2 - a) + 17*zn1*zzz<=
BR> z =3D z - pz/ppz,<BR> |pz| > r<BR>}</DIV> <DIV> <=
/DIV> <DIV>hrynewt17_m (XAXIS) { ; p3 is tolerance (if 0, will act like 0=
.001).<BR> &nb=
sp; ; n =3D 17.<BR> z =3D 0, a =3D pixel, r =3D p3<BR>&=
nbsp; IF(r =3D=3D 0)<BR> r =3D 0.001<BR> ENDIF<BR=
> :<BR> z2 =3D sqr(z)<BR> z3 =3D z*z2<BR> zn1 =3D=
sqr(sqr(sqr(z2)))<BR> zno =3D (z*zn1 - 1)<BR> zzz =3D (z3 - =
a*z - 1)<BR> pz =3D zno*zzz<BR> ppz =3D zno*(3*z2 - a) + 17*z=
n1*zzz<BR> z =3D z - pz/ppz,<BR> |pz| > r<BR>}</DIV> <DIV>=
</DIV> <DIV>hrynewt33_j { ; p1 is Julia parameter, p3 is tolerance =
(if 0, will act like 0.001).<BR>  =
; ; n =3D 33.<BR> z =3D pixel, =
a =3D p1, r =3D p3<BR> IF(r =3D=3D 0)<BR> r =3D 0=
.001<BR> ENDIF<BR> :<BR> z2 =3D sqr(z)<BR> z3 =3D=
z*z2<BR> zn1 =3D sqr(sqr(sqr(sqr(z2))))<BR> zno =3D (z*zn1 -=
1)<BR> zzz =3D (z3 - a*z - 1)<BR> pz =3D zno*zzz<BR> p=
pz =3D zno*(3*z2 - a) + 33*zn1*zzz<BR> z =3D z - pz/ppz,<BR> =
|pz| > r<BR>}</DIV> <DIV> </DIV> <DIV>hrynewt33_m (XAXIS) { ; p3 =
is tolerance (if 0, will act like 0.001).<BR> &nbs=
p; ; n =3D 33.<BR> =
z =3D 0, a =3D pixel, r =3D p3<BR> IF(r =3D=3D 0)<BR> &n=
bsp; r =3D 0.001<BR> ENDIF<BR> :<BR> z2 =3D sqr(z)<BR>&=
nbsp; z3 =3D z*z2<BR> zn1 =3D sqr(sqr(sqr(sqr(z2))))<BR> zno =
=3D (z*zn1 - 1)<BR> zzz =3D (z3 - a*z - 1)<BR> pz =3D zno*zzz=
<BR> ppz =3D zno*(3*z2 - a) + 33*zn1*zzz<BR> z =3D z - pz/ppz=
,<BR> |pz| > r<BR>}<BR></DIV></BODY></HTML><DIV><BR><br clear=3D=
all><hr>Get Your Private, Free E-mail from MSN Hotmail at <a href=3D"http=
://www.hotmail.com">http://www.hotmail.com</a>.<br></p></DIV>
- ------=_NextPart_001_0000_01C0F062.2AD5F4A0--
- --------------------------------------------------------------
Thanks for using Fractint, The Fractals and Fractint Discussion List
Post Message: fractint@lists.xmission.com
Get Commands: majordomo@lists.xmission.com "help"
Administrator: twegner@fractint.org
Unsubscribe: majordomo@lists.xmission.com "unsubscribe fractint"
------------------------------
Date: Fri, 8 Jun 2001 21:29:56 -0400
From: "Multiple Bogeys" <neo_1061@hotmail.com>
Subject: (fractint) Hairy Newton
- ------=_NextPart_001_0000_01C0F062.2AD5F4A0
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Yesterday evening I set out to find an interesting family of Newton-based=
Mandelbrot mappings. The result was the collection of formulae below.
You'll note the lack of 3D disease -- false alarm. All my posts to anothe=
r listserv had it, but I now think it must be that listserv rather than M=
SN Exploder. Or rather, some interaction between the two (since some post=
s from that listserv don't have it), like the weird "email laser" that ha=
ppened with this listserv last week (lots of peoples' messages were being=
duplicated, but for some reason mine were showing up in sets of five or =
six!)...Anyone who's an expert on listservs care to speculate further on =
what might be going on? One thing is clear: recent mail software and list=
servs have unnecessary complexity, and while we like complexity in our fr=
actals, we could do without chaos erupting in the mail system we depend o=
n to communicate here...
The hrynewt_j and hrynewt_m formulae iterate Newton's method for p(z) =3D=
(z^n - 1)(z^3 - az - 1). Both n and a are parameters, as is the toleranc=
e, an inverse bailout radius about the roots of p. The Mandelbrot variant=
has a vary over the screen while initial z is zero; this is a critical p=
oint but not a root of p for n real and greater than 2. You can plug in o=
ther values of n -- arbitrary negative or even complex values -- but won'=
t generally be able to find minibrots unless n has a positive real part g=
reater than two. If n is not an integer, there will be branch cuts in bot=
h the Mandelbrot and the Julia variants.
The hrynewtnnn_j and _m formulae are optimized versions with specific val=
ues for n, mostly small positive integers. They avoid a slow arbitrary ex=
ponentiation, and for the smaller values of n re-use powers that are used=
on both sides of the polynomial or its derivative. The hrynewt2_m formul=
a also has the feature of using a critical point for initial z, instead o=
f zero (which is *not* a critical point for n =3D 2). The result is a pro=
per Mandelbrot view, but it has a branch cut due to a square root in the =
calculation of the critical point, which is a-dependent. The branch cut h=
as been intentionally manipulated to put it in a fairly unobtrusive place=
, but can't be eliminated; the full Mandelbrot for this one lives on a tw=
o-layer Riemann sheet like that of the square root function.
The hrynewtnnn_m formulae also use an (XAXIS) symmetry declaration. (The =
generic hrynewt_m can't use this without trashing the output for non-real=
values of n.)
Observations:
* Certain choices of n produce three-fold-symmetric Mandelbrot sets. Find=
out which!
* Mangled and occasionally also intact Mandelbrots can be extracted when =
n is "strange" but has a real part greater than 2.
* You get radial petals with n real, concentric patterns with n imaginary=
, and logarithmic spirals with complex n; the ratio of
real to imaginary parts determines whether the spiral is steep (n clos=
e to real) or shallow (n close to imaginary).
* The Mandelbrots are always quadratic -- for real n > 2, the critical po=
int at zero is nondegenerate, and the critical point
pair for hrynewt2_m is degenerate only at one specific value of a.
The formula file begins with an extensive comment that details the mathem=
atical constructions that informed their design.
comment {
We want a Newton's method with a large number of basins, most of which =
are fixed and predictable.
This is accomplished by choosing a polynomial function to solve compose=
d of two factors, one with many fixed roots,
the other with a few mobile ones:
p(z) =3D (z^n-1)(z^3-az-1).
The Newton iteration is:
z -> r(z)
where
r(z) =3D z - p(z)/p'(z)
=3D (zp'(z) - p(z))/p'(z)
We easily discover p'(z) to be
p'(z) =3D (z^n-1)(3z^2-a) + (nz^(n-1))(z^3-az-1)
=3D (3+n)z^(n+2) - a(n+1)z^n - nz^(n-1) - 3z^2 + a
so
(zp'(z) - p(z)) =3D (z^n-1)(3z^3-az)+(nz^n)(z^3-az-1) - (z^n-1)(z^3-az-=
1)
=3D (z^n-1)(2z^3+1)+(nz^n)(z^3-az-1)
=3D (2+n)z^(n+3) - anz^(n+1) - (n-1)z^n - 2z^3 - 1
and
r(z) =3D ((2+n)z^(n+3) - anz^(n+1) - (n-1)z^n - 2z^3 - 1)/((3+n)z^(n+2)=
- a(n+1)z^n - nz^(n-1) - 3z^2 + a)
Using the quotient rule the numerator of r'(z) is
((3+n)z^(n+2) - a(n+1)z^n - nz^(n-1) - 3z^2 + a)((n+2)(n+3)z^(n+2) - an=
(n+1)z^n - n(n-1)z^(n-1) - 6z^2) -
((2+n)z^(n+3) - anz^(n+1) - (n-1)z^n - 2z^3 - 1)((n+2)(n+3)z^(n+1) - =
an(n+1)z^(n-1) - n(n-1)z^(n-2) - 6z)
which factors into
((n+2)(n+3)z^(n+1) - an(n+1)z^(n-1) - n(n-1)z^(n-2) - 6z)
and
((3+n)z^(n+3) - a(n+1)z^(n+1) - nz^n - 3z^3 + az) - ((2+n)z^(n+3) - anz=
^(n+1) - (n-1)z^n - 2z^3 - 1)
which simplifies to
z^(n+3) - az^(n+1) - z^n - z^3 + az + 1
Note that p(z) =3D z^(n+3) - az^(n+1) - z^n - z^3 + az + 1.
Thus the critical points of r(z) are the roots of p(z) and the roots of
q(z) :=3D ((n+2)(n+3)z^(n+1) - an(n+1)z^(n-1) - n(n-1)z^(n-2) - 6z)
These latter are the "interesting" critical points, as the other critic=
al points of r(z) are all superattracting.
Note that q(z) is divisible by z, so 0 is an "interesting" critical poi=
nt of r(z), for n not one of 2, 1, or -1.
This is the critical point used in the below hrynewt_m formulas except =
for hrynewt2_m. For n =3D 2,
q(z)/2 =3D 10z^3 - 3(a+1)z - 1
Put z =3D y + (a+1)/10y to get
q(z)/2 =3D y^6 - y^3/10 + (a+1)^3/1000
so
2y^3 =3D 1/10 +/-sqrt(1/100 - 4(a+1)^3/1000),
y =3D ((1/10 +/-sqrt(1/100 - 4(a+1)^3/1000))/2)^(1/3)
and
z =3D ((1/10 +/-sqrt(1/100 - 4(a+1)^3/1000))/2)^(1/3) + (a+1)/((1/10 +/=
- -sqrt(1/100 - 4(a+1)^3)/1000)/2)^(1/3)
}
hrynewt_j { ; p1 is Julia parameter, p2 is exponent n, p3 is tolerance (i=
f 0, will act like 0.001).
; SLOW. Use predefined hrynewtnnn_j where possible.
z =3D pixel, a =3D p1, n =3D p2, n1 =3D n - 1, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zn1 =3D z^n1
zno =3D (z*zn1 - 1)
zzz =3D (z3 - a*z - 1)
pz =3D zno*zzz
ppz =3D zno*(3*z2 - a) + n*zn1*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt_m { ; p2 is exponent n, p3 is tolerance (if 0, will act like 0.00=
1).
; SLOW. Use predefined hrynewtnnn_m where possible.
z =3D 0, a =3D pixel, n =3D p2, n1 =3D n - 1, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zn1 =3D z^n1
zn =3D z*zn1
zno =3D (zn - 1)
zzz =3D (z3 - a*z - 1)
pz =3D zno*zzz
ppz =3D zno*(3*z2 - a) + n*zn1*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt2_j { ; p1 is Julia parameter, p3 is tolerance (if 0, will act lik=
e 0.001).
; n =3D 2.
z =3D pixel, a =3D p1, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zno =3D (z2 - 1)
zzz =3D (z3 - a*z - 1)
pz =3D zno*zzz
ppz =3D zno*(3*z2 - a) + 2*z*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt2_m (XAXIS) { ; p3 is tolerance (if 0, will act like 0.001).
; n =3D 2.
a =3D pixel,
ap1 =3D a + 1,
IF((real(ap1) >=3D 0) || ((abs(real(ap1))*(3^(0.5))) < abs(imag(ap1))))
t =3D ((0.1 + (0.01 - 0.004*sqr(ap1)*ap1)^(0.5))/2)^(1/3),
ELSE
t =3D ((0.1 - (0.01 - 0.004*sqr(ap1)*ap1)^(0.5))/2)^(1/3),
ENDIF
z =3D t + 0.1*ap1/t, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zno =3D (z2 - 1)
zzz =3D (z3 - a*z - 1)
pz =3D zno*zzz
ppz =3D zno*(3*z2 - a) + 2*z*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt3_j { ; p1 is Julia parameter, p3 is tolerance (if 0, will act lik=
e 0.001).
; n =3D 3.
z =3D pixel, a =3D p1, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zno =3D (z3 - 1)
zzz =3D zno - a*z
tz2 =3D 3*z2
pz =3D zno*zzz
ppz =3D zno*(tz2 - a) + tz2*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt3_m (XAXIS) { ; p3 is tolerance (if 0, will act like 0.001).
; n =3D 3.
z =3D 0, a =3D pixel, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zno =3D (z3 - 1)
zzz =3D zno - a*z
tz2 =3D 3*z2
pz =3D zno*zzz
ppz =3D zno*(tz2 - a) + tz2*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt4_j { ; p1 is Julia parameter, p3 is tolerance (if 0, will act lik=
e 0.001).
; n =3D 4.
z =3D pixel, a =3D p1, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zno =3D (z*z3 - 1)
zzz =3D (z3 - a*z - 1)
pz =3D zno*zzz
ppz =3D zno*(3*z2 - a) + 4*z3*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt4_m (XAXIS) { ; p3 is tolerance (if 0, will act like 0.001).
; n =3D 4.
z =3D 0, a =3D pixel, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zno =3D (z*z3 - 1)
zzz =3D (z3 - a*z - 1)
pz =3D zno*zzz
ppz =3D zno*(3*z2 - a) + 4*z3*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt5_j { ; p1 is Julia parameter, p3 is tolerance (if 0, will act lik=
e 0.001).
; n =3D 5.
z =3D pixel, a =3D p1, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zn1 =3D sqr(z2)
zno =3D (z*zn1 - 1)
zzz =3D (z3 - a*z - 1)
pz =3D zno*zzz
ppz =3D zno*(3*z2 - a) + 5*zn1*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt5_m (XAXIS) { ; p3 is tolerance (if 0, will act like 0.001).
; n =3D 5.
z =3D 0, a =3D pixel, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zn1 =3D sqr(z2)
zno =3D (z*zn1 - 1)
zzz =3D (z3 - a*z - 1)
pz =3D zno*zzz
ppz =3D zno*(3*z2 - a) + 5*zn1*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt17_j { ; p1 is Julia parameter, p3 is tolerance (if 0, will act li=
ke 0.001).
; n =3D 17.
z =3D pixel, a =3D p1, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zn1 =3D sqr(sqr(sqr(z2)))
zno =3D (z*zn1 - 1)
zzz =3D (z3 - a*z - 1)
pz =3D zno*zzz
ppz =3D zno*(3*z2 - a) + 17*zn1*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt17_m (XAXIS) { ; p3 is tolerance (if 0, will act like 0.001).
; n =3D 17.
z =3D 0, a =3D pixel, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zn1 =3D sqr(sqr(sqr(z2)))
zno =3D (z*zn1 - 1)
zzz =3D (z3 - a*z - 1)
pz =3D zno*zzz
ppz =3D zno*(3*z2 - a) + 17*zn1*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt33_j { ; p1 is Julia parameter, p3 is tolerance (if 0, will act li=
ke 0.001).
; n =3D 33.
z =3D pixel, a =3D p1, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zn1 =3D sqr(sqr(sqr(sqr(z2))))
zno =3D (z*zn1 - 1)
zzz =3D (z3 - a*z - 1)
pz =3D zno*zzz
ppz =3D zno*(3*z2 - a) + 33*zn1*zzz
z =3D z - pz/ppz,
|pz| > r
}
hrynewt33_m (XAXIS) { ; p3 is tolerance (if 0, will act like 0.001).
; n =3D 33.
z =3D 0, a =3D pixel, r =3D p3
IF(r =3D=3D 0)
r =3D 0.001
ENDIF
:
z2 =3D sqr(z)
z3 =3D z*z2
zn1 =3D sqr(sqr(sqr(sqr(z2))))
zno =3D (z*zn1 - 1)
zzz =3D (z3 - a*z - 1)
pz =3D zno*zzz
ppz =3D zno*(3*z2 - a) + 33*zn1*zzz
z =3D z - pz/ppz,
|pz| > r
}<br clear=3Dall><hr>Get Your Private, Free E-mail from MSN Hotmail at <a=
href=3D"http://www.hotmail.com">http://www.hotmail.com</a>.<br></p>
- ------=_NextPart_001_0000_01C0F062.2AD5F4A0
Content-Type: text/html; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
<HTML><BODY STYLE=3D"font:10pt verdana; border:none;"><DIV>Yesterday even=
ing I set out to find an interesting family of Newton-based Mandelbrot ma=
ppings. The result was the collection of formulae below.</DIV> <DIV> =
;</DIV> <DIV>You'll note the lack of 3D disease -- false alarm. All my po=
sts to another listserv had it, but I now think it must be that listserv =
rather than MSN Exploder. Or rather, some interaction between the two (si=
nce some posts from that listserv don't have it), like the weird "em=
ail laser" that happened with this listserv last week (lots of peoples' m=
essages were being duplicated, but for some reason mine were showing up i=
n sets of five or six!)...Anyone who's an expert on listservs care to spe=
culate further on what might be going on? One thing is clear: recent mail=
software and listservs have unnecessary complexity, and while we like co=
mplexity in our fractals, we could do without chaos erupting in the mail =
system we depend on to communicate here...</DIV> <DIV> </DIV> <DIV>T=
he hrynewt_j and hrynewt_m formulae iterate Newton's method for p(z) =3D =
(z^n - 1)(z^3 - az - 1). Both n and a are parameters, as is the tolerance=
, an inverse bailout radius about the roots of p. The Mandelbrot variant =
has a vary over the screen while initial z is zero; this is a critical po=
int but not a root of p for n real and greater than 2. You can =
plug in other values of n -- arbitrary negative or even complex values --=
but won't generally be able to find minibrots unless n has a positive re=
al part greater than two. If n is not an integer, there will be branch cu=
ts in both the Mandelbrot and the Julia variants.</DIV> <DIV> </DIV>=
<DIV>The hrynewtnnn_j and _m formulae are optimized versions with specif=
ic values for n, mostly small positive integers. They avoid a slow arbitr=
ary exponentiation, and for the smaller values of n re-use powers that ar=
e used on both sides of the polynomial or its derivative. The hrynewt2_m =
formula also has the feature of using a critical point for initial z, ins=
tead of zero (which is *not* a critical point for n =3D 2). The result is=
a proper Mandelbrot view, but it has a branch cut due to a square root i=
n the calculation of the critical point, which is a-dependent. The branch=
cut has been intentionally manipulated to put it in a fairly unobtrusive=
place, but can't be eliminated; the full Mandelbrot for this one lives o=
n a two-layer Riemann sheet like that of the square root function.</DIV> =
<DIV> </DIV> <DIV>The hrynewtnnn_m formulae also use an (XAXIS) symm=
etry declaration. (The generic hrynewt_m can't use this without trashing =
the output for non-real values of n.)</DIV> <DIV> </DIV> <DIV>Observ=
ations:</DIV> <DIV>* Certain choices of n produce three-fold-symmetric Ma=
ndelbrot sets. Find out which!</DIV> <DIV>* Mangled and occasionally also=
intact Mandelbrots can be extracted when n is "strange" but has a real p=
art greater than 2.</DIV> <DIV>* You get radial petals with n real, conce=
ntric patterns with n imaginary, and logarithmic spirals with complex n; =
the ratio of</DIV> <DIV> real to imaginary parts determines w=
hether the spiral is steep (n close to real) or shallow (n close to imagi=
nary).</DIV> <DIV>* The Mandelbrots are always quadratic -- for real n &g=
t; 2, the critical point at zero is nondegenerate, and the critical point=
</DIV> <DIV> pair for hrynewt2_m is degenerate only at o=
ne specific value of a.</DIV> <DIV> </DIV> <DIV>The formula file beg=
ins with an extensive comment that details the mathematical constructions=
that informed their design.</DIV> <DIV> </DIV> <DIV>comment {<BR>&n=
bsp; We want a Newton's method with a large number of basins, most of whi=
ch are fixed and predictable.<BR> This is accomplished by choosing =
a polynomial function to solve composed of two factors, one with many fix=
ed roots,<BR> the other with a few mobile ones:<BR> p(z) =3D =
(z^n-1)(z^3-az-1).<BR> The Newton iteration is:<BR> z -> r=
(z)<BR> where<BR> r(z) =3D z - p(z)/p'(z)<BR> &nbs=
p; =3D (zp'(z) - p(z))/p'(z)<BR> We easily discov=
er p'(z) to be<BR> p'(z) =3D (z^n-1)(3z^2-a) + (nz^(n-1))(z^3-az-1)=
<BR> =3D (3+n)z^(n+2) - a(n+1)z=
^n - nz^(n-1) - 3z^2 + a<BR> so<BR> (zp'(z) - p(z)) =3D (z^n-=
1)(3z^3-az)+(nz^n)(z^3-az-1) - (z^n-1)(z^3-az-1)<BR> &nb=
sp; &nbs=
p; =3D (z^n-1)(2z^3+1)+(nz^n)(z^3-az-1)<BR> =
&=
nbsp; =3D (2+n)z^(n+3) - anz^(n+1) - (n-1)z^n - 2z^3 - 1<BR> and<BR=
> r(z) =3D ((2+n)z^(n+3) - anz^(n+1) - (n-1)z^n - 2z^3 - 1)/((3+n)z=
^(n+2) - a(n+1)z^n - nz^(n-1) - 3z^2 + a)<BR> Using the quotient ru=
le the numerator of r'(z) is<BR> ((3+n)z^(n+2) - a(n+1)z^n - nz^(n-=
1) - 3z^2 + a)((n+2)(n+3)z^(n+2) - an(n+1)z^n - n(n-1)z^(n-1) - 6z^2) -<B=
R> ((2+n)z^(n+3) - anz^(n+1) - (n-1)z^n - 2z^3 - 1)((n+=
2)(n+3)z^(n+1) - an(n+1)z^(n-1) - n(n-1)z^(n-2) - 6z)<BR> which fac=
tors into<BR> ((n+2)(n+3)z^(n+1) - an(n+1)z^(n-1) - n(n-1)z^(n-2) -=
6z)<BR> and<BR> ((3+n)z^(n+3) - a(n+1)z^(n+1) - nz^n - 3z^3 =
+ az) - ((2+n)z^(n+3) - anz^(n+1) - (n-1)z^n - 2z^3 - 1)<BR> which =
simplifies to<BR> z^(n+3) - az^(n+1) - z^n - z^3 + az + 1<BR> =
Note that p(z) =3D z^(n+3) - az^(n+1) - z^n - z^3 + az + 1.<BR> Th=
us the critical points of r(z) are the roots of p(z) and the roots of<BR>=
q(z) :=3D ((n+2)(n+3)z^(n+1) - an(n+1)z^(n-1) - n(n-1)z^(n-2) - 6z=
)<BR> These latter are the "interesting" critical points, as the ot=
her critical points of r(z) are all superattracting.<BR> Note that =
q(z) is divisible by z, so 0 is an "interesting" critical point of r(z), =
for n not one of 2, 1, or -1.<BR> This is the critical point used i=
n the below hrynewt_m formulas except for hrynewt2_m. For n =3D 2,<BR>&nb=
sp; q(z)/2 =3D 10z^3 - 3(a+1)z - 1<BR> Put z =3D y + (a+1)/10y to g=
et<BR> q(z)/2 =3D y^6 - y^3/10 + (a+1)^3/1000<BR> so<BR> =
; 2y^3 =3D 1/10 +/-sqrt(1/100 - 4(a+1)^3/1000),<BR> y =3D ((1/10 +/=
- -sqrt(1/100 - 4(a+1)^3/1000))/2)^(1/3)<BR> and<BR> z =3D ((1/=
10 +/-sqrt(1/100 - 4(a+1)^3/1000))/2)^(1/3) + (a+1)/((1/10 +/-sqrt(1/100 =
- - 4(a+1)^3)/1000)/2)^(1/3)<BR>}</DIV> <DIV> </DIV> <DIV>hrynewt_j { =
; p1 is Julia parameter, p2 is exponent n, p3 is tolerance (if 0, will ac=
t like 0.001).<BR> &=
nbsp; ; SLOW. Use predefined hrynewtnnn_j where possible.<BR> =
z =3D pixel, a =3D p1, n =3D p2, n1 =3D n - 1, r =3D p3<BR> IF(r =3D=
=3D 0)<BR> r =3D 0.001<BR> ENDIF<BR> :<BR>&=
nbsp; z2 =3D sqr(z)<BR> z3 =3D z*z2<BR> zn1 =3D z^n1<BR> =
; zno =3D (z*zn1 - 1)<BR> zzz =3D (z3 - a*z - 1)<BR> pz =3D z=
no*zzz<BR> ppz =3D zno*(3*z2 - a) + n*zn1*zzz<BR> z =3D z - p=
z/ppz,<BR> |pz| > r<BR>}</DIV> <DIV> </DIV> <DIV>hrynewt_m =
{ ; p2 is exponent n, p3 is tolerance (if 0, will act like 0.001).<BR>&nb=
sp; ; SLOW. U=
se predefined hrynewtnnn_m where possible.<BR> z =3D 0, a =3D pixel=
, n =3D p2, n1 =3D n - 1, r =3D p3<BR> IF(r =3D=3D 0)<BR> &nbs=
p; r =3D 0.001<BR> ENDIF<BR> :<BR> z2 =3D sqr(z)<=
BR> z3 =3D z*z2<BR> zn1 =3D z^n1<BR> zn =3D z*zn1<BR>&n=
bsp; zno =3D (zn - 1)<BR> zzz =3D (z3 - a*z - 1)<BR> pz =3D z=
no*zzz<BR> ppz =3D zno*(3*z2 - a) + n*zn1*zzz<BR> z =3D z - p=
z/ppz,<BR> |pz| > r<BR>}</DIV> <DIV> </DIV> <DIV>hrynewt2_j=
{ ; p1 is Julia parameter, p3 is tolerance (if 0, will act like 0.001).<=
BR> &nbs=
p; ; n =3D 2.<BR> z =3D pixel, a =3D p1, r =3D p3<BR> IF(r =3D=
=3D 0)<BR> r =3D 0.001<BR> ENDIF<BR> :<BR>&=
nbsp; z2 =3D sqr(z)<BR> z3 =3D z*z2<BR> zno =3D (z2 - 1)<BR>&=
nbsp; zzz =3D (z3 - a*z - 1)<BR> pz =3D zno*zzz<BR> ppz =3D z=
no*(3*z2 - a) + 2*z*zzz<BR> z =3D z - pz/ppz,<BR> |pz| > r=
<BR>}</DIV> <DIV> </DIV> <DIV>hrynewt2_m (XAXIS) { ; p3 is tolerance=
(if 0, will act like 0.001).<BR> &nbs=
p; ; n =3D 2.<BR> a =3D pixel,<BR>&nb=
sp; ap1 =3D a + 1,<BR> IF((real(ap1) >=3D 0) || ((abs(real(ap1))=
*(3^(0.5))) < abs(imag(ap1))))<BR> t =3D ((0.1 + (0.=
01 - 0.004*sqr(ap1)*ap1)^(0.5))/2)^(1/3),<BR> ELSE<BR> &=
nbsp; t =3D ((0.1 - (0.01 - 0.004*sqr(ap1)*ap1)^(0.5))/2)^(1/3),<BR> =
; ENDIF<BR> z =3D t + 0.1*ap1/t, r =3D p3<BR> IF(r =3D=3D 0)<=
BR> r =3D 0.001<BR> ENDIF<BR> :<BR> z=
2 =3D sqr(z)<BR> z3 =3D z*z2<BR> zno =3D (z2 - 1)<BR> z=
zz =3D (z3 - a*z - 1)<BR> pz =3D zno*zzz<BR> ppz =3D zno*(3*z=
2 - a) + 2*z*zzz<BR> z =3D z - pz/ppz,<BR> |pz| > r<BR>}</=
DIV> <DIV> </DIV> <DIV>hrynewt3_j { ; p1 is Julia parameter, p3 is t=
olerance (if 0, will act like 0.001).<BR> &n=
bsp; ; n =3D 3.<BR> z =3D pixel=
, a =3D p1, r =3D p3<BR> IF(r =3D=3D 0)<BR> r =3D=
0.001<BR> ENDIF<BR> :<BR> z2 =3D sqr(z)<BR> z3 =3D=
z*z2<BR> zno =3D (z3 - 1)<BR> zzz =3D zno - a*z<BR> tz=
2 =3D 3*z2<BR> pz =3D zno*zzz<BR> ppz =3D zno*(tz2 - a) + tz2=
*zzz<BR> z =3D z - pz/ppz,<BR> |pz| > r<BR>}</DIV> <DIV>&n=
bsp;</DIV> <DIV>hrynewt3_m (XAXIS) { ; p3 is tolerance (if 0, will act li=
ke 0.001).<BR>  =
; ; n =3D 3.<BR> z =3D 0, a =3D pixel, r =3D p3<BR>&nbs=
p; IF(r =3D=3D 0)<BR> r =3D 0.001<BR> ENDIF<BR>&n=
bsp; :<BR> z2 =3D sqr(z)<BR> z3 =3D z*z2<BR> zno =3D (z=
3 - 1)<BR> zzz =3D zno - a*z<BR> tz2 =3D 3*z2<BR> pz =3D=
zno*zzz<BR> ppz =3D zno*(tz2 - a) + tz2*zzz<BR> z =3D z - pz=
/ppz,<BR> |pz| > r<BR>}</DIV> <DIV> </DIV> <DIV>hrynewt4_j =
{ ; p1 is Julia parameter, p3 is tolerance (if 0, will act like 0.001).<B=
R>  =
; ; n =3D 4.<BR> z =3D pixel, a =3D p1, r =3D p3<BR> IF(r =3D=
=3D 0)<BR> r =3D 0.001<BR> ENDIF<BR> :<BR>&=
nbsp; z2 =3D sqr(z)<BR> z3 =3D z*z2<BR> zno =3D (z*z3 - 1)<BR=
> zzz =3D (z3 - a*z - 1)<BR> pz =3D zno*zzz<BR> ppz =3D=
zno*(3*z2 - a) + 4*z3*zzz<BR> z =3D z - pz/ppz,<BR> |pz| >=
; r<BR>}</DIV> <DIV> </DIV> <DIV>hrynewt4_m (XAXIS) { ; p3 is tolera=
nce (if 0, will act like 0.001).<BR> &=
nbsp; ; n =3D 4.<BR> z =3D 0, a =3D p=
ixel, r =3D p3<BR> IF(r =3D=3D 0)<BR> r =3D 0.001=
<BR> ENDIF<BR> :<BR> z2 =3D sqr(z)<BR> z3 =3D z*z=
2<BR> zno =3D (z*z3 - 1)<BR> zzz =3D (z3 - a*z - 1)<BR> =
pz =3D zno*zzz<BR> ppz =3D zno*(3*z2 - a) + 4*z3*zzz<BR> z =3D=
z - pz/ppz,<BR> |pz| > r<BR>}</DIV> <DIV> </DIV> <DIV>hryn=
ewt5_j { ; p1 is Julia parameter, p3 is tolerance (if 0, will act like 0.=
001).<BR> &nbs=
p; ; n =3D 5.<BR> z =3D pixel, a =3D p1, r =3D p3<BR> I=
F(r =3D=3D 0)<BR> r =3D 0.001<BR> ENDIF<BR> =
:<BR> z2 =3D sqr(z)<BR> z3 =3D z*z2<BR> zn1 =3D sqr(z2=
)<BR> zno =3D (z*zn1 - 1)<BR> zzz =3D (z3 - a*z - 1)<BR> =
; pz =3D zno*zzz<BR> ppz =3D zno*(3*z2 - a) + 5*zn1*zzz<BR> z=
=3D z - pz/ppz,<BR> |pz| > r<BR>}</DIV> <DIV> </DIV> <DIV>=
hrynewt5_m (XAXIS) { ; p3 is tolerance (if 0, will act like 0.001).<BR>&n=
bsp; ; =
n =3D 5.<BR> z =3D 0, a =3D pixel, r =3D p3<BR> IF(r =3D=3D 0=
)<BR> r =3D 0.001<BR> ENDIF<BR> :<BR> =
z2 =3D sqr(z)<BR> z3 =3D z*z2<BR> zn1 =3D sqr(z2)<BR> =
zno =3D (z*zn1 - 1)<BR> zzz =3D (z3 - a*z - 1)<BR> pz =3D zno=
*zzz<BR> ppz =3D zno*(3*z2 - a) + 5*zn1*zzz<BR> z =3D z - pz/=
ppz,<BR> |pz| > r<BR>}</DIV> <DIV> </DIV> <DIV>hrynewt17_j =
{ ; p1 is Julia parameter, p3 is tolerance (if 0, will act like 0.001).<B=
R>  =
; ; n =3D 17.<BR> z =3D pixel, a =3D p1, r =3D p3<BR> I=
F(r =3D=3D 0)<BR> r =3D 0.001<BR> ENDIF<BR> =
:<BR> z2 =3D sqr(z)<BR> z3 =3D z*z2<BR> zn1 =3D sqr(sq=
r(sqr(z2)))<BR> zno =3D (z*zn1 - 1)<BR> zzz =3D (z3 - a*z - 1=
)<BR> pz =3D zno*zzz<BR> ppz =3D zno*(3*z2 - a) + 17*zn1*zzz<=
BR> z =3D z - pz/ppz,<BR> |pz| > r<BR>}</DIV> <DIV> <=
/DIV> <DIV>hrynewt17_m (XAXIS) { ; p3 is tolerance (if 0, will act like 0=
.001).<BR> &nb=
sp; ; n =3D 17.<BR> z =3D 0, a =3D pixel, r =3D p3<BR>&=
nbsp; IF(r =3D=3D 0)<BR> r =3D 0.001<BR> ENDIF<BR=
> :<BR> z2 =3D sqr(z)<BR> z3 =3D z*z2<BR> zn1 =3D=
sqr(sqr(sqr(z2)))<BR> zno =3D (z*zn1 - 1)<BR> zzz =3D (z3 - =
a*z - 1)<BR> pz =3D zno*zzz<BR> ppz =3D zno*(3*z2 - a) + 17*z=
n1*zzz<BR> z =3D z - pz/ppz,<BR> |pz| > r<BR>}</DIV> <DIV>=
</DIV> <DIV>hrynewt33_j { ; p1 is Julia parameter, p3 is tolerance =
(if 0, will act like 0.001).<BR>  =
; ; n =3D 33.<BR> z =3D pixel, =
a =3D p1, r =3D p3<BR> IF(r =3D=3D 0)<BR> r =3D 0=
.001<BR> ENDIF<BR> :<BR> z2 =3D sqr(z)<BR> z3 =3D=
z*z2<BR> zn1 =3D sqr(sqr(sqr(sqr(z2))))<BR> zno =3D (z*zn1 -=
1)<BR> zzz =3D (z3 - a*z - 1)<BR> pz =3D zno*zzz<BR> p=
pz =3D zno*(3*z2 - a) + 33*zn1*zzz<BR> z =3D z - pz/ppz,<BR> =
|pz| > r<BR>}</DIV> <DIV> </DIV> <DIV>hrynewt33_m (XAXIS) { ; p3 =
is tolerance (if 0, will act like 0.001).<BR> &nbs=
p; ; n =3D 33.<BR> =
z =3D 0, a =3D pixel, r =3D p3<BR> IF(r =3D=3D 0)<BR> &n=
bsp; r =3D 0.001<BR> ENDIF<BR> :<BR> z2 =3D sqr(z)<BR>&=
nbsp; z3 =3D z*z2<BR> zn1 =3D sqr(sqr(sqr(sqr(z2))))<BR> zno =
=3D (z*zn1 - 1)<BR> zzz =3D (z3 - a*z - 1)<BR> pz =3D zno*zzz=
<BR> ppz =3D zno*(3*z2 - a) + 33*zn1*zzz<BR> z =3D z - pz/ppz=
,<BR> |pz| > r<BR>}<BR></DIV></BODY></HTML><DIV><BR><br clear=3D=
all><hr>Get Your Private, Free E-mail from MSN Hotmail at <a href=3D"http=
://www.hotmail.com">http://www.hotmail.com</a>.<br></p></DIV>
- ------=_NextPart_001_0000_01C0F062.2AD5F4A0--
- --------------------------------------------------------------
Thanks for using Fractint, The Fractals and Fractint Discussion List
Post Message: fractint@lists.xmission.com
Get Commands: majordomo@lists.xmission.com "help"
Administrator: twegner@fractint.org
Unsubscribe: majordomo@lists.xmission.com "unsubscribe fractint"
------------------------------
End of fractint-digest V1 #570
******************************