Dynamics of the family of complex maps F (z) z n z n , n2 (why the case n = 2 is ) with: Paul Blanchard Toni Garijo Matt Holzer U. Hoomiforgot Dan Look Sebastian Marotta Mark Morabito Monica Moreno Rocha Kevin Pilgrim Elizabeth Russell Yakov Shapiro David Uminsky F (z) z n The case n > 2 is great because: zn F (z) z n zn The case n > 2 is great because: There exists a McMullen domain around = 0 .... Parameter plane for n=3 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. F (z) z n zn The case n > 2 is great because: There exists a McMullen domain around = 0 .... by infinitely many “Mandelpinski” necklaces... ... surrounded QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. F (z) z n zn The case n > 2 is great because: There exists a McMullen domain around = 0 .... by infinitely many “Mandelpinski” necklaces... ... surrounded ... and the Julia sets behave nicely as 0 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. .01 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. .0001 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. .000001 F (z) z n zn The case n = 2 is crazy because: There is no McMullen domain.... QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. F (z) z n zn The case n = 2 is crazy because: There is no McMullen domain.... ... and no “Mandelpinski” necklaces... QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. F (z) z n zn The case n = 2 is crazy because: There is no McMullen domain.... ... and no “Mandelpinski” necklaces... ... and the Julia sets “go crazy” as 0 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. .01 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. .0001 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. .000001 Some definitions: Julia set of F (z) z n zn J = boundary of {orbits that escape to = closure {repelling periodic orbits} = {chaotic set} Fatou set = complement of J = predictable set } F (z) z 3 Computation of J: Color points that escape to infinity shades of red orange yellow green blue violet Black points do not escape. J = boundary of the black region. z3 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. .08i F (z) z 3 Easy computations: is superattracting, so have immediate basin B mapped n-to-1 to itself. z3 B QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. .08i F (z) z 3 Easy computations: is superattracting, so have immediate basin B mapped n-to-1 to itself. z3 B QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. T 0 is a pole, so have trap door T mapped n-to-1 to B. .08i F (z) z 3 Easy computations: is superattracting, so have immediate basin B mapped n-to-1 to itself. z3 B QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. T 0 is a pole, so have trap door T mapped n-to-1 to B. The Julia set has 2n-fold symmetry. .08i F (z) z 3 Easy computations: 2n free critical points c 1/2n z3 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. .08i F (z) z 3 Easy computations: 2n free critical points c 1/2n z3 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. .08i F (z) z 3 Easy computations: 2n free critical points c 1/2n z3 Only 2 critical values v 2 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. .08i F (z) z 3 Easy computations: 2n free critical points c 1/2n z3 Only 2 critical values v 2 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. .08i F (z) z 3 Easy computations: 2n free critical points c 1/2n z3 Only 2 critical values v 2 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. .08i F (z) z 3 Easy computations: 2n free critical points c 1/2n z3 Only 2 critical values v 2 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. But really only 1 free critical orbit .08i F (z) z 3 Easy computations: 2n free critical points c 1/2n z3 Only 2 critical values v 2 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. But really only 1 free critical orbit And 2n prepoles p ( )1/2n .08i The Escape Trichotomy (with D. Look & D Uminsky) There are three possible ways that the critical orbits can escape to infinity, and each yields a different type of Julia set. The Escape Trichotomy (with D. Look & D Uminsky) v B J(F ) is a Cantor set The Escape Trichotomy (with D. Look & D Uminsky) v B v T J(F ) is a Cantor set J(F ) is a Cantor set of simple closed curves (n > 2) (McMullen) The Escape Trichotomy (with D. Look & D Uminsky) v B v T J(F ) is a Cantor set J(F ) is a Cantor set of simple closed curves (n > 2) (McMullen) cases J(F ) is a connected set, and if In all other F (v ) T k J(F ) is a Sierpinski curve Case 1: v B QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. parameter plane when n = 3 J(F ) is a Cantor set v B Case 1: QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. parameter plane when n = 3 J(F ) is a Cantor set v B Case 1: J(F ) is a Cantor set QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. parameter plane when n = 3 J is a Cantor set v B Case 1: J(F ) is a Cantor set QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. v QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. c parameter plane when n = 3 J is a Cantor set Case 2: v T J(F ) is a Cantor set of simple closed curves QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. parameter plane when n = 3 Case 2: v T J(F ) is a Cantor set of simple closed curves QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. parameter plane when n = 3 Case 2: v T J(F ) is a Cantor set of simple closed curves QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. The central disk is the McMullen domain Case 2: v T J(F ) is a Cantor set of simple closed curves B QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. T parameter plane when n = 3 J is a Cantor set of simple closed curves Case 2: v T J(F ) is a Cantor set of simple closed curves v parameter plane when n = 3 c QuickTime™ and a TIFF (LZW) decompressor this picture. are needed to see QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. J is a Cantor set of simple closed curves k Case 3: F (v ) T QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. parameter plane when n = 3 J(F ) is a Sierpinski curve k Case 3: F (v ) T J(F ) is a Sierpinski curve QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. parameter plane when n = 3 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. A Sierpinski curve is a planar set homeomorphic to the Sierpinski carpet fractal Sierpinski curves are important for two reasons: 1. There is a “topological characterization” of the carpet 2. A Sierpinski curve is a “universal plane continuum” Topological Characterization Any planar set that is: 1. compact 2. connected 3. locally connected 4. nowhere dense 5. any two complementary domains are bounded by simple closed curves that are pairwise disjoint is a Sierpinski curve. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. The Sierpinski Carpet Universal Plane Continuum Any planar, one-dimensional, compact, connected set can be homeomorphically embedded in a Sierpinski curve. For example.... This set QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. This set QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. can be embedded inside QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. k Case 3: F (v ) T QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. parameter plane when n = 3 J(F ) is a Sierpinski curve k Case 3: F (v ) T QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. A Sierpinski “hole” J(F ) is a Sierpinski curve k Case 3: F (v ) T J(F ) is a Sierpinski curve QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. A Sierpinski “hole” A Sierpinski curve k Case 3: F (v ) T QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. J(F ) is a Sierpinski curve v c QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. F (v ) A Sierpinski “hole” A Sierpinski curve Escape time 3 k Case 3: F (v ) T J(F ) is a Sierpinski curve QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Another Sierpinski “hole” A Sierpinski curve k Case 3: F (v ) T J(F ) is a Sierpinski curve v QuickTime™ and a TIFF (LZW) 2 decompressor are needed to see this picture. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. F (v ) Another Sierpinski “hole” A Sierpinski curve Escape time 4 c k Case 3: F (v ) T J(F ) is a Sierpinski curve QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Another Sierpinski “hole” QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. A Sierpinski curve Escape time 7 k Case 3: F (v ) T J(F ) is a Sierpinski curve QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Another Sierpinski “hole” A Sierpinski curve Escape time 5 So to show that is homeomorphic to QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Need to show: QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. compact connected nowhere dense locally connected bounded by disjoint s.c.c.’s Need to show: QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. compact connected nowhere dense locally connected bounded by disjoint s.c.c.’s Fatou set is the union of the preimages of B; all disjoint, open disks. Need to show: QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. compact connected nowhere dense locally connected bounded by disjoint s.c.c.’s Fatou set is the union of the preimages of B; all disjoint, open disks. Need to show: QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. compact connected nowhere dense locally connected bounded by disjoint s.c.c.’s If J contains an open set, then J = C. Need to show: QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. compact connected nowhere dense locally connected bounded by disjoint s.c.c.’s If J contains an open set, then J = C. Need to show: QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. compact connected nowhere dense locally connected bounded by disjoint s.c.c.’s No recurrent critical orbits and no parabolic points. Need to show: QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. compact connected nowhere dense locally connected bounded by disjoint s.c.c.’s No recurrent critical orbits and no parabolic points. Need to show: QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. compact connected nowhere dense locally connected bounded by disjoint s.c.c.’s J locally connected, so the boundaries are locally connected. Need to show they are s.c.c.’s. Can only meet at (preimages of) critical points, hence disjoint. Need to show: QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. compact connected nowhere dense locally connected bounded by disjoint s.c.c.’s So J is a Sierpinski curve. Remark: All Julia sets drawn from Sierpinski holes are homeomorphic, but only those in symmetrically located Sierpinski holes have the same dynamics. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. The maps on all these Julia sets are dynamically different. Remark: All Julia sets drawn from Sierpinski holes are homeomorphic, but only those in symmetrically located Sierpinski holes have the same dynamics. In fact, there are exactly (n-1)(2n)k-3 Sierpinski holes with escape time k, and (2n)k-3 different conjugacy classes (n odd). (with K.Pilgrim) QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. The maps on all these Julia sets are dynamically different. Topics 1. The McMullen domain 2. Mandelpinski necklaces 3. Julia sets near 0 Part 1: The McMullen Domain Why is there no McMullen domain when n = 2? First suppose v T: What is the preimage of T? Why is there no McMullen domain when n = 2? First suppose v T: What is the preimage of T? Can the preimage be 2n disjoint disks, each of which contains a critical point? QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Why is there no McMullen domain when n = 2? First suppose v T: What is the preimage of T? Can the preimage be 2n disjoint disks, each of which contains a critical point? No --- there would then be 4n preimages of any point in T, but the map has degree 2n. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Why is there no McMullen domain when n = 2? So some of the preimages of T must overlap, and by 2n-fold symmetry, all must intersect. Why is there no McMullen domain when n = 2? So some of the preimages of T must overlap, and by 2n-fold symmetry, all must intersect. v By Riemann-Hurwitz, the preimage of T must then be an annulus. c QuickTime™ and a TIFF (LZW) decompressor this picture. are needed to see Why is there no McMullen domain when n = 2? So here is the picture: B A is the annulus separating B and T A T Why is there no McMullen domain when n = 2? So here is the picture: B A is the annulus separating B and T X F maps X 2n-to-1 onto T A T Why is there no McMullen domain when n = 2? So here is the picture: B A0 A is the annulus separating B and T X F maps X 2n-to-1 onto T A F maps both A0 and A1 as an n-to-1 covering onto A T A1 Why is there no McMullen domain when n = 2? So mod(A0) = 1/n mod(A) And mod(A1) = 1/n mod(A) B A0 X A T A1 Why is there no McMullen domain when n = 2? So mod(A0) = 1/n mod(A) And mod(A1) = 1/n mod(A) B A0 X When n = 2, mod(A0) + mod(A1) = mod(A) A T A1 Why is there no McMullen domain when n = 2? So mod(A0) = 1/n mod(A) And mod(A1) = 1/n mod(A) B A0 X When n = 2, mod(A0) + mod(A1) = mod(A) A T A1 So there is no room for X, i.e., v does not lie in T Why is there no McMullen domain when n = 2? Here is another reason: v 2 F (v ) 2 n n /2 1 n n /2 n n /2 2 n (n /21) 2 2 Why is there no McMullen domain when n = 2? Here is another reason: v 2 F (v ) 2 n n /2 1 n n /2 n n /2 2 n (n /21) 2 2 n 2 F (v ) as 0 so v lies in T when n > 2 Why is there no McMullen domain when n = 2? Here is another reason: v 2 F (v ) 2 n n /2 1 n n /2 n n /2 2 n (n /21) 2 2 n 2 F (v ) as 0 so v lies in T when n > 2 1 n 2 F (v ) 4 4 Why is there no McMullen domain when n = 2? Here is another reason: v 2 F (v ) 2 n n /2 1 n n /2 n n /2 2 n (n /21) 2 2 n 2 F (v ) as 0 so v lies in T when n > 2 1 n 2 F (v ) 4 4 so F (v ) 1/ 4 (???) as 0 Part 2: Mandelpinski Necklaces A “Mandelpinski necklace” is a simple closed curve in the parameter plane that passes alternately through k centers of baby Mandelbrot sets and k centers of Sierpinski holes. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Parameter plane for n = 3 A “Mandelpinski necklace” is a simple closed curve in the parameter plane that passes alternately through k centers of baby Mandelbrot sets and k centers of Sierpinski holes. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Parameter plane for n = 3 C1 passes through the centers of 2 M-sets and 2 S-holes A “Mandelpinski necklace” is a simple closed curve in the parameter plane that passes alternately through k centers of baby Mandelbrot sets and k centers of Sierpinski holes. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Parameter plane for n = 3 A “Mandelpinski necklace” is a simple closed curve in the parameter plane that passes alternately through k centers of baby Mandelbrot sets and k centers of Sierpinski holes. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. C2 passes through the centers of 4 M-sets* and 4 S-holes * only exception: 2 centers of period 2 bulbs, not M-sets Parameter plane for n = 3 A “Mandelpinski necklace” is a simple closed curve in the parameter plane that passes alternately through k centers of baby Mandelbrot sets and k centers of Sierpinski holes. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Parameter plane for n = 3 C3 passes through the centers of 10 M-sets and 10 S-holes A “Mandelpinski necklace” is a simple closed curve in the parameter plane that passes alternately through k centers of baby Mandelbrot sets and k centers of Sierpinski holes. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Parameter plane for n = 3 C4 passes through the centers of 28 M-sets and 28 S-holes A “Mandelpinski necklace” is a simple closed curve in the parameter plane that passes alternately through k centers of baby Mandelbrot sets and k centers of Sierpinski holes. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Parameter plane for n = 3 C5 passes through the centers of 82 M-sets and 82 S-holes Theorem: There exist closed curves Cj, j 1,..., surrounding the McMullen domain. Each Cj passes alternately through (n-2)nj-1 +1 centers of baby Mandelbrot sets and centers of Sierpinski holes. Theorem: There exist closed curves Cj, j 1,..., surrounding the McMullen domain. Each Cj passes alternately through (n-2)nj-1 +1 centers of baby Mandelbrot sets and centers of Sierpinski holes. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Parameter plane for n = 3 C14 passes through the centers of 4,782,969 M-sets and S-holes Theorem: There exist closed curves Cj, j 1,..., surrounding the McMullen domain. Each Cj passes alternately through (n-2)nj-1 +1 centers of baby Mandelbrot sets and centers of Sierpinski holes. C1: 3 holes and M-sets QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Parameter plane for n = 4 Theorem: There exist closed curves Cj, j 1,..., surrounding the McMullen domain. Each Cj passes alternately through (n-2)nj-1 +1 centers of baby Mandelbrot sets and centers of Sierpinski holes. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Parameter plane for n = 4 C2: 9 holes and M-sets* C3: 33 holes and M-sets F (z) z 3 Easy computations: All of the critical points and prepoles lie on the “critical circle” 0 : |z| = | | 1/2n z3 0 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. .08i F (z) z 3 Easy computations: All of the critical points and prepoles lie on the “critical circle” 0 : |z| = | | 1/2n QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. v z3 0 v which is mapped 2n-to-1 onto value line” the “critical connecting v .08i F (z) z 3 Easy computations: Any other circle around 0 is mapped n-to-1 to an ellipse whose foci are 2 z3 0 v QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. v .08i F (z) z 3 Easy computations: Any other circle around 0 is mapped n-to-1 to an ellipse whose foci are 2 z3 0 v QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. v .08i F (z) z 3 Easy computations: Any other circle around 0 is mapped n-to-1 to an ellipse whose foci are 2 So the exterior of 0 is mapped as an n-to-1 covering of the exterior of the critical value line. z3 0 v QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. v .08i F (z) z 3 Easy computations: Any other circle around 0 is mapped n-to-1 to an ellipse whose foci are 2 So the exterior of 0 is mapped as an n-to-1 covering of the exterior of the critical value line. 0. Same with the interior of z3 0 v QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. v .08i Simplest case: C1 Assume that v both sit on the critical circle. 2 | |1/2 | |1/2n (i.e., | v | | c | | p | ) p v v 0 Simplest case: C1 Assume that v both sit on the critical circle. 2 | |1/2 | |1/2n 2 2n | |n | | || 2 2n n1 p v v 0 Simplest case: C1 Assume that v both sit on the critical circle. 2 | |1/2 | |1/2n 2 2n | |n | | || 2 2n n1 This is the “dividing circle” in the parameter plane r = 2-3 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Parameter plane n = 3 Simplest case: C1 Assume that v both sit on the critical circle. 2 | |1/2 | |1/2n 2 2n | |n | | || 2 2n n1 This is the “dividing circle” in the parameter plane r = 2-8/3 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Parameter plane n = 4 Simplest case: C1 Assume that lies on the dividing circle, so v both sit on the critical circle. p v In this picture, is real and n = 3 v 0 Simplest case: C1 Assume that lies on the dividing circle, so v both sit on the critical circle. As rotates around the dividing circle, v rotates a half-turn, while p and c rotate 1/2n of a turn. So each v meets exactly n - 1 prepoles and critical points. p v In this picture, is real and n = 3 v 0 Simplest case: C1 Assume that lies on the dividing circle, so v both sit on the critical circle. As rotates around the dividing circle, v rotates a half-turn, while p and c rotate 1/2n of a turn. So each v meets exactly n - 1 prepoles and critical points. So the dividing circle is C1 p v In this picture, is real and n = 3 v 0 The dividing circle when n = 5 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. n-1 = 4 centers of Sierpinski holes; n-1 = 4 centers of baby Mandelbrot sets Now assume that v lies inside the critical circle: v v 0 Now assume that v lies inside the critical circle: The exterior of 0 is mapped n-to-1 onto the exterior of the critical value is a preimage 1 ray, so there mappedn-to-1 to 0 , v v 0 Now assume that v lies inside the critical circle: The exterior of 0 is mapped n-to-1 onto the exterior of the critical value is a preimage 1 ray, so there mappedn-to-1 to 0 , v v 0 1 Now assume that v lies inside the critical circle: The exterior of 0 is mapped n-to-1 onto the exterior of the critical value is a preimage 1 ray, so there mappedn-to-1 to 0 , then 2 is mapped n-to-1 to 1 , v v 0 1 2 Now assume that v lies inside the critical circle: The exterior of 0 is mapped n-to-1 onto the exterior of the critical value is a preimage 1 ray, so there mappedn-to-1 to 0 , then 2 is mapped n-to-1 to 1 , on and on and out to B v v 0 1 2 B 0 contains 2n critical points and 2n prepoles, so 1 contains 2n2 pre-critical points and pre-prepoles v v 0 1 0 contains 2n critical points and 2n prepoles, so 1 contains 2n2 pre-critical points and pre-prepoles k contains 2nk+1 points that map to the critical points and pre-prepoles under Fk v v 0 1 2 B As rotates by one turn, these 2nk+1 points on k each rotate by 1/2nk+1 of a turn. v v 0 k As rotates by one turn, these 2nk+1 points on k each rotate by 1/2nk+1 of a turn. Since 1 1n /2 F (v ) n n /2 n 2 2 the second iterate of the critical points rotate by 1 - n/2 of a turn v v 0 k As rotates by one turn, these 2nk+1 points on k each rotate by 1/2nk+1 of a turn. Since 1 1n /2 F (v ) n n /2 n 2 2 the second iterate of the critical points rotate by 1 - n/2 of a turn, so this point hits exactly (n / 2 1)(2n k 1 ) 1 (n 2)n k 1 v v 1 0 preimages of the critical points and prepoles on k k The real proof involves the Schwarz Lemma: There is a natural parametrization k ( ) of each k k ( ) v QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. v 0 k The real proof involves the Schwarz Lemma: There is a natural parametrization k ( ) of each k k ( ) v QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. v Best to restrict to a “symmetry region” inside the dividing circle, so that k ( ) is well-defined. 0 k Then we have a second map from the parameter plane to the dynamical plane, namely G( ) F (v ) which is invertible on the symmetry sector G 1 k ( ) QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. v v Best to restrict to a “symmetry region” inside the dividing circle, so that k ( ) is well-defined. 0 k Then we have a second map from the parameter plane to the dynamical plane, namely G( ) F (v ) which is invertible on the symmetry sector G 1 k ( ) QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. v v So consider the composition G 1 ( k ( ) ) a map from a “disk” to itself. 0 k Schwarz implies that G 1 ( k ( ) ) has a unique fixed point, i.e., a parameter for which the second iterate of the critical point lands on the point k ( ), so this proves the existence of the centers of the S-holes and M-sets. G 1 k ( ) QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. v v So consider the composition G 1 ( k ( ) ) a map from a “disk” to itself. 0 k Remarks: This proves the existence of centers of Sierpinski holes and Mandelbrot sets. Producing the entire M-set involves polynomial-like maps; while the entire S-hole involves qc-surgery. Part 3: Behavior of the Julia sets Part 3: Behavior of the Julia sets n = 2: When 0, the Julia set of F is the unit circle. But, as 0 , the Julia set of F converges to the closed unit disk Part 3: Behavior of the Julia sets n = 2: When 0, the Julia set of F is the unit circle. But, as 0 , the Julia set of F converges to the closed unit disk n > 2: J is always a Cantor set of “circles” when is small. Part 3: Behavior of the Julia sets n = 2: When 0, the Julia set of F is the unit circle. But, as 0 , the Julia set of F converges to the closed unit disk n > 2: J is always a Cantor set of “circles” when is small. Moreover, there is a 0 such that there is always a “round” annulus of “thickness” between two of these circles in the Fatou set. So J does not converge to the unit disk when n > 2. Theorem: When n = 2, the Julia sets converge to the unit disk as 0 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. n=2 Sketch of the proof: Suppose the Julia sets do not converge to the unit disk D as 0 Sketch of the proof: Suppose the Julia sets do not converge to the unit disk D as 0 Then there exists 0 and a sequence i 0 such that, for each i, there is a point zi D such that B (zi ) lies in the Fatou set. Sketch of the proof: Suppose the Julia sets do not converge to the unit disk D as 0 Then there exists 0 and a sequence i 0 such that, for each i, there is a point zi D such that B (zi ) lies in the Fatou set. B (z1 ) B (z3 ) B (z2 ) B (z4 ) Sketch of the proof: Suppose the Julia sets do not converge to the unit disk D as 0 Then there exists 0 and a sequence i 0 such that, for each i, there is a point zi D such that B (zi ) lies in the Fatou set. The zi must accumulate on some nonzero point, say z*, may assume that so we B (z* ) lies in the Fatou set for all i. B (z1 ) B (z3 ) B (z2 ) B (z4 ) Sketch of the proof: Suppose the Julia sets do not converge to the unit disk D as 0 Then there exists 0 and a sequence i 0 such that, for each i, there is a point zi D such that B (zi ) lies in the Fatou set. The zi must accumulate on some nonzero point, say z*, may assume that so we B (z* ) lies in the Fatou set for all i. B (z* ) Sketch of the proof: Suppose the Julia sets do not converge to the unit disk D as 0 Then there exists 0 and a sequence i 0 such that, for each i, there is a point zi D such that B (zi ) lies in the Fatou set. The zi must accumulate on some nonzero point, say z*, may assume that so we B (z* ) lies in the Fatou set for all i. 2 F z But for large i, i k F B (z* ) so i stretches into an “annulus” that surrounds the origin, so this disconnects the Julia set. F k B (z* ) So the Fatou components must become arbitrarily small: QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. 0.0001 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. 0.0000001 n > 2: Note the “round” annuli in the Fatou set; there is always such an annulus of some fixed width for | | small. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. .01 .0001 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. .00000001 .000001 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. .0000000001 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. .000000000001 Say n = 3: | | small T is tiny mod A0 = m is huge B A0 T Say n = 3: | | small T is tiny mod A0 = m is huge and the boundary of A0 is very close to S1 B A0 T Say n = 3: | | small T is tiny mod A0 = m is huge and the boundary of A0 is very close to S1 ~ A1 and A1 mapped to A0; X1 is mapped to T B A1 A0 ~ mod A1 = mod A 1 = mod X1 = m/3; ~ T XX 11 A1 A1 S1 Say n = 3: | | small T is tiny mod A0 = m is huge and the boundary of A0 is very close to S1 ~ A1 and A1 mapped to A0; X1 is mapped to T B A1 A2 T X1 X2 A2 is mapped to A1; X2 is mapped to X1 ~ mod A1 = mod A A1 S1 1 = mod X1 = m/3; mod A2 = mod X2 = m/32; A2 S1 Say n = 3: | | small T is tiny mod A0 = m is huge and the boundary of A0 is very close to S1 ~ A1 and A1 mapped to A0; X1 is mapped to T A2 is mapped to A1; X2 is mapped to X1 B Ak-1 Ak N Xk M Ak is mapped to Ak-1; Xk to Xk-1 ~ mod A1 = mod A1 = mod X1 = m/3; A1 S1 mod A2 = mod X2 = m/32; A2 S1 mod Ak = mod Xk = m/3k; Ak S1 Say n = 3: | | small T is tiny mod A0 = m is huge and the boundary of A0 is very close to S1 B Ak-1 Ak Xk Eventually find k so that 1 mod Ak < 3 and Ak S1 Say n = 3: | | small T is tiny mod A0 = m is huge and the boundary of A0 is very close to S1 B Ak-1 Ak Xk Eventually find k so that 1 mod Ak < 3 and Ak S1 Ak must contain a round annulus of modulus 1/2 1/2 (Ble, Douady, and Henriksen) Say n = 3: | | small T is tiny mod A0 = m is huge and the boundary of A0 is very close to S1 B Ak-1 Ak Xk Eventually find k so that 1 mod Ak < 3 and Ak S1 Ak must contain a round annulus of modulus 1/2 1/2 (Ble, Douady, and Henriksen) But does this annulus have definite “thickness?” Ak in here QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Ak in here mod Ak says that the inner boundary of Ak cannot be inside 0 or outside 1 , so the round annulus in Ak is “thick” Ak QuickTime™ and a 1 TIFF 0 (LZW) decompressor are needed to see this picture. Ak in here Same argument says that Ak Xk is twice as thick Ak Xk and a 1TIFFQuickTime™ 0(LZW) decompressor are needed to see this picture. So Xk must have definite thickness as well Xk QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Part 4: A major application Here’s the parameter plane when n = 2: QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Rotate it by 90 degrees: and this object appears everywhere..... QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture.