Benoit Mandelbrot

Download Report

Transcript Benoit Mandelbrot

FRACTAL_2
英國的海岸線有多長?
• http://atlas-zone.com/complex/fractals/dimension/coastline.htm
• 碎形幾何學之父 Benoit Mandelbrot 於
1967 年,在一篇幾乎算是他思想轉捩點的論
文中,如此地發問:「英國的海岸線有多
長?」他之所以會想到海岸線的問題,靈感
來自於英國數學家 Lewis Fry Richardson 遺
稿中一篇晦澀的論文,其中他所摸索的一大
堆爭議性主題,後來成為混沌理論(Chaos
Theory)的一部份。
• 當初 Lewis Fry Richardson 為了想要瞭解
一些國家鋸齒形的海岸線長度,所以翻閱
西班牙、葡萄牙、比利時與荷蘭的百科全
書,他發現書上在估計同一個國家的海岸
線長度時,竟然有百分之二十的誤差,
Lewis Fry Richardson 指出:這種誤差是因
為他們使用不同長度的量尺所導致的。
他同時發現海岸線長度 L 與測量尺度 s 的關
係如下,其中,值得注意的是 log(1/s) 與
log(L) 呈線性關係,其斜率為一定值 d:
log(L)
log(1/s)
•如果我們試圖實際測量英國海岸線的長度,我們
可以拿著兩腳器,先撐開一碼長,然後沿著英國
的海岸線行進,會測量出某個海岸線長度值,如
果我們將兩腳器的距離調小,再次測量海岸線,
那麼必定會得到比原先所測量的更長的海岸線長
度,那是因為,較小尺寸的量度可以掌握更多海
岸線的細節,我們可以這樣推論:用越小的量尺
來量度海岸線長度,所量出的結果會越長 。
• Benoit Mandelbrot 說,其實任何海岸線的長度在
某個意義下皆為無限長 ,或者說,海岸線的長度
是依量尺的長短而定。實際量度英國的海岸線實
在是太麻煩了,我們可以嘗試用不同的量尺去測
量 Koch Curve 的長度,Koch Curve 同樣具有海
岸線般的扭曲與轉折。如下圖所示,當我們用越
長的量尺去測量 Koch Curve 的長度,就會有越
多的細節無法量到,而當 s 趨近於無限小時,L
顯然也會趨近於無限長 ,但是,1/s 與 L 並不是
正比關係, 而是呈現指數關係 ,如左下方的關係
圖所示。其中,值得注意的是, L 相當於量尺的
測量次數(我們定義做 N(s))乘上量尺的長度,
可以寫成 L=N(s)*s。
Exploring Patterns in Nature Tutorials
• http://physionet.ph.biu.ac.il/tutorials/epn/
Center for Polymer Studies at Boston University
"Walking" Along a Coastline
EXERCISE:測量英國海岸線長
• 用ruler method測量英國地圖海岸線長,
畫圖找出1/s 與 L 的關係
• You will need:
a map of the coastline to be measured
and a pair of calipers(測徑器) .
EXERCISE:畫出海岸線
• Roll the die.
• If you get 1, push the midpoint left by 4 centimeters, hold it with the
thumb tack.
• If you get 2, push the midpoint left by 2 centimeters, hold it with the
thumb tack.
• If you get 3, leave the midpoint where it is, but hold it with a thumb
tack.
• If you get 4, push the midpoint right by 2 centimeters, hold it with the
thumb tack.
• If you get 5, push the midpoint right by 4 centimeters, it with the
thumb tack.
• If you get 6, roll again.
IRELAND
IRELAND
Grid method
Box Counting Method
• An alternative method is called the grid
method or box counting method or
covering method. The grid method is
a bit more versatile than the ruler
method, and can be used for different
kinds of fractals.
Box Counting Method
Covering the Coastline with Boxes
邊長
16
8
4
2
1
預測
測量
方塊數目
方塊數目
Fractal Dimension
How many disks does it take to cover the Koch
coastline? Well, it depends on their size of course. 1
disk with diameter 1 is sufficient to cover the whole
thing, 4 disks with diameter 1/3, 16 disks with
diameter 1/9, 64 disks with diameter 1/27, and so on.
In general, it takes 4n disks of radius (1/3)n to cover
the Koch coastline. If we apply this procedure to any
entity in any metric space we can define a quantity
that is the equivalent of a dimension. The HausdorffBesicovitch dimension of an object in a metric space
is given by the formula
ln N (h)
D  lim
h0 ln( 1 / h)
(1 / h)  N (h)
D
ln N (h)
D  lim
h0 ln( 1 / h)
where N(h) is the number of disks of radius h needed to
cover the object. Thus the Koch coastline has a
Hausdorff-Besicovitch dimension which is the limit of the
sequence
Is this really a dimension? Apply the procedure to the unit
line segment. It takes 1 disk of diameter 1, 2 disks of
diameter 1/2, 4 disks of diameter 1/4, and so on to cover
the unit line segment. In the limit we find a dimension of
This agrees with the topological dimension of the space.
The problem now is, how do we interpret a result like
1.261859507...? This does not agree with the
topological dimension of 1 but neither is it 2. The Koch
coastline is somewhere between a line and a plane. Its
dimension is not a whole number but a fraction. It is a
fractal. Actually fractals can have whole number
dimensions so this is a bit of a misnomer. A better
definition is that a fractal is any entity whose
Hausdorff-Besicovitch dimension strictly exceeds its
topological dimension (D > DT). Thus, the Peano spacefilling curve is also a fractal as we would expect it to be.
Even though its Hausdorff-Besicovitch dimension is a
whole number (D = 2) its topological dimension
(DT = 1) is strictly less than this.
→The monster has been tamed.
Surrounding the Koch Coastline with Boxes
(a way to determine its dimension)
Koch Coastline
log (1/h)
0
log N(h)
7.60837
-0.693147
-1.38629
-2.56495
-3.09104
7.04054
6.32972
4.85981
4.21951
-3.49651
-3.78419
-4.00733
3.52636
3.29584
3.04452
-4.18965
-4.34381
-4.47734
-5.17615
2.99573
2.70805
2.56495
1.60944
dimension (experimental) = 1.18
dimension (analytical) = 1.26
deviation = 6.35%
San Marco Dragon
log (1/h)
0
-0.693147
-1.38629
-2.63906
-3.21888
-3.61092
-3.91202
-4.12713
-4.31749
-4.46591
-4.60517
log N(h)
8.02355
7.29438
6.52209
5.03044
4.29046
4.00733
3.52636
3.3322
2.94444
2.83321
2.63906
dimension (experimental) = 1.16
dimension (analytical) = ???
deviation = ???
正方形
log (1/h)
0
-0.693147
-1.38629
-2.63906
-3.21888
-3.61092
-3.91202
-4.12713
-4.31749
-4.46591
-4.60517
-5.29832
log N(h)
11.0904
9.71962
8.31777
5.99146
4.79579
4.15888
3.58352
3.4012
3.21888
2.77259
2.77259
1.38629
dimension (experimental) = 1.82
dimension (analytical) = 2.00
deviation = 9.00%
練習題
• 使用Box Counting Method計算IRELAND海岸線
的碎形維度
log (1/h)
64
32
16
8
4
2
1
log N(h)
IRELAND 1 x 1
IRELAND 2 x 2
IRELAND 4 x 4
IRELAND 8 x 8
IRELAND 16 x 16
IRELAND 32 x 32
FRACTAL_3
• 1970年代左右,數學家 Benoit Mandelbrot 在一
篇幾乎算是他思想轉捩點的論文「英國的海岸線
有多長?」中,發展出了新的維度觀念 ─ 幾何學:
碎形(分形)。
• 三十年間,碎形幾何,與混沌理論,複雜性科學
共同匯合,試圖解釋過去科學家們所忽略的非線
性現象,與大自然的複雜結構,把觸角伸入,除
了物理、化學之外的生理學、經濟學、社會學、
氣象學,乃至於天文學所談及的星體分布。
• 搖身一變,碎形幾何已經變成了主要能描述大自
然的幾何學了。這些研究開拓了人們對於維度、
尺度、結構的新看法,大致歸納如下:
◆碎形具有分數維度:不同於整數維度的一維線段,
二維矩形,碎形所具有的維度是分數的,例如無
窮擴張三分之四的卡區曲線,其維度是
1.2618…。
◆碎形具有尺度無關性:對於「同一個」碎形結構,
以不同大小的量尺來量度「可觀察的區域」,碎
形會具有一致的碎形維度。例如,如果我們不同
程度地放大或縮小 Mandelbrot Set,我們會發現
圖形的複雜度,或摺疊程度,或粗糙程度並未因
此而改變。
◆碎形具有自我模仿性:對於「同一個」碎形結構,
自我模仿就是尺度一層一層縮小的結構重複性,
它們不僅在越來越小的尺度裡重複細節,而且是
以某種固定的方式將細節縮小尺寸,造成某種循
環重現的複雜現象。
◆碎形代表有限區域的無限結構:例如,卡區的雪
花曲線,是一條無限長,而結構不斷重複的線段,
被限制在最初三角形的正圓區域內。例如,原本
是一固定線段的 Cantor Set,最後變成一系列數
量無窮,但總長度卻為零的點集合。
◆碎形隱含一種整體性:我們可以從某一尺度的碎
形,來推知另一尺度的「同一個」碎形的大致樣
子,這意味著一種整體性,小細節的
傾向可以透露大細節的傾向,大細節的絲毫改變
可以令所有小細節全面改觀,再造成整個碎形圖
形的變化。
◆碎形是觀察手段的相對結果:回到 Mandelbrot 的
那篇論文「英國的海岸線有多長?」,作為碎形
結構的海岸線本身,在某種意義下是無限長,但
是對於不同的觀察者而言,海岸線長度卻端視其
手中的量尺(不同的觀察手段)而定,
Mandelbrot 說:「數據結果是依觀察者與其對象
而改變。」也正是這個觀念,才促使他發展出不
同於過去科學家的維度量度的新理論。
◆碎形是非線性動力過程的結果:大自然的外貌、結
構是非線性動力過程所造成的結果,我們也只能在
非線性現象中,才能找到碎形的蹤跡,於是碎形幾
何與非線性動力學有著密不可分的關係。
•
典型的碎形
Pythagorean Trees
Cantor Set
Sierpinski Gasket And Carpet
Koch Curve
Cesaro Curve
Levy Curve
Dragon Curve
Peano Curve
Hilbert Curve
H-Fractal
Tree Fractal
•
http://www.atlas-zone.com/complex/fractals/index.html
• 繪製碎形的方法
起始元與生成元疊代法(Initiator and Generator
Iteration Method)
幾何變換疊代函數系統(Deterministic Iterated
Function System)
隨機疊代函數系統(Random Iterated Function
Sysytems)
規範疊代函數系統(Formula Iterated Function
Sysytems)
Mandelbrot and Julia Sets
• The definition of the Mandelbrot set is :
the set of all the complex numbers, c,
such that the iteration of f ( z )  z 2  c is
bounded (starting with z =0 ).
• The Mandelbrot set is the graph of all
the complex numbers c, that do not go
to infinity when iterated in f ( z )  z 2  c ,
with a starting value of z =0 .
f ( z)  z  c
2
Mandelbrot set
• Named after Benoit B. Mandelbrot.
• A fractal generated by iterating:
2
z n 1  z n  c , z0  0
and plotting how fast it diverges to infinity for
different values of the complex number c
(speed represented as colours).
• The black set represents the "prisoner" points
that do not diverge: it is the Mandelbrot set.
Julia sets
• Named after Gaston Julia (1893--1978).
• A fractal generated by iterating:
z n 1  z n  c
2
• and plotting how fast it diverges to infinity for different
values of the complex number z (speed represented
as colours) for a set value of c. The black set
represents the "prisoner" points that do not diverge:
the border of this set is the Julia set.
• Values of c that lie within the Mandelbrot set result in
connected Julia sets; values of c from outside result in
disconnected Julia sets. We can draw an array of
Julia sets for various values of c, and map out the
Mandelbrot set.
Julia set
• A Julia set is almost the same thing. It
is defined to be : the set of all the
complex numbers, z , such that the
2
iteration of f ( z )  z  c is bounded
for a particular value of c. Again, more
simply put it is the graph of all the
complex numbers z, that do not go to
infinity when iterated in f ( z )  z 2  c ,
where c is constant.
Mandelbrot Set Explorer
• http://math.bu.edu/DYSYS/explorer/page1.html
Mandelbrot Explorer Gallery
• Julia and Mandelbrot Sets
http://aleph0.clarku.edu/~djoyce/julia/julia.html
• Mandelbrot and Julia Sets
http://www.cut-the-knot.org/blue/julia.shtml
• Julia Set Fractal (2D)
http://local.wasp.uwa.edu.au/~pbourke/fractals/juliaset/
• Color Cycling on the Mandelbrot Set
http://www.cut-the-knot.org/Curriculum/Magic/MandelCycle.shtml
• The Mandelbrot/Julia Set Applet
http://math.bu.edu/DYSYS/applets/JuliaIteration.html
Java Julia Set Generator
複動態系統的Julia Set
• http://www.emath.pu.edu.tw/celebrate/celebrate1/p3.htm
• http://www.ibiblio.org/e-notes/MSet/Anim/ManJuOrb.htm
•碎形與藝術
http://www.fractalus.com/