Monte Carlo Elliptic Integrals, Summaries of Geometry

Monte Carlo for elliptic integrals first looked at by Gary Gipson (1982) as part of his PhD thesis, 'The Coupling of Monte Carlo Integration with the.

Typology: Summaries

2022/2023

Uploaded on 02/28/2023

geek45
geek45 🇺🇸

4.4

(10)

274 documents

1 / 13

Toggle sidebar

This page cannot be seen from the preview

Don't miss anything!

bg1
1/13
Monte Carlo Elliptic Integrals
Circumference of an ellipse? Can’t do it mate.
Josh Fogg (they/them)
2021–10–15
pf3
pf4
pf5
pf8
pf9
pfa
pfd

Partial preview of the text

Download Monte Carlo Elliptic Integrals and more Summaries Geometry in PDF only on Docsity!

Monte Carlo Elliptic Integrals

Circumference of an ellipse? Can’t do it mate.

Josh Fogg (they/them)

Background

r

Circle Area = π r 2 Circumference = 2 π r

a

b

Ellipse Area = π ab Circumference = ???

Rough Approximations

The greats had a go:

  1. Geometric Average s = 2 π

√ ab (Johannes Kepler, 1609)

  1. Arithmetic Average

s = 2 π a^ + 2 b (Leonhard Euler, 1773)

  1. Root Mean Square

s = 2 π

√ a^2 + b^2 2 (Leonhard Euler, 1773)

Exact Solutions

Colin Maclaurin (1742):

s = 2 π a

∑^ ∞

m=

(2m)! m!m!

k^2 m 16 m(1 − 2 m)

where k =

b^2 a^2

Leonhard Euler (1776):

s = π

2(a^2 + b^2 )

∑^ ∞

m=

δ 16

)m · (4m − 3)!! (m!)^2

where δ =

a^2 − b^2 a^2 + b^2

James Ivory (1796):

s = π (a + b)

∑^ ∞

m=

(2m)! m!m!

hm 16 m(1 − 2 m)^2

where h =

a − b a + b

Carl Gauss (1812) & Eduard Kummer (1836) showed these could be expressed in terms of the hypergeometric function:

2 π a · F

, 1 , k^2

= π · F

, 1 , δ

= π (a + b) · F

, 1 , h

Better Approximations

Srinivasa Ramanujan’s Approximations (1914)

He gave two:

s = π

3(a + b) −

(3a + b)(a + 3b)

s = π (a + b)

3 h 10 +

4 − 3 h

where h =

a − b a + b

David Cantrell’s Approximation (2001)

Dropped this equation on a Geometry Research Google group:

s = 4(a + b) − 2(4 − π ) ab Hp

where Hp =

ap^ + bp 2

) p^1

. (3)

The value of p can be optimized for the type of ellipse:

pround = 3 π − 8 8 − 2 π , plong = ln(2) ln(2 / (4 − π )) , pgeneral = 0_._ 825_._

Better Approximations

Parker’s Approximation (2020)

s = π

a +

b −

269 a^2 + 667ab + 371b^2

where a > b

Lazy Parker’s Approximation (2020)

s = π

a +

b

where a > b

Monte Carlo: Method

s ≈ 4 a ×

( π 2

× 1

×

#red #red + #blue

Monte Carlo: Convergence

Plots of sample count against absolute percentage error for varying a.

(a) a = 1 (b) a = 1_._ 01 (c) a = 1_._ 1

(d) a = 2 (e) a = 5 (f) a = 80

Conclusion

  1. Don’t use Monte Carlo to do Elliptic Integrals in practice.
  2. Like really don’t... 3.... but it is a cool exercise!

The theory of elliptic functions is the fairyland of mathematics. The mathematician who once gazes upon this enchanting and wondrous domain crowded with the most beautiful relations and concepts is forever captivated. -Richard Bellman

Figure: Ellipses with b = 1 and a = 1, 5, 20, and 80.

Figure: Ellipse with b = 1 and a = 3_._ 93; has the same eccentricity as Halley’s comet.

Figure: Ellipse with b = 1 and a = 76_._ 7, has the eccentricity of most eccentric comet.