Fast algorithms based on asymptotic expansions of special functions Alex Townsend MIT Cornell University, 23rd January 2015, CAM Colloquium
Fast algorithms based on asymptotic expansions of special functions
Alex Townsend
MIT
Cornell University, 23rd January 2015, CAM Colloquium
Introduction
Two trends in the computing era
Alex Townsend @MIT
Tabulations Software
Logarithms
Gauss quadrature
Special functions
1/22
Introduction
Two trends in the computing era
Alex Townsend @MIT
Classicalanalysis
Numericalanalysis
= ~~
Fast multipole method
Butterfly schemes
Iterative algorithms
Finite element method
Tabulations Software
Logarithms
Gauss quadrature
Special functions
1/22
Introduction
A selection of my research
Alex Townsend @MIT
๐(๐ฅ, ๐ฆ)
๐(๐ฅ, ๐ฆ)= 0
[Nakatsukasa, Noferini, & T., 2015]
2/22
๐(๐ฅ, ๐ฆ) โ
๐=1
๐พ
๐๐ ๐ฆ ๐๐(๐ฅ)
[Trefethen & T., 2014]
Algebraic geometry
Approx. theory
Introduction
A selection of my research
Alex Townsend @MIT
Algebraic geometry
๐(๐ฅ, ๐ฆ)
๐(๐ฅ, ๐ฆ)= 0
[Nakatsukasa, Noferini, & T., 2015]
Approx. theory
[Battels & Trefethen, 2004]
๐(๐ฅ, ๐ฆ) โ
๐=1
๐พ
๐๐ ๐ฆ ๐๐(๐ฅ)
[Trefethen & T., 2014]
2/22
Introduction
Many special functions are trigonometric-like
Alex Townsend @MIT
Trigonometric functionscos(ฯx), sin(๐๐ฅ)
Chebyshev polynomialsTn(x)
Legendre polynomialsPn(x)
Bessel functionsJv(z)
Airy functionsAi(x)
Also, Jacobi polynomials, Hermite polynomials, parabolic cylinder functions, etc.
3/22
Introduction
Many special functions are trigonometric-like
Alex Townsend @MIT
Trigonometric functionscos(ฯx), sin(๐๐ฅ)
Chebyshev polynomialsTn(x)
Legendre polynomialsPn(x)
Bessel functionsJv(z)
Airy functionsAi(x)
Also, Jacobi polynomials, Hermite polynomials, parabolic cylinder functions, etc.
3/22
Introduction
Many special functions are trigonometric-like
Alex Townsend @MIT
Trigonometric functionscos(ฯx), sin(๐๐ฅ)
Chebyshev polynomialsTn(x)
Legendre polynomialsPn(x)
Bessel functionsJv(z)
Airy functionsAi(x)
Also, Jacobi polynomials, Hermite polynomials, parabolic cylinder functions, etc.
3/22
Introduction
Many special functions are trigonometric-like
Alex Townsend @MIT
Trigonometric functionscos(ฯx), sin(๐๐ฅ)
Chebyshev polynomialsTn(x)
Legendre polynomialsPn(x)
Bessel functionsJv(z)
Airy functionsAi(x)
Also, Jacobi polynomials, Hermite polynomials, parabolic cylinder functions, etc.
3/22
Introduction
Many special functions are trigonometric-like
Alex Townsend @MIT
Trigonometric functionscos(ฯx), sin(๐๐ฅ)
Chebyshev polynomialsTn(x)
Legendre polynomialsPn(x)
Bessel functionsJv(z)
Airy functionsAi(x)
Also, Jacobi polynomials, Hermite polynomials, parabolic cylinder functions, etc.
3/22
Introduction
Many special functions are trigonometric-like
Alex Townsend @MIT
Trigonometric functionscos(ฯx), sin(๐๐ฅ)
Chebyshev polynomialsTn(x)
Legendre polynomialsPn(x)
Bessel functionsJv(z)
Airy functionsAi(x)
Also, Jacobi polynomials, Hermite polynomials, parabolic cylinder functions, etc.
3/22
Introduction
Asymptotic expansions of special functions
Alex Townsend @MIT
Legendre polynomials
Bessel functions
Thomas Stieltjes
Hermann Hankel
๐๐(cos ๐) ~ ๐ถ๐
๐=0
๐โ1
โ๐,๐
cos ๐ + ๐ +12๐ โ ๐ +
12๐2
(2 sin ๐)๐+12
, ๐ โ โ
๐ฝ0 ๐ง ~2
๐๐ง(cos ๐ง โ
๐
4
๐=0
๐โ1โ1 ๐๐2๐๐ง2๐
โ sin(๐ง โ๐
4)
๐=0
๐โ1(โ1)๐๐2๐+1๐ง2๐+1
)
๐ง โ โ
4/22
Introduction
Numerical pitfalls of an asymptotic expansionist
Alex Townsend @MIT
Fix M. Where is the asymptotic expansion accurate?
๐ฝ0 ๐ง =2
๐๐ง(cos ๐ง โ
๐
4
๐=0
๐โ1โ1 ๐๐2๐๐ง2๐
โ sin(๐ง โ๐
4)
๐=0
๐โ1(โ1)๐๐2๐+1๐ง2๐+1
) + ๐ ๐(๐ง)
5/22
Introduction
Talk outline
Alex Townsend @MIT
Using asymptotics expansions of special functions:
I. Computing Gauss quadrature rules. [Hale & T., 2013], [T., Trodgon, & Olver, 2015]
II. Fast transforms based on asymptotic expansions. [Hale & T., 2014], [T., 2015]
6/22
Part I: Computing Gauss quadrature rules
Alex Townsend @MIT
Part I: Computing Gauss quadrature rules
Pn(x) = Legendre polynomial of degree n
7/22
Part I: Computing Gauss quadrature rules
Definition
Alex Townsend @MIT
n-point quadrature rule: Gauss quadrature rule:
โ1
1
๐ ๐ฅ ๐๐ฅ โ
๐=1
๐
๐ค๐๐ ๐ฅ๐๐ฅ๐ = ๐๐กโ zero of ๐๐
๐ค๐ = 2(1 โ ๐ฅ๐2)-1 [๐๐
โฒ(๐ฅ๐)]-2
1. Exactly integrates polynomials of
degree โค 2n - 1. Best possible.
2. Computed by the GolubโWelsch
algorithm. [Golub & Welsch, 1969]
3. In 2004: no scheme for
computing Gauss rules in O(n). [Bailey, Jeyabalan, & Li, 2004]
8/22
Part I: Computing Gauss quadrature rules
Gauss quadrature experiments
Alex Townsend @MIT
Quadrature error Execution time
Qu
ad
ratu
re e
rro
r
Ex
ecu
tio
n t
ime
The GolubโWelsch algorithm is beautiful, but not the state-of-the-art.
There is another way...
n n
9/22
Part I: Computing Gauss quadrature rules
Newtonโs method
Alex Townsend @MIT
Newtonโs method
Isaac Newton
1. Evaluation scheme for Pn(๐๐๐ ฮธ) (๐ ๐๐ ฮธ)1/2
2. Evaluation scheme for ๐
๐๐Pn( ๐๐๐ ฮธ) (๐ ๐๐ ฮธ)1/21/2
3. Initial guesses for cos-1(xk)
Solve Pn(๐๐๐ ฮธ) (๐ ๐๐ ฮธ)1/2 = 0, ฮธ โ (0, ฯ).
(with asymptotics and a change of variables)
10/22
1. Evaluation scheme for Pn(๐๐๐ ฮธ) (๐ ๐๐ ฮธ)1/2
2. Evaluation scheme for ๐
๐๐Pn( ๐๐๐ ฮธ) (๐ ๐๐ ฮธ)1/2)1/2
3. Initial guesses for cos-1(xk)
Part I: Computing Gauss quadrature rules
Newtonโs methodโฆ with asymptotics and a change of variables
Alex Townsend @MIT
Newtonโs method
Isaac Newton
(with asymptotics and a change of variables)
Pn(x)
Solve Pn(๐๐๐ ฮธ) (๐ ๐๐ ฮธ)1/2 = 0, ฮธ โ (0, ฯ).
10/22
Part I: Computing Gauss quadrature rules
Newtonโs methodโฆ with asymptotics and a change of variables
Alex Townsend @MIT
Newtonโs method
Isaac Newton
1. Evaluation scheme for Pn(๐๐๐ ฮธ) (๐ ๐๐ ฮธ)1/2
2. Evaluation scheme for ๐
๐๐Pn( ๐๐๐ ฮธ) (๐ ๐๐ ฮธ)1/2)1/2
3. Initial guesses for cos-1(xk)
Solve Pn(๐๐๐ ฮธ) (๐ ๐๐ ฮธ)1/2 = 0, ฮธ โ (0, ฯ).
Pn(x)
(with asymptotics and a change of variables)
10/22
Part I: Computing Gauss quadrature rules
Newtonโs methodโฆ with asymptotics and a change of variables
Alex Townsend @MIT
Isaac Newton
Pn(๐๐๐ ฮธ) (๐ ๐๐ ฮธ)1/2
1. Evaluation scheme for Pn(๐๐๐ ฮธ) (๐ ๐๐ ฮธ)1/2
2. Evaluation scheme for ๐
๐๐Pn( ๐๐๐ ฮธ) (๐ ๐๐ ฮธ)1/2
3. Initial guesses for cos-1(xk)
Newtonโs method
Solve Pn(๐๐๐ ฮธ) (๐ ๐๐ ฮธ)1/2 = 0, ฮธ โ (0, ฯ).
(with asymptotics and a change of variables)
10/22
Part I: Computing Gauss quadrature rules
Newtonโs methodโฆ with asymptotics and a change of variables
Alex Townsend @MIT
Isaac Newton
Pn(๐๐๐ ฮธ) (๐ ๐๐ ฮธ)1/2
1. Evaluation scheme for Pn(๐๐๐ ฮธ) (๐ ๐๐ ฮธ)1/2
2. Evaluation scheme for ๐
๐๐Pn( ๐๐๐ ฮธ) (๐ ๐๐ ฮธ)1/2
3. Initial guesses for cos-1(xk)
Newtonโs method
Solve Pn(๐๐๐ ฮธ) (๐ ๐๐ ฮธ)1/2 = 0, ฮธ โ (0, ฯ).
(with asymptotics and a change of variables)
10/22
Part I: Computing Gauss quadrature rules
Newtonโs methodโฆ with asymptotics and a change of variables
Alex Townsend @MIT
Isaac Newton
Pn(๐๐๐ ฮธ) (๐ ๐๐ ฮธ)1/2
1. Evaluation scheme for Pn(๐๐๐ ฮธ) (๐ ๐๐ ฮธ)1/2
2. Evaluation scheme for ๐
๐๐Pn( ๐๐๐ ฮธ) (๐ ๐๐ ฮธ)1/2
3. Initial guesses for cos-1(xk)
Newtonโs method
(with asymptotics and a change of variables)
Solve Pn(๐๐๐ ฮธ) (๐ ๐๐ ฮธ)1/2 = 0, ฮธ โ (0, ฯ).
10/22
Part I: Computing Gauss quadrature rules
Evaluation scheme for Pn
Alex Townsend @MIT
Interior (in grey region):Pn(๐๐๐ ฮธ) (๐ ๐๐ ฮธ)1/2
Boundary:
[Baratella & Gatteschi, 1988]
11/22
๐๐(cos ๐)(sin ๐)1/2 โ ๐
12(๐ฝ0(๐๐)
๐=0
๐โ1๐ด๐(๐)
๐2๐+ ๐๐ฝ1(๐๐)
๐=0
๐โ1๐ต๐(๐)
๐2๐+1), ๐ = ๐ +
1
2
๐๐(cos ๐)(sin ๐)1/2
โ ๐ถ๐
๐=0
๐โ1
โ๐,๐
cos ๐ + ๐ +12๐ โ ๐ +
12๐2
2(2 sin ๐)๐
Theorem
For ๐ > 0, ๐ โฅ 200, and ฮธ๐๐๐๐๐ก= (๐ โ
1
2)๐
๐, Newton converges and the nodes are
computed to an accuracy of ๐ช(๐) in ๐ช(๐log(1/๐)) operations.
Part I: Computing Gauss quadrature rules
Evaluation scheme for Pn
Alex Townsend @MIT
Interior (in grey region):Pn(๐๐๐ ฮธ) (๐ ๐๐ ฮธ)1/2
Boundary:
[Baratella & Gatteschi, 1988]
๐๐(cos ๐)(sin ๐)1/2 โ ๐
12(๐ฝ0(๐๐)
๐=0
๐โ1๐ด๐(๐)
๐2๐+ ๐๐ฝ1(๐๐)
๐=0
๐โ1๐ต๐(๐)
๐2๐+1), ๐ = ๐ +
1
2
๐๐(cos ๐)(sin ๐)1/2
โ ๐ถ๐
๐=0
๐โ1
โ๐,๐
cos ๐ + ๐ +12๐ โ ๐ +
12๐2
2(2 sin ๐)๐
11/22
Theorem
For ๐ > 0, ๐ โฅ 200, and ฮธ๐๐๐๐๐ก= (๐ โ
1
2)๐
๐, Newton converges and the nodes are
computed to an accuracy of ๐ช(๐) in ๐ช(๐log(1/๐)) operations.
Part I: Computing Gauss quadrature rules
Evaluation scheme for Pn
Alex Townsend @MIT
Interior (in grey region):Pn(๐๐๐ ฮธ) (๐ ๐๐ ฮธ)1/2
Boundary:
Theorem
For ๐ โฅ 200 and ฮธ๐๐๐๐๐ก= (๐ โ
1
2)๐
๐, Newton converges to the nodes in ๐ช(๐) operations.
[Baratella & Gatteschi, 1988]
11/22
๐๐(cos ๐)(sin ๐)1/2 โ ๐
12(๐ฝ0(๐๐)
๐=0
๐โ1๐ด๐(๐)
๐2๐+ ๐๐ฝ1(๐๐)
๐=0
๐โ1๐ต๐(๐)
๐2๐+1), ๐ = ๐ +
1
2
๐๐(cos ๐)(sin ๐)1/2
โ ๐ถ๐
๐=0
๐โ1
โ๐,๐
cos ๐ + ๐ +12๐ โ ๐ +
12๐2
2(2 sin ๐)๐
Part I: Computing Gauss quadrature rules
Initial guesses for the nodes
Alex Townsend @MIT
Max error in the initial guesses
There are now better initial guesses. [Bogaert, 2014]
Interior: Use Tricomiโs
initial guesses. [Tricomi, 1950]
Boundary: Use Olverโs
initial guesses. [Olver, 1974]
Ignace Bogaert
Ma
x e
rro
r
12/22
Part I: Computing Gauss quadrature rules
The race to compute high-order Gauss quadrature
Alex Townsend @MIT
โThe race to computehigh-order Gauss quadratureโ
[T., 2015]
13/22
Part I: Computing Gauss quadrature rules
Applications
Alex Townsend @MIT
1. Psychological: Gauss quadrature
rules are cheap.
2. Experimental mathematics: High
precision quadratures. [Bailey, Jeyabalan,
& Li, 2004]
3. Cosmic microwave background:
Noise removal by L2-projection.
Generically, composite rules converge algebraically.
โ11 1
1+1002๐ฅ2dxQ
ua
dra
ture
err
or
14/22
Part II: Fast transforms based on asymptotic expansions
Alex Townsend @MIT
Part II: Fast transforms based on asymptoticexpansions
๐ท๐ต๐ =๐0 cos ๐0 โฏ ๐๐โ1 cos ๐0โฎ โฑ โฎ
๐0 cos ๐๐โ1 โฏ ๐๐โ1 cos ๐๐โ1
๐1โฎ๐๐โ1
, ๐๐=๐๐
๐ โ 1
15/22
๐0 ๐1
Part II: Fast transforms based on asymptotic expansions
Fast transforms as evaluation schemes
Alex Townsend @MIT 16/22
DFT ๐๐ = ๐=0๐โ1๐๐๐ง๐๐ , ๐ง๐ = ๐
โ2๐๐๐/๐
DCT ๐๐ = ๐=0๐โ1๐๐๐๐๐ (๐๐๐) , ๐๐=
๐๐
๐โ1
DCT ๐๐ = ๐=0๐โ1๐๐๐๐(๐๐๐ ๐๐)
DLT ๐๐ = ๐=0๐โ1๐๐๐๐(๐๐๐ ๐๐)
DHT ๐๐ = ๐=0๐โ1๐๐๐ฝ๐ฃ ๐๐ฃ,๐๐๐ , ๐๐ =
๐๐ฃ,๐
๐๐ฃ,๐, ๐๐ฃ,๐ = ๐๐กโ ๐๐๐๐ก ๐๐ ๐ฝ๐ฃ ๐ง
๐๐
๐๐๐๐
Part II: Fast transforms based on asymptotic expansions
Fast transforms as evaluation schemes
Alex Townsend @MIT 16/22
DFT ๐๐ = ๐=0๐โ1๐๐๐ง๐๐ , ๐ง๐ = ๐
โ2๐๐๐/๐
DCT ๐๐ = ๐=0๐โ1๐๐๐๐๐ (๐๐๐) , ๐๐=
๐๐
๐โ1
DCT ๐๐ = ๐=0๐โ1๐๐๐๐(๐๐๐ ๐๐)
DLT ๐๐ = ๐=0๐โ1๐๐๐๐(๐๐๐ ๐๐)
DHT ๐๐ = ๐=0๐โ1๐๐๐ฝ๐ฃ ๐๐ฃ,๐๐๐ , ๐๐ =
๐๐ฃ,๐
๐๐ฃ,๐, ๐๐ฃ,๐ = ๐๐กโ ๐๐๐๐ก ๐๐ ๐ฝ๐ฃ ๐ง
๐0 ๐1
๐๐
๐๐๐๐
Part II: Fast transforms based on asymptotic expansions
Fast transforms as evaluation schemes
Alex Townsend @MIT 16/22
DFT ๐๐ = ๐=0๐โ1๐๐๐ง๐๐ , ๐ง๐ = ๐
โ2๐๐๐/๐
DCT ๐๐ = ๐=0๐โ1๐๐๐๐๐ (๐๐๐) , ๐๐=
๐๐
๐โ1
DCT ๐๐ = ๐=0๐โ1๐๐๐๐(๐๐๐ ๐๐)
DLT ๐๐ = ๐=0๐โ1๐๐๐๐(๐๐๐ ๐๐)
DHT ๐๐ = ๐=0๐โ1๐๐๐ฝ๐ฃ ๐๐ฃ,๐๐๐ , ๐๐ =
๐๐ฃ,๐
๐๐ฃ,๐, ๐๐ฃ,๐ = ๐๐กโ ๐๐๐๐ก ๐๐ ๐ฝ๐ฃ ๐ง
๐๐
๐๐๐๐
๐0 ๐1
Part II: Fast transforms based on asymptotic expansions
Fast transforms as evaluation schemes
Alex Townsend @MIT
DFT ๐๐ = ๐=0๐โ1๐๐๐ง๐๐ , ๐ง๐ = ๐
โ2๐๐๐/๐
DCT ๐๐ = ๐=0๐โ1๐๐๐๐๐ (๐๐๐) , ๐๐=
๐๐
๐โ1
DCT ๐๐ = ๐=0๐โ1๐๐๐๐(๐๐๐ ๐๐)
DLT ๐๐ = ๐=0๐โ1๐๐๐๐(๐๐๐ ๐๐)
DHT ๐๐ = ๐=0๐โ1๐๐๐ฝ๐ฃ ๐๐ฃ,๐๐๐ , ๐๐ =
๐๐ฃ,๐
๐๐ฃ,๐, ๐๐ฃ,๐ = ๐๐กโ ๐๐๐๐ก ๐๐ ๐ฝ๐ฃ ๐ง
16/22
๐๐
๐๐๐๐
๐0 ๐1
Part II: Fast transforms based on asymptotic expansions
Related work
Alex Townsend @MIT
Many others: Candel, Cree, Johnson, Mori, Orszag, Suda, Sugihara, Takami, etc.
Fast transforms are very important.
Emmanuel Candes
Boag, Demanet,
Michielssen,
OโNeil, Ying
James Cooley
Ahmed, Tukey, Gauss,
Gentleman, Natarajan,
Rao
Leslie Greengard
Alpert, Barnes, Hut,
Martinsson, Rokhlin,
Mohlenkamp, Tygert
Daniel Potts
Driscoll, Healy,
Steidl, Tasche
17/22
๐๐(cos ๐๐) = ๐ถ๐
๐=0
๐โ1
โ๐,๐
cos ๐ + ๐ +12๐๐ โ ๐ +
12๐2
(2 sin ๐๐)๐ + 1/2 + ๐ ๐,๐(๐๐)
Part II: Fast transforms based on asymptotic expansions
Asymptotic expansions as a matrix decomposition
Alex Townsend @MIT
The asymptotic expansion
gives a matrix decomposition (sum of diagonally scaled DCTs and DSTs):
๐ท๐ต ๐ท๐ต๐จ๐บ๐ ๐น๐ด,๐ต= +
๐ท๐ต=
๐=0
๐โ1
๐ท๐ข๐๐ช๐ต๐ท๐ถโ๐ + ๐ท๐ฃ๐
0 0 00 ๐บ๐ตโ๐ 00 0 0
๐ท๐ถโ๐ + ๐น๐ด,๐ต
18/22
Part II: Fast transforms based on asymptotic expansions
Be careful and stay safe
Alex Townsend @MIT
Error curve: ๐ ๐,๐ ๐๐ = ๐
Fix M. Where is the asymptotic expansion accurate?
๐น๐ด,๐ต =
๐ ๐,๐(๐๐) โค2๐ถ๐โ๐,๐
(2 sin ๐๐)๐+1/2
19/22
Part II: Fast transforms based on asymptotic expansions
Partitioning and balancing competing costs
Alex Townsend @MIT
Theorem
The matrix-vector product ๐ = PN๐ can
be computed in ๐ช (๐((log๐)2/ log log๐)
operations.
ฮฑ too small
20/22
0 1
ฮฑ too small
Part II: Fast transforms based on asymptotic expansions
Partitioning and balancing competing costs
Alex Townsend @MIT
Theorem
The matrix-vector product ๐ = PN๐ can
be computed in ๐ช(๐((log๐)2/ log log๐)
operations.
20/22
0 1
Part II: Fast transforms based on asymptotic expansions
Partitioning and balancing competing costs
Alex Townsend @MIT
The matrix-vector product ๐ = ๐๐๐ can
be computed in ๐ช(๐((log๐)2/ log log๐)
operations.
ฮฑ too small
Theorem
20/22
0 1
Part II: Fast transforms based on asymptotic expansions
Partitioning and balancing competing costs
Alex Townsend @MIT
ฮฑ too large
Theorem
The matrix-vector product ๐ = PN๐ can
be computed in ๐ช(๐((log๐)2/ log log๐)
operations.
20/22
0 1
Part II: Fast transforms based on asymptotic expansions
Partitioning and balancing competing costs
Alex Townsend @MIT
ฮฑ = O (1/ log๐)
PN(j)๐ PN
EVAL๐
Theorem
๐ช (N (log N)
2
log log ๐) ๐ช (
N (log N)2
log log ๐)
The matrix-vector product ๐ = PN๐ can
be computed in ๐ช(๐((log๐)2/ log log๐)
operations.
20/22
0 1
Part II: Fast transforms based on asymptotic expansions
Numerical results
Alex Townsend @MIT 21/22
105
No precomputation.
0.15
0.1
0.05
0
-0.05
-0.1
-0.15
-0.2
-0.250.3 0.32 0.34
x
Modification of rough signals
โ1
1
๐๐ ๐ฅ ๐โ๐๐๐ก๐๐ฅ = ๐๐
2๐
โ๐๐ฝ๐+12(โ๐)
Direct approachUsing asymptotics
10210-3
10-2
10-1
100
103 104
Ex
ecu
tio
n t
ime
N
Future work
Fast spherical harmonic transform
Alex Townsend @MIT 22/22
Spherical harmonic transform:
๐ ๐, ๐ =
๐=0
๐โ1
๐=โ๐
๐
๐ผ๐๐๐๐๐ (cos ๐)๐๐๐๐
[Mohlenkamp, 1999]
[Rokhlin & Tygert, 2006]
[Tygert, 2008]
10-2
100
102
104
104E
xec
uti
on
tim
eN
105
PrecomputationFast transformDirect approach
A new generation of fast algorithms with no precomputation.
Existing approaches
103
Alex Townsend @MIT
Thank you
References
Alex Townsend @MIT
โข D. H. Bailey, K. Jeyabalan, and X. S. Li, A comparison of three high-precision quadrature schemes, 2005.
โข P. Baratella and I. Gatteschi, The bounds for the error term of an asymptotic approximation of Jacobipolynomials, Lecture notes in Mathematics, 1988.
โข Z. Battels and L. N. Trefethen, An extension of Matlab to continuous functions and operators, SISC,2004.
โข I. Bogaert, Iteration-free computation of GaussโLegendre quadrature nodes and weights, SISC, 2014.
โข G. H. Golub and J. H. Welsch, Calculation of Gauss quadrature rules, Math. Comput., 1969.
โข N. Hale and A. Townsend, Fast and accurate computation of GaussโLegendre and GaussโJacobiquadrature nodes and weights, SISC, 2013.
โข N. Hale and A. Townsend, A fast, simple, and stable ChebyshevโLegendre transform using an asymptoticformula, SISC, 2014.
โข Y. Nakatsukasa, V. Noferini, and A. Townsend, Computing the common zeros of two bivariate functionsvia Bezout resultants, Numerische Mathematik, 2014.
References
Alex Townsend @MIT
โข F. W. J. Olver, Asymptotics and Special Functions, Academic Press, New York, 1974.
โข A. Townsend and L. N. Trefethen, An extension of Chebfun to two dimensions, SISC, 2013.
โข A. Townsend, T. Trogdon, and S. Olver, Fast computation of Gauss quadrature nodes and weights on thewhole real line, to appear in IMA Numer. Anal., 2015.
โข A. Townsend, The race to compute high-order Gauss quadrature, SIAM News, 2015.
โข A. Townsend, A fast analysis-based discrete Hankel transform using asymptotics expansions, submitted,2015.
โข F. G. Tricomi, Sugli zeri dei polinomi sferici ed ultrasferici, Ann. Mat. Pura. Appl., 1950.