SIAM CSE 2017 talk

Adaptive Methods for Cubature
Fred J. Hickernell
Department of Applied Mathematics, Illinois Institute of Technology
hickernell@iit.edu mypages.iit.edu/~hickernell
Thanks to Lan Jiang, Tony Jiménez Rugama, Jagadees Rathinavel,
and the rest of the the Guaranteed Automatic Integration Library (GAIL) team
Supported by NSF-DMS-1522687
For more details see H. (2017+), H. et al. (2017+), and Choi et al. (2013–2015)

Problem IID Monte Carlo Low Discrepancy Sampling Bayesian Cubature Numerical Examples References
Problem
answer(µ) =?, µ =
ż
Rd
f(x) ν(dx) P Rp
, pµn =
nÿ
i=1
wif(xi)
Given εa, εr, choose n and {answer dependent on pµn to guarantee
answer(µ) ´ {answer ď max(εa, εr |answer(µ)|) (with high probability)
adaptively and automatically
2/10

Problem
answer(µ) =?, µ =
ż
Rd
f(x) ν(dx) P Rp
, pµn =
nÿ
i=1
wif(xi)
E.g.,
option price =
ż
Rd
payoﬀ(x)
e´xT
Σ´1
x/2
(2π)d/2 |Σ|1/2
dx, Σ = min(i, j)T/d
d
i,j=1
Gaussian probability =
ż
[a,b]
e´xT
Σ´1
x/2
(2π)d/2 |Σ|1/2
dx
Genz (1993)
=
ż
[0,1]d´1
f(x) dx
Sobol’ indexj =
ş
[0,1]2d output(x) ´ output(xj : x1
´j) output(x1
) dx dx1
ş
[0,1]d output(x)2 dx ´
ş
[0,1]d output(x) dx
2
Bayesian estimatej =
ş
Rd βj prob(data|β) probprior(β) dβ
ş
Rd prob(data|β) probprior(β) dβ 2/10

Problem
answer(µ) =?, µ =
ż
Rd
f(x) ν(dx) P Rp
, pµn =
nÿ
i=1
wif(xi)
If µ P [pµn ´ errn, pµn + errn] (with high probability), then the optimal and
successful choice is
{answer =
ans´ max(εa, εr |ans+|) + ans+ max(εa, εr |ans´|)
max(εa, εr |ans+|) + max(εa, εr |ans´|)
where ans˘ :=
"
sup
inf
*
µ P [pµn ´ errn, pµn + errn]
answer(µ)
provided
|ans+ ´ ans´|
max(εa, εr |ans+|) + max(εa, εr |ans´|)
ď 1 (H. et al., 2017+)
2/10

Berry-Esseen Stopping Rule for IID Monte Carlo
µ =
ż
Rd
f(x) ν(dx)
pµn =
1
n
nÿ
i=1
f(xi), xi
IID
„ ν
Need µ P [^µn ´ errn, ^µn + errn]
with high probability
P[|µ ´ ^µn| ď errn] « 99%
for Φ ´
?
n errn /(1.2^σ) = 0.005
by the Central Limit Theorem
where ^σ2
is the sample variation
3/10

Berry-Esseen Stopping Rule for IID Monte Carlo
µ =
ż
Rd
f(x) ν(dx)
pµn =
1
n
nÿ
i=1
f(xi), xi
IID
„ ν
P[|µ ´ ^µn| ď errn] ě 99%
for Φ ´
?
n errn /(1.2^σnσ
)
+ ∆n(´
?
n errn /(1.2^σnσ
), κmax) = 0.0025
by the Berry-Esseen Inequality
where ^σ2
nσ
is the sample variation using an independent sample, and provided
that kurt(f(X)) ď κmax(nσ) (H. et al., 2013; Jiang, 2016)
3/10

Adaptive Low Discrepancy Sampling Cubature
µ =
ż
[0,1]d
f(x) dx
pµn =
1
n
nÿ
i=1
f(xi), xi Sobol’ or lattice
4/10

Adaptive Low Discrepancy Sampling Cubature
µ =
ż
[0,1]d
f(x) dx
pµn =
1
n
nÿ
i=1
f(xi), xi Sobol’ or lattice
Express µ ´ ^µn in terms of the Fourier coeﬃcients of f. Assuming that these
coeﬃcients do not decay erratically, the discrete transform, rfn,κ
(n´1
κ=0
, may be
used to bound the error reliably (H. and Jiménez Rugama, 2016; Jiménez Rugama and
H., 2016; H. et al., 2017+):
|µ ´ ^µn| ď errn := C(n, )
2 ´1ÿ
κ=2 ´1
rfn,κ
4/10

Bayesian Cubature—f Is Random
µ =
ż
Rd
f(x) ν(dx)
pµn =
nÿ
i=1
wi f(xi)
Assume f „ GP(0, C). Choose the wi to integrate the best estimate of f given the
data txi, f(xi)un
i=1 (Diaconis, 1988; O’Hagan, 1991; Ritter, 2000; Rasmussen and
Ghahramani, 2003)
P[|µ ´ ^µn| ď errn] = 99% for errn = an expression involving C and txi, f(xi)un
i=1
A de-randomized interpretation exists (H., 2017+)
5/10

Gaussian Probability
µ =
ż
[a,b]
exp ´1
2 tT
Σ´1
t
a
(2π)d det(Σ)
dt
Genz (1993)
=
ż
[0,1]d´1
f(x) dx
For some typical choice of a, b, Σ, d = 3, εa = 0; µ « 0.6763
Worst 10% Worst 10%
εr Method % Accuracy n Time (s)
IID Monte Carlo 100% 8.1E4 1.8E´2
1E´2 Sobol’ Sampling 100% 1.0E3 5.1E´3
Bayesian Lattice 100% 1.0E3 2.8E´3
IID Monte Carlo 100% 2.0E6 3.8E´1
6/10

Sobol’ Indices
Y = output(X), where X „ U[0, 1]d
; Sobol’ Indexj(µ) describes how much
coordinate j of input X inﬂuences output Y (Sobol’, 1990; 2001):
Sobol’ Indexj(µ) :=
µ1
µ2 ´ µ2
3
, j = 1, . . . , d
µ1 :=
ż
[0,1)2d
[output(x) ´ output(xj : x1
´j)]output(x1
) dx dx1
µ2 :=
ż
[0,1)d
output(x)2
dx, µ3 :=
ż
[0,1)d
output(x) dx.
output(x) = ´x1 + x1x2 ´ x1x2x3 + ¨ ¨ ¨ + x1x2x3x4x5x6 (Bratley et al., 1992)
εa = 1E´3, εr = 0 j 1 2 3 4 5 6
n 65 536 32 768 16 384 16 384 2 048 2 048
Sobol’ Indexj 0.6529 0.1791 0.0370 0.0133 0.0015 0.0015
{Sobol’ Indexj 0.6528 0.1792 0.0363 0.0126 0.0010 0.0012
Sobol’ Indexj(pµn) 0.6492 0.1758 0.0308 0.0083 0.0018 0.0039
7/10

Thank you
Slides available at
www.slideshare.net/fjhickernell/siam-cse-2017-talk

References I
Bratley, P., B. L. Fox, and H. Niederreiter. 1992. Implementation and tests of low-discrepancy
sequences, ACM Trans. Model. Comput. Simul. 2, 195–213.
Choi, S.-C. T., Y. Ding, F. J. H., L. Jiang, Ll. A. Jiménez Rugama, X. Tong, Y. Zhang, and X. Zhou.
2013–2015. GAIL: Guaranteed Automatic Integration Library (versions 1.0–2.1).
Cools, R. and D. Nuyens (eds.) 2016. Monte Carlo and quasi-Monte Carlo methods: MCQMC,
Leuven, Belgium, April 2014, Springer Proceedings in Mathematics and Statistics, vol. 163,
Springer-Verlag, Berlin.
Diaconis, P. 1988. Bayesian numerical analysis, Statistical decision theory and related topics IV,
Papers from the 4th Purdue symp., West Lafayette, Indiana 1986, pp. 163–175.
Genz, A. 1993. Comparison of methods for the computation of multivariate normal probabilities,
Computing Science and Statistics 25, 400–405.
H., F. J. 2017+. Error analysis of quasi-Monte Carlo methods. submitted for publication,
arXiv:1702.01487.
H., F. J., L. Jiang, Y. Liu, and A. B. Owen. 2013. Guaranteed conservative ﬁxed width conﬁdence
intervals via Monte Carlo sampling, Monte Carlo and quasi-Monte Carlo methods 2012, pp. 105–128.
H., F. J. and Ll. A. Jiménez Rugama. 2016. Reliable adaptive cubature using digital sequences,
Monte Carlo and quasi-Monte Carlo methods: MCQMC, Leuven, Belgium, April 2014, pp. 367–383.
arXiv:1410.8615 [math.NA].
9/10

References II
H., F. J., Ll. A. Jiménez Rugama, and D. Li. 2017+. Adaptive quasi-Monte Carlo methods. submitted
for publication, arXiv:1702.01491 [math.NA].
Jiang, L. 2016. Guaranteed adaptive Monte Carlo methods for estimating means of random
variables, Ph.D. Thesis.
Jiménez Rugama, Ll. A. and F. J. H. 2016. Adaptive multidimensional integration based on rank-1
lattices, Monte Carlo and quasi-Monte Carlo methods: MCQMC, Leuven, Belgium, April 2014,
pp. 407–422. arXiv:1411.1966.
O’Hagan, A. 1991. Bayes-Hermite quadrature, J. Statist. Plann. Inference 29, 245–260.
Rasmussen, C. E. and Z. Ghahramani. 2003. Bayesian Monte Carlo, Advances in Neural Information
Processing Systems, pp. 489–496.
Ritter, K. 2000. Average-case analysis of numerical problems, Lecture Notes in Mathematics,
vol. 1733, Springer-Verlag, Berlin.
Sobol’, I. M. 1990. On sensitivity estimation for nonlinear mathematical models, Matem. Mod. 2,
no. 1, 112–118.
. 2001. Global sensitivity indices for nonlinear mathematical models and their monte carlo
estimates, Math. Comput. Simul. 55, no. 1-3, 271–280.
10/10

SIAM CSE 2017 talk

More Related Content

What's hot (20)

Viewers also liked (11)

Similar to SIAM CSE 2017 talk (20)

Recently uploaded (20)

SIAM CSE 2017 talk