From CFD to computational finance

(and back again?)

Mike Giles

University of Oxford
Mathematical Institute

MIT Center for Computational Engineering Seminar

March 14th, 2013

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 1 / 47
Outline

a short personal history

computational finance
“smoking adjoints”
multilevel Monte Carlo

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 2 / 47
Short personal history

BA (Maths) at Cambridge, 1978–81

SM/PhD (Aero/Astro) at MIT, 1981–85
Assistant/Associate Prof in Aero/Astro, 1985–92
Reader/Prof in Computing Laboratory, Oxford, 1992–2004
moved to Mathematical Institute in 2004
acting head of the Mathematical and Computational Finance Group

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 3 / 47
Research history

CFD: 1981 – 1992

compressible flow CFD
turbomachinery applications (GTL and Rolls-Royce)
non-reflecting boundary conditions
vector/multi-threaded parallel computing, and visualisation

CFD: 1992 – 2004

distributed-memory parallel computing
adjoint methods for design
HYDRA CFD code – now the main CFD code used by Rolls-Royce
for all turbomachinery design
started getting into uncertainty quantification

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 4 / 47
Research history
Developing HYDRA was too much like hard work, not enough fun
— time for a change

Considered computational biology (but I know nothing about biology)

Instead moved to computational finance:

lots of opportunities (at least there were in 2004!)
close to London (one of top two international finance centres)
excellent colleagues in Mathematics working on modelling side
thought I could exploit my computational PDE knowledge

Currently:
50% effort on Monte Carlo methods – half in finance
50% effort on HPC, primarily using GPUs

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 5 / 47
Computational Finance

Options pricing – investment banks

Monte Carlo methods (60%)
PDEs / finite difference methods (30%)
other semi-analytic methods (10%)

High-frequency algorithmic trading – hedge funds

Might seem a bad time to be in this business, but as an academic it’s fine:
clear need for better models
regulators and internal risk management demanding more simulation
computational finance accounts for 10% of Top500 “supercomputers”
only problem is lack of research funding

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 6 / 47
Computational Finance

Computational finance reminds me of CFD about 20 years ago

not many academics working on numerical methods
codes are small – my biggest is probably 1000 lines
lots of low-hanging fruit, surprisingly more on the Monte Carlo side
than on the PDE side
Olivier Pironneau, Peter Forsyth and others moved earlier
from CFD to finance, but kept to PDEs
Monte Carlo researchers have mainly come from theoretical physics
(and don’t know about design optimisation)
in the past, each product group within a bank often had its own codes
— past 5 years have seen consolidation, often resulting in a single
corporate Monte Carlo system for both London and New York

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 7 / 47
SDEs in Finance

In computational finance, stochastic differential equations are used

to model the behaviour of
stocks
interest rates
exchange rates
weather
electricity/gas demand
crude oil prices
...

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 8 / 47
SDEs in Finance

Stochastic differential equations are just ordinary differential equations

plus an additional random source term.

The stochastic term accounts for the uncertainty of unpredictable

day-to-day events.

The aim is not to predict exactly what will happen in the future, but
to predict the probability of a range of possible things that might happen,
and compute some averages, or the probability of an excessive loss.

This is really just uncertainty quantification, and they’ve been doing it

for quite a while because they have so much uncertainty.

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 9 / 47
SDEs in Finance
multiple Geometric Brownian Motion paths
250

200

150
asset value

100

50

0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
years

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 10 / 47
SDEs in Finance

Examples:
Geometric Brownian motion (Black-Scholes model for stock prices)

dS = r S dt + σ S dW

Cox-Ingersoll-Ross model (interest rates)

√
dr = α(b − r ) dt + σ r dW

Heston stochastic volatility model (stock prices)

√
dS = r S dt + V S dW1
√
dV = λ (σ 2 −V ) dt + ξ V dW2

with correlation ρ between dW1 and dW2

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 11 / 47
Generic Problem

Stochastic differential equation with general drift and volatility terms:

dS(t) = a(S, t) dt + b(S, t) dW (t)

W (t) is a Wiener variable with the properties that for any q < r < s < t,
W (t)−W (s) is Normally distributed with mean 0 and variance t −s,
independent of W (r )−W (q).

In many finance applications, we want to compute the expected value of

an option dependent on the terminal state P(S(T ))

Other options depend on the average, minimum or maximum over the

whole time interval.

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 12 / 47
Euler Discretisation
Euler-Maruyama discretisation with timestep h:
bn+1 = S
S bn + a(S
bn , tn ) h + b(S
bn , tn ) ∆Wn

In the scalar case, each ∆Wn is a Normal random variable with mean 0
and variance h.

May seem very simple-minded but it’s hard to improve on the Euler
discretisation, and many codes are this simple.

Two different ways of measuring the discretisation error:

h i
Strong error: bn )2 = O(h)
E (S(tn ) − S
h i
Weak error: bn ) = O(h)
E P(S(tn )) − P(S

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 13 / 47
The “Greeks”

As well as estimating the value V = E[P], also very important to estimate

various first and second derivatives:
∂V ∂2V ∂V
∆= , Γ= , Vega =
∂S0 ∂S02 ∂σ

These are needed for “hedging” – banks try to hold a portfolio of different
financial products so that the effects of the random terms all cancel.

(Contrary to the public perception, banks generally do not intentionally

“gamble” with their assets.)

In some cases, can need 100 or more first order derivatives

=⇒ use of adjoints is natural

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 14 / 47
Monte Carlo simulation
If we want the expected value of a payoff function P which depends
on an underlying quantity S, so that
Z
V = E[P(S)] = P(S) pS (θ; S) dS

where pS is the probability distribution for S which depends on an input

parameter θ, then the Monte Carlo estimate is
N
X
Y =N −1
P(S (n) )
n=1

where the S (n) are generated independently, so V[Y ] = N −1 V[P(S)]

giving an O(N −1/2 ) sampling error.

To achieve an RMS error of ε requires O(ε−2 ) samples.

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 15 / 47
LRM sensitivity analysis
If pS is a differentiable function of θ, then
Z Z
∂V ∂pS ∂ log pS
= P(S) dS = P(S) pS (S) dS
∂θ ∂θ ∂θ
 
∂ log pS
= E P(S)
∂θ

which can be estimated as

N
X (n)
∂ log pS
N −1 P(S (n) )
∂θ
n=1

This is the Likelihood Ratio Method for computing sensitivities:

can handle discontinuous payoffs P(S)
requires a known distribution pS (e.g. log-normal)
usually has a much larger variance than alternatives

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 16 / 47
Pathwise sensitivity analysis
Alternatively, if S depends on θ and an independent random variable Z ,
then Z
V = E[P( S(θ; Z ) )] = P( S(θ; Z ) ) pZ (Z ) dZ

and Z  
∂V ∂P ∂S ∂P ∂S
= pZ (Z ) dZ = E ,
∂θ ∂S ∂θ ∂S ∂θ
which gives the pathwise sensitivity estimate
N
X ∂P
N −1
Ṡ (n)
∂S
n=1

where Ṡ is the path sensitivity keeping fixed all of the random numbers.
This is the “natural” sensitivity analysis which comes from differentiating
the original Monte Carlo estimate – this derivation shows the need for
P(S) to be at least continuous.
Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 17 / 47
“Smoking Adjoints”

My first finance paper in 2006 was with Paul Glasserman from Columbia
Business School:
“Smoking Adjoints: fast Monte Carlo Greeks” in Risk, the leading
monthly publication for the finance industry
explains how to do a discrete adjoint implementation of pathwise
sensitivity analysis for an application needing 100’s of Greeks,
at a cost less than double the original MC cost
Yves Achdou and Olivier Pironneau had previously used adjoints for
finance PDEs, but the technique hadn’t been transferred over to the
Monte Carlo side
absolutely nothing novel from an academic point of view, but has had
an impact in the industry – I think many banks now use it

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 18 / 47
Unsteady adjoints

Suppose that we have

Sn+1 = fn (Sn , θ), n = 0, 1, . . . , N − 1

given starting value S0 and parameter θ, and we want to compute P(SN ).

Differentiation gives this recurrence for Ṡn ≡ ∂Sn /∂θ,

∂fn ∂fn ∂P
Ṡn+1 = Ṡn + ≡ An Ṡn + bn , Ṗ = ṠN ,
∂Sn ∂θ ∂SN
and therefore, after some rearrangement,
N−1
∂P X
Ṗ = (AN−1 AN−2 . . . An+2 An+1 ) bn .
∂SN
n=0

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 19 / 47
Unsteady adjoints

Equivalently, we have
N−1 N−1
∂P X X
T
Ṗ = (AN−1 AN−2 . . . An+2 An+1 ) bn = vn+1 bn
∂SN
n=0 n=0

where vn are the adjoint variables defined by

 T
∂P
vN = , vn = A T
n vn+1 .
∂SN

Key points:
separate Ṡn calculation required for each input parameter θ
vn doesn’t depend on θ, so just one adjoint computation required
nothing here depends on whether the application is CFD or finance

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 20 / 47
Unsteady adjoints

need to store all of the Sn – this could be a problem in CFD

(use “checkpointing”) but not in finance

automatic differentiation tools can automate the generation of

the code (we used Tapenade for the Rolls-Royce HYDRA code)
but finance applications are simple enough to do by hand
(One of the early uses of adjoint AD tool was in EAPS in the
early 1990’s for an oceanographic application)

automatic differentiation theory assures you that the cost for

an unlimited number of first order sensitivities is no more than
a factor 4 greater than the original simulation cost

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 21 / 47
“Vibrato” Monte Carlo

Pathwise sensitivity analysis fails if P is discontinuous.

“Vibrato” Monte Carlo combines LRM for the final simulation timestep
with pathwise analysis for the rest of the path calculation.

Combines the strengths of the two methods:

can handle discontinuous payoffs
variance is always better than LRM, and similar to pathwise when
the payoff is continuous
has an efficient adjoint implementation when there are lots of
sensitivities to be computed

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 22 / 47
“Binning”

This was introduced by a collaborator, Luca Capriotti at Credit Suisse.

Monte Carlo analysis computes an estimate and a confidence interval.

Differentiating everything gives the sensitivity of both the estimate and the
confidence interval, but we want the confidence interval for the sensitivity.

In simple cases, compute the sensitivity analysis for each path, then
calculate the confidence interval.

In more complex cases with expensive pre-computations (e.g. Cholesky

factorisation of correlation matrix) need to group paths, compute the
sensitivity for each group, then combine into a confidence interval.

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 23 / 47
More on adjoints for finance

works naturally for finite difference applications

Black-Scholes PDE goes backwards in time from known payoff
to present value
adjoint goes forward in time – potentially a bit confusing
linked to natural duality between forward and backward Kolmogorov
equations

Anyone interested to learn more, please contact me for lecture notes, and
see
http://people.maths.ox.ac.uk/gilesm/libor/

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 24 / 47
Multilevel Monte Carlo

Coming from CFD, the use of adjoints was very natural.

What else is there? Multigrid!

But there’s no iterative solver here – instead just keep the central ideas of
a nested sequence of grids
fine grid accuracy at coarse grid cost

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 25 / 47
Multilevel Monte Carlo

Consider multiple levels of simulations with different timesteps

bℓ
hℓ = 2−ℓ T , ℓ = 0, 1, . . . , L, and payoff P

The expected value on the finest level, which is what we want, can be
expressed as a telescoping sum:
L
X
bL ] = E[P
E[P b0 ] + bℓ − P
E[P bℓ−1 ]
ℓ=1

The aim is to estimate the quantity on the left by independently

estimating each of the expectations on the right, and do so in a way
which minimises the overall variance for a fixed computational cost.

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 26 / 47
Multilevel Monte Carlo
bℓ − P
Key idea: approximate E[P bℓ−1 ] using Nℓ simulations with P
bℓ and P
bℓ−1
obtained using same Brownian path:
Nℓ 
X 
bℓ = N −1
Y P b (i )
b (i ) − P
ℓ ℓ ℓ−1
i =1

Why is this helpful?

bℓ ≈ P
P bℓ−1 since both approximate same P
Vℓ ≡ V[P bℓ − P
bℓ−1 ] is small, especially on finer levels
fewer samples needed to estimate E[P bℓ − P
bℓ−1 ]
end up using many, cheap samples on coarse levels,
and a few, expensive samples on fine levels

Easy implementation: generate Brownian increments ∆W for level ℓ,

then sum them pairwise to get increments for level ℓ−1

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 27 / 47
Multilevel Monte Carlo
Discrete Brownian path at different levels (with offset for clarity)
3

2.5

1.5

0.5

−0.5
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 28 / 47
Multilevel Monte Carlo
GBM paths at different levels (without offsets)
140

135

130

125

120

115

110

105

100

95
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 29 / 47
Multilevel Monte Carlo
If Cℓ is cost of a sample on level ℓ, the variance of the combined estimator
is
X L
Nℓ−1 Vℓ
ℓ=0
and its computational cost is
L
X
Nℓ Cℓ
ℓ=0

so the variance is minimised for fixed cost by choosing Nℓ√proportional to

p
Vℓ /Cℓ , and then the cost on level ℓ is proportional to Vℓ Cℓ .

In an SDE application with timestep hℓ , typically have Cℓ = O(hℓ−1 ) and

Vℓ = O(hℓ ), so the computational effort is spread evenly across all levels.

To achieve an O(ε) RMS error ends up requiring a cost which is

O(ε−2 (log ε)2 ), instead of usual O(ε−3 ).
Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 30 / 47
MLMC Theorem

Theorem: If there exist independent estimators Y bℓ based on Nℓ Monte

Carlo samples, each costing Cℓ , and positive constants α, β, γ, c1 , c2 , c3
such that α ≥ 12 min(β, γ) and

i) E[Pbℓ −P] ≤ c1 2−α ℓ


b0 ],
 E[P l =0
bℓ ] =
ii) E[Y
 E[P
bℓ − Pbℓ−1 ], l > 0
bℓ ] ≤ c2 N −1 2−β ℓ
iii) V[Y ℓ

iv) Cℓ ≤ c3 2γ ℓ

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 31 / 47
MLMC Theorem
then there exists a positive constant c4 such that for any ε < 1 there exist
L and Nℓ for which the multilevel estimator
L
X
b =
Y bℓ ,
Y
ℓ=0
 2 
b
has a mean-square-error with bound E Y − E[P] < ε2

and a computational cost C with bound



 c4 ε−2 , β > γ,

C≤ c4 ε−2 (log ε)2 , β = γ,



c4 ε−2−(γ−β)/α , 0 < β < γ.

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 32 / 47
MLMC Theorem

This is a powerful, general theorem:

applies to a wide range of stochastic processes

(SDEs, SPDEs, Lévy processes, Poisson processes, etc.)
applies also to a wide choice of numerical approximations, leaving
lots of flexibility for creating multilevel estimators with a variance Vℓ
which converges rapidly to zero

. . . and also a total “cheat”:

hard bit is constructing numerical approximations with good

properties
even harder bit is proving it – the numerical analysis can be really
tough

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 33 / 47
Multilevel Monte Carlo
Finance applications:
SDEs with Euler and Milstein discretisations
jump-diffusion and Lévy processes, either by directly simulating
increments, or by simulating all but the smallest jumps
support for a variety of path-dependent options
◮ digital – discontinuous function of final value
◮ Asian – based on an average over time interval
◮ lookback – based on minimum / maximum over interval
◮ barrier – knocked out if path crosses a certain level
◮ American options – much tougher because of optional early exercise
numerical analysis now exists for most of this
also some work on Multilevel quasi-Monte Carlo (using Sobol points
or rank-1 lattices instead of pseudo-random numbers)
10-15 groups worldwide working on different aspects

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 34 / 47
Back to CFD

I’m now working with Rob Scheichl (Bath) and Andrew Cliffe
(Nottingham) applying these ideas to the modelling of oil reservoirs and
groundwater contamination in nuclear waste repositories.

Here we have an elliptic SPDE coming from Darcy’s law:

 
∇· κ(x)∇p = 0

where the permeability κ(x) is uncertain, and log κ(x) is often modelled as
being Normally distributed with a spatial covariance such as

cov(log κ(x1 ), log κ(x2 )) = σ 2 exp(−kx1 −x2 k/λ)

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 35 / 47
Elliptic SPDE

A typical realisation of κ for λ = 0.001, σ = 1.

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 36 / 47
Elliptic SPDE
Samples of log k are provided by a Karhunen-Loève expansion:

X
∞ p
log k(x, ω) = θn ξn (ω) fn (x),
n=0

where θn , fn are eigenvalues / eigenfunctions of the correlation function:

Z
R(x, y) fn (y) dy = θn fn (x)

and ξn (ω) are standard Normal random variables.

Numerical experiments truncate the expansion.

(Latest 2D/3D work uses an efficient FFT construction based on a

circulant embedding.)

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 37 / 47
Elliptic SPDE
Decay of 1D eigenvalues
0
10
λ=0.01
λ=0.1
λ=1
−2
10
eigenvalue

−4
10

−6
10

0 1 2 3
10 10 10 10
n

When λ = 1, can use a low-dimensional polynomial chaos approach, but

it’s impractical for smaller λ.
Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 38 / 47
Elliptic SPDE

Discretisation:
cell-centred finite volume discretisation on a uniform grid – for rough
coefficients we need to make grid spacing very small on finest grid
each level of refinement has twice as many grid points in each
direction
current numerical experiments use a direct solver for simplicity,
but in 3D will use an efficient AMG multigrid solver with a cost
roughly proportional to the total number of grid points

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 39 / 47
2D Results

Boundary conditions for unit square [0, 1]2 :

– fixed pressure: p(0, x2 ) = 1, p(1, x2 ) = 0
– Neumann b.c.: ∂p/∂x2 (x1 , 0) = ∂p/∂x2 (x1 , 1) = 0
Z
∂p
Output quantity – mass flux: − k dx2
∂x1
Correlation length: λ = 0.2

Coarsest grid: h = 1/8 (comparable to λ)

Finest grid: h = 1/128

Karhunen-Loève truncation: mKL = 4000

Cost taken to be proportional to number of nodes

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 40 / 47
2D Results

2 2

0 0

−2 −2
log2 variance

log2 |mean|
−4 −4

−6 −6

−8 −8

P P
−10 l −10 l
Pl− Pl−1 Pl− Pl−1
−12 −12
0 1 2 3 4 0 1 2 3 4
level l level l

bℓ − P
V[P bℓ−1 ] ∼ h2 bℓ − P
E[P bℓ−1 ] ∼ h2
ℓ ℓ

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 41 / 47
2D Results

8
10
ε=0.0005 Std MC
7 ε=0.001 MLMC
10
ε=0.002
2
ε=0.005 10
6
10 ε=0.01

ε2 Cost
5
Nl

10

1
4 10
10

3
10

2 0
10 10
0 1 2 3 4 −3 −2
10 10
level l accuracy ε

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 42 / 47
Complexity analysis
Relating things back to the MLMC theorem:
bℓ −P] ∼ 2−2ℓ =⇒ α = 2
E[P
Vℓ ∼ 2−2ℓ =⇒ β = 2
Cℓ ∼ 2dℓ =⇒ γ = d (dimension of PDE)

To achieve r.m.s. accuracy ε requires finest level grid spacing h ∼ ε1/2

and hence we get the following complexity:

dim MC MLMC
1 ε−2.5 ε−2
2 ε−3 ε (log ε)2
−2

3 ε−3.5 ε−2.5

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 43 / 47
Other SPDE applications

For more on multilevel for SPDEs, see the work of Christoph Schwab and
his group (ETH Zurich):
http://www.math.ethz.ch/∼schwab/

elliptic, parabolic and hyperbolic PDEs

stochastic coefficients, initial data, boundary data

Schwab used to work on alternative techniques such as “polynomial chaos”

so I think the fact that he has now switched to multilevel is significant.

For other papers on multilevel, see my MLMC community homepage:

http://people.maths.ox.ac.uk/gilesm/mlmc community.html

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 44 / 47
Bio-chemical reactions

Another interesting application area is bio-chemical reactions.

Chemical reactions are usually modelled by ODEs:

X
ċs = λr (c) νr ,s
r

where λr is the rate for reaction r , and νr ,s is its effect on species s.

However, when concentrations are extremely small, the modelling is

stochastic, at the level of individual molecular reactions:
 
P reaction r occurs in time interval [t, t +dt] = λr (c) dt

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 45 / 47
Bio-chemical reactions
This leads to a hierarchy of models:
SSA (Stochastic Simulation Algorithm)
◮ models each individual reaction
τ -leaping method
◮ time-stepping with Poisson model for reactions in each time interval
Langevin equations
◮ an SDE with Brownian noise
◮ limit as Poisson distribution approaches Normal at high rates
“standard” ODEs
◮ limit as std. dev. of Normal distribution becoming negligible relative to
mean

Anderson (Wisconsin) & Higham (Strathclyde) have used multilevel for

SSA and τ -leaping, and obtained large computational savings.

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 46 / 47
Conclusions

Moving from CFD to Monte Carlo simulation, I have been fortunate

in being able to adapt ideas from CFD to new challenges:
◮ adjoints for sensitivity calculations – very natural
◮ multilevel Monte Carlo – not as natural because it is different from
multigrid, but a similar philosophy
The multilevel Monte Carlo development is now feeding back into
CFD for uncertainty quantification in applications such as nuclear
waste repositories and oil servoir simulation
I think stochastic modelling and simulation is an important growth
area in applied mathematics, engineering and science

Mike Giles (Oxford) CFD to finance (and back?) March 14th, 2013 47 / 47