Machine Learning For Fluid Mechanics

FL52CH19_Koumoutsakos ARjats.
cls December 7, 2019 16:29
Annual Review of Fluid Mechanics

Machine Learning for
Fluid Mechanics
Access provided by Peter the Great the St. Petersburg Polytechnic University on 03/18/21. For personal use only.
Steven L. Brunton,1 Bernd R. Noack,2,3

Annu. Rev. Fluid Mech. 2020.52:477-508. Downloaded from www.annualreviews.org
and Petros Koumoutsakos4

1
Department of Mechanical Engineering, University of Washington, Seattle,
Washington 98195, USA
2
LIMSI (Laboratoire d’Informatique pour la Mécanique et les Sciences de l’Ingénieur),
CNRS UPR 3251, Université Paris-Saclay, F-91403 Orsay, France
3
Institut für Strömungsmechanik und Technische Akustik, Technische Universität Berlin,
D-10634 Berlin, Germany
4
Computational Science and Engineering Laboratory, ETH Zurich, CH-8092 Zurich,
Switzerland; email: [email protected]
Annu. Rev. Fluid Mech. 2020. 52:477–508 Keywords

First published as a Review in Advance on
machine learning, data-driven modeling, optimization, control
September 12, 2019
The Annual Review of Fluid Mechanics is online at Abstract

fluid.annualreviews.org
The field of fluid mechanics is rapidly advancing, driven by unprecedented
https://doi.org/10.1146/annurev-fluid-010719-
volumes of data from experiments, field measurements, and large-scale sim-
060214
ulations at multiple spatiotemporal scales. Machine learning (ML) offers a
Copyright © 2020 by Annual Reviews.
wealth of techniques to extract information from data that can be trans-
All rights reserved
lated into knowledge about the underlying fluid mechanics. Moreover, ML
algorithms can augment domain knowledge and automate tasks related to
flow control and optimization. This article presents an overview of past his-
tory, current developments, and emerging opportunities of ML for fluid me-
chanics. We outline fundamental ML methodologies and discuss their uses
for understanding, modeling, optimizing, and controlling fluid flows. The
strengths and limitations of these methods are addressed from the perspec-
tive of scientific inquiry that considers data as an inherent part of model-
ing, experiments, and simulations. ML provides a powerful information-
processing framework that can augment, and possibly even transform,
current lines of fluid mechanics research and industrial applications.
477
FL52CH19_Koumoutsakos ARjats.cls December 7, 2019 16:29
1. INTRODUCTION
Fluid mechanics has traditionally dealt with massive amounts of data from experiments, field mea-
Machine learning: surements, and large-scale numerical simulations. Indeed, in the past few decades, big data have
algorithms that been a reality in fluid mechanics research (Pollard et al. 2016) due to high-performance comput-
process and extract ing architectures and advances in experimental measurement capabilities. Over the past 50 years,
information from data;
many techniques were developed to handle such data, ranging from advanced algorithms for data
they facilitate
automation of tasks processing and compression to fluid mechanics databases (Perlman et al. 2007, Wu & Moin 2008).
and augment human However, the analysis of fluid mechanics data has relied, to a large extent, on domain expertise,
domain knowledge statistical analysis, and heuristic algorithms.
Supervised learning: The growth of data today is widespread across scientific disciplines, and gaining insight and
learning from data actionable information from data has become a new mode of scientific inquiry as well as a com-
labeled with expert mercial opportunity. Our generation is experiencing an unprecedented confluence of (a) vast and
knowledge, providing increasing volumes of data; (b) advances in computational hardware and reduced costs for com-
corrective information
putation, data storage, and transfer; (c) sophisticated algorithms; (d) an abundance of open source
to the algorithm
software and benchmark problems; and (e) significant and ongoing investment by industry on
Semisupervised data-driven problem solving. These advances have, in turn, fueled renewed interest and progress
learning: learning
in the field of machine learning (ML) to extract information from these data. ML is now rapidly
with partially labeled
data (generative making inroads in fluid mechanics. These learning algorithms may be categorized into supervised,
adversarial networks) semisupervised, and unsupervised learning (see Figure 1), depending on the information available
or by interactions of about the data to the learning machine (LM).
the machine with its ML provides a modular and agile modeling framework that can be tailored to address many
environment
challenges in fluid mechanics, such as reduced-order modeling, experimental data processing,
(reinforcement
learning) shape optimization, turbulence closure modeling, and control. As scientific inquiry shifts from
first principles to data-driven approaches, we may draw a parallel with the development of nu-
Unsupervised
merical methods in the 1940s and 1950s to solve the equations of fluid dynamics. Fluid mechanics
learning: learning
without labeled stands to benefit from learning algorithms and in return presents challenges that may further
training data advance these algorithms to complement human understanding and engineering intuition.
Supervised Semisupervised Unsupervised
Optimization Reinforcement Generative Dimensionality

Classification Regression Clustering
and control learning models reduction
Support vector Linear Linear control Q-learning Generative k-means POD/PCA

machines Generalized linear Genetic Markov decision adversarial Spectral Autoencoder
Decision trees algorithms processes networks clustering
Gaussian process Self-organizing
Random forests Deep model Deep reinforce- maps
Neural networks predictive ment learning Diffusion maps
control
k-nearest
neighbor Estimation of
distribution
algorithms
Evolutionary
strategies
Figure 1
Machine learning algorithms may be categorized into supervised, unsupervised, and semisupervised, depending on the extent and type
of information available for the learning process. Abbreviations: PCA, principal component analysis; POD, proper orthogonal
decomposition.
478 Brunton • Noack • Koumoutsakos

In addition to outlining successes, we must note the importance of understanding how learning
algorithms work and when these methods succeed or fail. It is important to balance excitement
about the capabilities of ML with the reality that its application to fluid mechanics is an open
Reduced-order
and challenging field. In this context, we also highlight the benefit of incorporating domain model: representation
knowledge about fluid mechanics into learning algorithms. We envision that the fluid mechanics of a high-dimensional
community can contribute to advances in ML reminiscent of advances in numerical methods in system in terms of a
the last century. low-dimensional one,
balancing accuracy and
efficiency
1.1. Historical Overview Perceptron: the first
learning machine; a
The interface between ML and fluid dynamics has a long and possibly surprising history. In the
network of binary
early 1940s, Kolmogorov, a founder of statistical learning theory, considered turbulence as one of decision units used for
its prime application domains (Kolmogorov 1941). Advances in ML in the 1950s and 1960s were classification
characterized by two distinct developments. On one side, we may distinguish cybernetics (Wiener
1965) and expert systems designed to emulate the thinking process of the human brain, and on
the other side, machines such as the perceptron (Rosenblatt 1958) aimed to automate processes
such as classification and regression. The use of perceptrons for classification created significant
excitement. However, this excitement was quenched by findings that their capabilities had severe
limitations (Minsky & Papert 1969): Single-layer perceptrons were only able to learn linearly sep-
arable functions and were not capable of learning the XOR function. It was known that multilayer
perceptrons could learn the XOR function, but perhaps their advancement was limited given the
computational resources of the times (a recurring theme in ML research). The reduced inter-
est in perceptrons was soon accompanied by a reduced interest in artificial intelligence (AI) in
general.
Another branch of ML, closely related to budding ideas of cybernetics in the early 1960s, was
pioneered by two graduate students: Ingo Rechenberg and Hans-Paul Schwefel at the Technical
University of Berlin. They performed experiments in a wind tunnel on a corrugated structure
composed of five linked plates with the goal of finding their optimal angles to reduce drag (see
Figure 2). Their breakthrough involved adding random variations to these angles, where the
randomness was generated using a Galton board (an analog random number generator). Most
importantly, the size of the variance was learned (increased/decreased) based on the success rate
(positive/negative) of the experiments. Despite its brilliance, the work of Rechenberg and Schwefel
has received little recognition in the fluid mechanics community, even though a significant number
of applications in fluid mechanics and aerodynamics use ideas that can be traced back to their work.
Renewed interest in the potential of AI for aerodynamics applications materialized almost simulta-
neously with the early developments in computational fluid dynamics in the early 1980s. Attention
was given to expert systems to assist in aerodynamic design and development processes (Mehta &
Kutler 1984).
An indirect link between fluid mechanics and ML was the so-called Lighthill report in 1974
that criticized AI programs in the United Kingdom as not delivering on their grand claims. This
report played a major role in the reduced funding and interest in AI in the United Kingdom and
subsequently in the United States that is known as the AI winter. Lighthill’s main argument was
based on his perception that AI would never be able to address the challenge of the combinatorial
explosion between possible configurations in the parameter space. He used the limitations of lan-
guage processing systems of that time as a key demonstration of the failures for AI. In Lighthill’s
defense, 40 years ago the powers of modern computers as we know them today may have been
difficult to fathom. Indeed, today one may watch Lighthill’s speech on the internet while an ML
algorithm automatically provides the captions.
www.annualreviews.org • Machine Learning for Fluid Mechanics 479

Figure 2
First example of learning and automation in experimental fluid mechanics: Rechenberg’s (1964) experiments for optimally corrugated
plates for drag reduction using the Galtonbrett (Galton board) as an analog random number generator. Figure reprinted with
permission from Rechenberg (1973).
The reawakening of interest in ML, and in neural networks (NNs) in particular, came in the
late 1980s with the development of the backpropagation algorithm (Rumelhart et al. 1986). This
enabled the training of NNs with multiple layers, even though in the early days at most two layers
were the norm. Other sources of stimulus were the works by Hopfield (1982), Gardner (1988),
and Hinton & Sejnowski (1986), who developed links between ML algorithms and statistical me-
chanics. However, these developments did not attract many researchers from fluid mechanics. In
the early 1990s a number of applications of NNs in flow-related problems were developed in the
context of trajectory analysis and classification for particle tracking velocimetry (PTV) and par-
Neural network:
a computational ticle image velocimetry (PIV) (Teo et al. 1991, Grant & Pan 1995) as well as for identifying the
architecture, based phase configurations in multiphase flows (Bishop & James 1993). The link between proper or-
loosely on biological thogonal decomposition (POD) and linear NNs (Baldi & Hornik 1989) was exploited in order
networks of neurons, to reconstruct turbulence flow fields and the flow in the near-wall region of a channel flow using
for nonlinear
wall-only information (Milano & Koumoutsakos 2002). This application was one of the first to
regression
also use multiple layers of neurons to improve compression results, marking perhaps the first use
Deep learning: neural of deep learning, as it is known today, in the field of fluid mechanics.
networks with multiple
In the past few years, we have experienced a renewed blossoming of ML applications in fluid
layers; used to create
powerful hierarchical mechanics. Much of this interest is attributed to the remarkable performance of deep learning
representations at architectures, which hierarchically extract informative features from data. This has led to several
varying levels of advances in data-rich and model-limited fields such as the social sciences and in companies for
abstraction which prediction is a key financial factor. Fluid mechanics is not a model-limited field, and it is
rapidly becoming data rich. We believe that this confluence of first principles and data-driven
approaches is unique and has the potential to transform both fluid mechanics and ML.

1.2. Challenges and Opportunities for Machine Learning in Fluid Dynamics

Fluid dynamics presents challenges that differ from those tackled in many applications of ML,
such as image recognition and advertising. In fluid flows it is often important to precisely quantify Reinforcement
the underlying physical mechanisms in order to analyze them. Furthermore, fluid flows exhibit learning: an agent
complex, multiscale phenomena the understanding and control of which remain largely unre- learns a policy to
maximize its
solved. Unsteady flow fields require algorithms capable of addressing nonlinearities and multiple
long-term rewards by
spatiotemporal scales that may not be present in popular ML algorithms. In addition, many promi- interacting with its
nent applications of ML, such as playing the game Go, rely on inexpensive system evaluations and environment
an exhaustive categorization of the process that must be learned. This is not the case in fluids,
where experiments may be difficult to repeat or automate and where simulations may require
large-scale supercomputers operating for extended periods of time.
ML has also become instrumental in robotics, and algorithms such as reinforcement learning
(RL) are used routinely in autonomous driving and flight. While many robots operate in fluids,
it appears that the subtleties of fluid dynamics are not presently a major concern in their design.
Reminiscent of the pioneering days of flight, solutions imitating natural forms and processes are
often the norm (see the sidebar titled Learning Fluid Mechanics: From Living Organisms to Ma-
chines). We believe that deeper understanding and exploitation of fluid mechanics will become
critical in the design of robotic devices when their energy consumption and reliability in complex
flow environments become a concern.
In the context of flow control, actively or passively manipulating flow dynamics for an en-
gineering objective may change the nature of the system, making predictions based on data of
uncontrolled systems impossible. Although flow data are vast in some dimensions, such as spatial
resolution, they may be sparse in others; for example, it may be expensive to perform parametric
studies. Furthermore, flow data can be highly heterogeneous, requiring special care when choos-
ing the type of LM. In addition, many fluid systems are nonstationary, and even for stationary
flows it may be prohibitively expensive to obtain statistically converged results.
Fluid dynamics is central to transportation, health, and defense systems, and it is therefore
essential that ML solutions are interpretable, explainable, and generalizable. Moreover, it is
LEARNING FLUID MECHANICS: FROM LIVING ORGANISMS TO MACHINES

Birds, bats, insects, fish, whales, and other aquatic and aerial life-forms perform remarkable feats of fluid manip-
ulation, optimizing and controlling their shape and motion to harness unsteady fluid forces for agile propulsion,
efficient migration, and other exquisite maneuvers. The range of fluid flow optimization and control observed in
biology is breathtaking and has inspired humans for millennia. How do these organisms learn to manipulate the
flow environment?
To date, we know of only one species that manipulates fluids through knowledge of the Navier–Stokes equations.
Humans have been innovating and engineering devices to harness fluids since before the dawn of recorded history,
from dams and irrigation to mills and sailing. Early efforts were achieved through intuitive design, although recent
quantitative analysis and physics-based design have enabled a revolution in performance over the past hundred
years. Indeed, physics-based engineering of fluid systems is a high-water mark of human achievement. However,
there are serious challenges associated with equation-based analysis of fluids, including high dimensionality and
nonlinearity, which defy closed-form solutions and limit real-time optimization and control efforts. At the beginning
of a new millennium, with increasingly powerful tools in machine learning and data-driven optimization, we are
again learning how to learn from experience.

often necessary to provide guarantees on performance, which are presently rare. Indeed, there
is a poignant lack of convergence results, analysis, and guarantees in many ML algorithms. It is
also important to consider whether the model will be used for interpolation within a parameter
regime or for extrapolation, which is considerably more challenging. Finally, we emphasize the
importance of cross-validation on withheld data sets to prevent overfitting in ML.
We suggest that this nonexhaustive list of challenges need not be a barrier; to the contrary, it
should provide a strong motivation for the development of more effective ML techniques. These
techniques will likely impact several disciplines if they are able to solve fluid mechanics problems.
The application of ML to systems with known physics, such as fluid mechanics, may provide
deeper theoretical insights into algorithms. We also believe that hybrid methods that combine
ML and first principles models will be a fertile ground for development.
This review is structured as follows: Section 2 outlines the fundamental algorithms of ML,
followed by discussions of their applications to flow modeling (Section 3) and optimization and
control (Section 4). We provide a summary and outlook of this field in Section 5.
2. MACHINE LEARNING FUNDAMENTALS

The learning problem can be formulated as the process of estimating associations between inputs,
outputs, and parameters of a system using a limited number of observations (Cherkassky & Mulier
2007). We distinguish between a generator of samples, the system in question, and an LM, as in
Figure 3. We emphasize that the approximations by LMs are fundamentally stochastic, and their
learning process can be summarized as the minimization of a risk functional:

R(w) = L[y, φ(x, y, w)] p(x, y) dxdy, 1.
where the data x (inputs) and y (outputs) are samples from a probability distribution p, φ(x, y, w)
defines the structure and w the parameters of the LM, and the loss function L balances the var-
ious learning objectives (e.g., accuracy, simplicity, smoothness, etc.). We emphasize that the risk
functional is weighted by a probability distribution p(x, y) that also constrains the predictive ca-
pabilities of the LM. The various types of learning algorithms can be grouped into three major
categories: supervised, unsupervised, and semisupervised, as in Figure 1. These distinctions sig-
nify the degree to which external supervisory information from an expert is available to the LM.
Learning
Sample machine
generator Input vector, x Learning machine output, ŷ
Functional
Probability of form with
input, p(x)
weights w, φ(x,y,w)
System
Conditional System output, y
probability of
input, p(y|x)
Figure 3
The learning problem. A learning machine uses inputs from a sample generator and observations from a
system to generate an approximation of its output. Figure based on an idea from Cherkassky & Mulier
(2007).

2.1. Supervised Learning

Supervised learning implies the availability of corrective information to the LM. In its simplest
and most common form, this implies labeled training data, with labels corresponding to the output
of the LM. Minimization of the cost function, which depends on the training data, will determine
the unknown parameters of the LM. In this context, supervised learning dates back to the regres-
sion and interpolation methods proposed centuries ago by Gauss (Meijering 2002). A commonly
employed loss function is
L[y, φ(x, y, w)] = ||y − φ(x, y, w)||2 . 2.

Alternative loss functions may reflect different constraints on the LM such as sparsity (Hastie et al.
2009, Brunton & Kutz 2019). The choice of the approximation function reflects prior knowledge
about the data, and the choice between linear and nonlinear methods directly bears on the com-
putational cost associated with the learning methods.
2.1.1. Neural networks. NNs are arguably the most well-known methods in supervised learn-
ing. They are fundamental nonlinear function approximators, and in recent years several efforts
have been dedicated to understanding their effectiveness. The universal approximation theo-
rem (Hornik et al. 1989) states that any function may be approximated by a sufficiently large
and deep network. Recent work has shown that sparsely connected, deep NNs are information
theoretic–optimal nonlinear approximators for a wide range of functions and systems (Bölcskei
et al. 2019).
The power and flexibility of NNs emanate from their modular structure based on the neuron
as a central building element, a caricature of neurons in the human brain. Each neuron receives
an input, processes it through an activation function, and produces an output. Multiple neurons
can be combined into different structures that reflect knowledge about the problem and the type
of data. Feedforward networks are among the most common structures, and they are composed
of layers of neurons, where a weighted output from one layer is the input to the next layer. NN
architectures have an input layer that receives the data and an output layer that produces a pre-
diction. Nonlinear optimization methods, such as backpropagation (Rumelhart et al. 1986), are
used to identify the network weights to minimize the error between the prediction and labeled
training data. Deep NNs involve multiple layers and various types of nonlinear activation func-
tions. When the activation functions are expressed in terms of convolutional kernels, a powerful
class of networks emerges, namely convolutional neural networks (CNNs), with great success in
image and pattern recognition (Krizhevsky et al. 2012, Goodfellow et al. 2016), Grossberg et al.
1988).
Recurrent neural networks (RNNs), depicted in Figure 4, are of particular interest to fluid me-
chanics. They operate on sequences of data (e.g., images from a video, time series, etc.), and their
weights are obtained by backpropagation through time. RNNs have been quite successful for natu-
ral language processing and speech recognition. Their architecture takes into account the inherent
order of the data, thus augmenting some of the pioneering applications of classical NNs on signal
processing (Rico-Martinez et al. 1992). However, their effectiveness has been hindered by dimin-
ishing or exploding gradients that emerge during their training. The renewed interest in RNNs
is largely attributed to the development of the long short-term memory (LSTM) (Hochreiter &
Schmidhuber 1997) algorithms that deploy cell states and gating mechanisms to store and forget
information about past inputs, thus alleviating the problems with gradients and the transmission
of long-term information from which standard RNNs suffer. An extended architecture called the

ht–1 ht ht+1 ht+1 ht ht+1
ht–1 ht ct–1 ct
+
tanh
RNN tanh RNN LSTM LSTM
σ σ tanh
ht–1 ht
σ
x t–1 xt x t+1 x t–1 xt x t+1

Figure 4
Recurrent neural networks (RNNs) for time series predictions and the long short-term memory (LSTM) regularization. Abbreviations:
ct−1 , previous cell memory; ct , current cell memory; ht−1 , previous cell output; ht , current cell output; xt , input vector; σ , sigmoid.
Figure based on an idea from Hochreiter & Schmidhuber (1997).
multidimensional LSTM network (Graves et al. 2007) was proposed to efficiently handle high-
dimensional spatiotemporal data. Several potent alternatives to RNNs have appeared over the
years; the echo state network has been used with success in predicting the output of several dy-
namical systems (Pathak et al. 2018).
2.1.2. Classification: support vector machines and random forests. Classification is a su-
pervised learning task that can determine the label or category of a set of measurements from a
priori labeled training data. It is perhaps the oldest method for learning, starting with the per-
ceptron (Rosenblatt 1958), which could classify between two types of linearly separable data. Two
fundamental classification algorithms are support vector machines (SVMs) (Schölkopf & Smola
2002) and random forests (Breiman 2001), which have been widely adopted in industry. The prob-
lem can be specified by the following loss functional, which is expressed here for two classes:

0, if y = φ(x, y, w),
L[y, φ(x, y, w)] = 3.
1, if y = φ(x, y, w).
The output of the LM is an indicator of the class to which the data belong. The risk functional
quantifies the probability of misclassification, and the task is to minimize the risk based on the
training data by suitable choice of φ(x, y, w). Random forests are based on an ensemble of decision
trees that hierarchically split the data using simple conditional statements; these decisions are
interpretable and fast to evaluate at scale. In the context of classification, an SVM maps the data
into a high-dimensional feature space on which a linear classification is possible.
2.2. Unsupervised Learning

This learning task implies the extraction of features from the data by specifying certain global
criteria, without the need for supervision or a ground-truth label for the results. The types of
problems involved here include dimensionality reduction, quantization, and clustering.
2.2.1. Dimensionality reduction I: proper orthogonal decomposition, principal compo-

nent analysis, and autoencoders. The extraction of flow features from experimental data
and large-scale simulations is a cornerstone of flow modeling. Moreover, identifying lower-
dimensional representations for high-dimensional data can be used as preprocessing for all
tasks in supervised learning algorithms. Dimensionality reduction can also be viewed as an

PCA/POD Deep encoder Deep decoder

N
∑x
Encoder Decoder
x= 1 n U V φ(x) ψ(z)
N
n=1
N
S= 1
N ∑(x – x)(x – x)
n=1
n n
T
z
Sui = λi ui Latent
variables
Retain M < D
eigenvectors
x x̂ x
x̂
Input Output Input Output
Figure 5
PCA/POD (left) versus shallow autoencoders (center) and deep autoencoders (right). If the node activation functions in the shallow
autoencoder are linear, then U and V are matrices that minimize the loss function, x̂ − VUx. The node activation functions may be
nonlinear, minimizing the loss function, x − ψ[φ(x)]. The input x ∈ RD is reduced to z ∈ RM , with M D. Note that PCA/POD
requires the solution of a problem-specific eigenvalue equation, while the neuron modules can be extended to nonlinear activation
functions and multiple nodes and layers. Abbreviations: PCA, principal component analysis; POD, proper orthogonal decomposition;
S, covariance matrix of mean-subtracted data; U, linear encoder; ui , eigenvector; V, linear decoder; x, input vector; xn , n-th input
vector; x̄, mean of input data; x̂, autoencoder reconstruction; z, latent variable; λi , eigenvalue; φ(x), deep encoder; ψ(x), deep decoder.
Figure based on an idea from Bishop & James (1993).
information-filtering bottleneck where the data are processed through a lower-dimensional rep-
resentation before being mapped back to the ambient dimension. The classical POD algorithm
belongs to this category of learning and is discussed more in Section 3. The POD, or linear princi-
pal component analysis (PCA) as it is more widely known, can be formulated as a two-layer NN (an
autoencoder) with a linear activation function for its linearly weighted input, which can be trained
by stochastic gradient descent (see Figure 5). This formulation is an algorithmic alternative to lin-
ear eigenvalue/eigenvector problems in terms of NNs, and it offers a direct route to the nonlinear
regime and deep learning by adding more layers and a nonlinear activation function on the net-
work. Unsupervised learning algorithms have seen limited use in the fluid mechanics community,
and we believe that they deserve further exploration. In recent years, the ML community has devel-
oped numerous autoencoders that, when properly matched with the possible features of the flow
field, can lead to significant insight for reduced-order modeling of stationary and time-dependent
data.
2.2.2. Dimensionality reduction II: discrete principal curves and self-organizing maps.
The mapping between high-dimensional data and a low-dimensional representation can be struc-
Autoencoder:
tured through an explicit shaping of the lower-dimensional space, possibly reflecting an a priori
a neural network
knowledge about this subspace. These techniques can be seen as extensions of the linear autoen- architecture used to
coders, where the encoder and decoder can be nonlinear functions. This nonlinearity may, how- compress and
ever, come at the expense of losing the inverse relationship between the encoder and decoder decompress
functions that is one of the strengths of linear PCA. An alternative is to define the decoder as an high-dimensional data;
linear and nonlinear
approximation of the inverse of the encoder, leading to the method of principal curves. Principal
alternative to the
curves are structures on which the data are projected during the encoding step of the learning proper orthogonal
algorithm. In turn, the decoding step amounts to an approximation of the inverse of this mapping decomposition
by adding, for example, some smoothing onto the principal curves. An important version of this
process is the self-organizing map (SOM) introduced by Grossberg (1976) and Kohonen (1995).
In SOMs the projection subspace is described into a finite set of values with specified connectivity

architecture and distance metrics. The encoder step amounts to identifying for each data point the
closest node point on the SOM, and the decoder step is a weighted regression estimate using, for
example, kernel functions that take advantage of the specified distance metric between the map
nodes. This modifies the node centers, and the process can be iterated until the empirical risk of
the autoencoder has been minimized. The SOM capabilities can be exemplified by comparing it to
linear PCA for a two-dimensional set of points. The linear PCA will provide as an approximation
the least squares straight line between the points, whereas the SOM will map the points onto a
curved line that better approximates the data. We note that SOMs can be extended to areas be-
yond floating point data, and they offer an interesting way for creating databases based on features
of flow fields.
2.2.3. Clustering and vector quantization. Clustering is an unsupervised learning technique

that identifies similar groups in the data. The most common algorithm is k-means clustering,
which partitions data into k clusters; an observation belongs to the cluster with the nearest cen-
troid, resulting in a partition of data space into Voronoi cells.
Vector quantizers identify representative points for data that can be partitioned into a prede-
termined number of clusters. These points can then be used instead of the full data set so that
future samples can be approximated by them. The vector quantizer φ(x, w) provides a mapping
between the data x and the coordinates of the cluster centers. The loss function is usually the
squared distortion of the data from the cluster centers, which must be minimized to identify the
parameters of the quantizer,
L[φ(x, w)] = ||x − φ(x, w)||2 . 4.
We note that vector quantization is a data reduction method not necessarily employed for dimen-
sionality reduction. In the latter, the learning problem seeks to identify low-dimensional features
in high-dimensional data, whereas quantization amounts to finding representative clusters of the
data. Vector quantization must also be distinguished from clustering, as in the former the number
of desired centers is determined a priori, whereas clustering aims to identify meaningful group-
ings in the data. When these groupings are represented by some prototypes, then clustering and
quantization have strong similarities.
2.3. Semisupervised Learning

Semisupervised learning algorithms operate under partial supervision, either with limited la-
beled training data or with other corrective information from the environment. Two algorithms
in this category are generative adversarial networks (GANs) and RL. In both cases, the LM is
(self-)trained through a game-like process discussed below.
2.3.1. Generative adversarial networks. GANs are learning algorithms that result in a gener-
ative model, i.e., a model that produces data according to a probability distribution that mimics
that of the data used for its training. The LM is composed of two networks that compete with each
other in a zero-sum game (Goodfellow et al. 2014). The generative network produces candidate
data examples that are evaluated by the discriminative, or critic, network to optimize a certain
task. The generative network’s training objective is to synthesize novel examples of data to fool
the discriminative network into misclassifying them as belonging to the true data distribution.

The weights of these networks are obtained through a process, inspired by game theory, called
adversarial learning. The final objective of the GAN training process is to identify the genera-
tive model that produces an output that reflects the underlying system. Labeled data are provided
by the discriminator network, and the function to be minimized is the Kullback–Liebler diver-
gence between the two distributions. In the ensuing game, the discriminator aims to maximize
the probability of discriminating between true data and data produced by the generator, while the
generator aims to minimize the same probability. Because the generative and discriminative net-
works essentially train themselves, after initialization with labeled training data, this procedure
is often called self-supervised. This self-training process adds to the appeal of GANs, but at the
same time one must be cautious about whether an equilibrium will ever be reached in the above-
mentioned game. As with other training algorithms, large amounts of data help the process, but
at the moment, there is no guarantee of convergence.
2.3.2. Reinforcement learning. RL is a mathematical framework for problem solving (Sutton

& Barto 2018) that implies goal-directed interactions of an agent with its environment. In RL the
agent has a repertoire of actions and perceives states. Unlike in supervised learning, the agent does
not have labeled information about the correct actions but instead learns from its own experiences
in the form of rewards that may be infrequent and partial; thus, this is termed semisupervised
learning. Moreover, the agent is concerned not only with uncovering patterns in its actions or in
the environment but also with maximizing its long-term rewards. RL is closely linked to dynamic
programming (Bellman 1952), as it also models interactions with the environment as a Markov
decision process. Unlike dynamic programming, RL does not require a model of the dynamics,
such as a Markov transition model, but proceeds by repeated interaction with the environment
through trial and error. We believe that it is precisely this approximation that makes it highly
suitable for complex problems in fluid dynamics. The two central elements of RL are the agent’s
policy, a mapping a = π (s) between the state s of the system and the optimal action a, and the
value function V (s) that represents the utility of reaching the state s for maximizing the agent’s
long-term rewards.
Games are one of the key applications of RL that exemplify its strengths and limitations. One of
the early successes of RL is the backgammon learner of Tesauro (1992). The program started out
from scratch as a novice player, trained by playing a couple of million times against itself, won the
computer backgammon olympiad, and eventually became comparable to the three best human
players in the world. In recent years, advances in high-performance computing and deep NN
architectures have produced agents that are capable of performing at or above human performance
at video games and strategy games much more complicated than backgammon, such as Go (Mnih
et al. 2015) and the AI gym (Mnih et al. 2015, Silver et al. 2016). It is important to emphasize that
RL requires significant computational resources due to the large numbers of episodes required to
properly account for the interaction of the agent and the environment. This cost may be trivial
for games, but it may be prohibitive in experiments and flow simulations, a situation that is rapidly
changing (Verma et al. 2018).
A core remaining challenge for RL is the long-term credit assignment (LTCA) problem, espe-
cially when rewards are sparse or very delayed in time (for example, consider the case of a perching
bird or robot). LTCA implies inference, from a long sequence of states and actions, of causal rela-
tions between individual decisions and rewards. Several efforts address these issues by augmenting
the original sparsely rewarded objective with densely rewarded subgoals (Schaul et al. 2015). A re-
lated issue is the proper accounting of past experience by the agent as it actively forms a new policy
(Novati et al. 2019).

2.4. Stochastic Optimization: A Learning Algorithms Perspective

Optimization is an inherent part of learning, as a risk functional is minimized in order to identify
the parameters of the LM. There is, however, one more link that we wish to highlight in this re-
view: that optimization (and search) algorithms can be cast in the context of learning algorithms
and more specifically as the process of learning a probability distribution that contains the design
points that maximize a certain objective. This connection was pioneered by Rechenberg (1973)
and Schwefel (1977), who introduced evolution strategies (ES) and adapted the variance of their
search space based on the success rate of their experiments. This process is also reminiscent of the
operations of selection and mutation that are key ingredients of genetic algorithms (GAs) (Holland
1975) and genetic programming (Koza 1992). ES and GAs can be considered as hybrids between
gradient search strategies, which may effectively march downhill toward a minimum, and Latin
hypercube or Monte Carlo sampling methods, which maximally explore the search space. Genetic
programming was developed in the late 1980s by J.R. Koza, a PhD student of John Holland. Ge-
netic programming generalized parameter optimization to function optimization, initially coded

as a tree of operations (Koza 1992). A critical aspect of these algorithms is that they rely on an it-
erative construction of the probability distribution, based on data values of the objective function.
This iterative construction can be lengthy and practically impossible for problems with expensive
objective function evaluations.
Over the past 20 years, ES and GAs have begun to converge into estimation of distribution
algorithms (EDAs). The covariance matrix adaptation ES (CMA-ES) algorithm (Ostermeier et al.
1994, Hansen et al. 2003) is a prominent example of ES using an adaptive estimation of the co-
variance matrix of a Gaussian probability distribution to guide the search for optimal parameters.
This covariance matrix is adapted iteratively using the best points in each iteration. The CMA-
ES is closely related to several other algorithms, such as mixed Bayesian optimization algorithms
(Pelikan et al. 2004), and the reader is referred to Kern et al. (2004) for a comparative review. In
recent years, this line of work has evolved into the more generalized information-geometric opti-
mization (IGO) framework (Ollivier et al. 2017). IGO algorithms allow for families of probability
distributions whose parameters are learned during the optimization process and maintain the cost
function invariance as a major design principle. The resulting algorithm makes no assumption
on the objective function to be optimized, and its flow is equivalent to a stochastic gradient de-
scent. These techniques have proven to be effective on several simplified benchmark problems;
however, their scaling remains unclear, and there are few guarantees for convergence in cost func-
tion landscapes such as those encountered in complex fluid dynamics problems. We note also that
there is an interest in deploying these optimization methods to minimize the cost functions often
associated with classical ML tasks (Salimans et al. 2017).
2.5. Important Topics We Have Not Covered: Bayesian Inference

and Gaussian Processes
There are several learning algorithms that this review does not address but that demand particular
attention from the fluid mechanics community. First and foremost, we wish to mention Bayesian
inference, which aims to inform the model structure and its parameters from data in a proba-
bilistic framework. Bayesian inference is fundamental for uncertainty quantification, and it is also
fundamentally a learning method, as data are used to adapt the models. In fact, the alternative
view is also possible, where every ML framework can be cast in a Bayesian framework (Barber
2012, Theodoridis 2015). The optimization algorithms outlined in this review provide a direct
link. Whereas optimization algorithms aim to provide the best parameters of a model for given
data in a stochastic manner, Bayesian inference aims to provide the full probability distribution. It
may be argued that Bayesian inference may be even more powerful than ML, as it provides proba-
bility distributions for all parameters, leading to robust predictions, rather than single values, as is
usually the case with classical ML algorithms. However, a key drawback for Bayesian inference is
its computational cost, as it involves sampling and integration in high-dimensional spaces, which
can be prohibitive for expensive function evaluations (e.g., wind tunnel experiments or large-scale
direct numerical simulation). Along the same lines, one must mention Gaussian processes (GPs),
which resemble kernel-based methods for regression. However, GPs develop these kernels adap-
tively based on the available data. They also provide probability distributions for the respective
model parameters. GPs have been used extensively in problems related to time-dependent prob-
lems, and they may be considered competitors, albeit more costly, to RNNs. Finally, we note
the use of GPs as surrogates for expensive cost functions in optimization problems using ES and
GAs.
3. FLOW MODELING WITH MACHINE LEARNING

First principles, such as conservation laws, have been the dominant building blocks for flow mod-
eling over the past centuries. However, for high Reynolds numbers, scale-resolving simulations
using the most prominent model in fluid mechanics, the Navier–Stokes equations, are beyond our
current computational resources. An alternative is to perform simulations based on approxima-
tions of these equations (as often practiced in turbulence modeling) or laboratory experiments for
a specific configuration. However, simulations and experiments are expensive for iterative opti-
mization, and simulations are often too slow for real-time control (Brunton & Noack 2015). Con-
sequently, considerable effort has been placed on obtaining accurate and efficient reduced-order
models that capture essential flow mechanisms at a fraction of the cost (Rowley & Dawson 2016).
ML provides new avenues for dimensionality reduction and reduced-order modeling in fluid me-
chanics by providing a concise framework that complements and extends existing methodologies.
We distinguish here two complementary efforts: dimensionality reduction and reduced-order
modeling. Dimensionality reduction involves extracting key features and dominant patterns that
may be used as reduced coordinates where the fluid is compactly and efficiently described (Taira
et al. 2017). Reduced-order modeling describes the spatiotemporal evolution of the flow as a
parametrized dynamical system, although it may also involve developing a statistical map from
parameters to averaged quantities, such as drag.
There have been significant efforts to identify coordinate transformations and reductions that
simplify dynamics and capture essential flow physics; the POD is a notable example (Lumley 1970).
Model reduction, such as Galerkin projection of the Navier–Stokes equations onto an orthogonal
basis of POD modes, benefits from a close connection to the governing equations; however, it is
intrusive, requiring human expertise to develop models from a working simulation. ML provides
modular algorithms that may be used for data-driven system identification and modeling. Unique
aspects of data-driven modeling of fluid flows include the availability of partial prior knowledge of
the governing equations, constraints, and symmetries. With advances in simulation capabilities and
experimental techniques, fluid dynamics is becoming a data-rich field, thus becoming amenable
to ML algorithms.
In this review, we distinguish ML algorithms to model flow (a) kinematics through the extrac-
tion flow features and (b) dynamics through the adoption of various learning architectures.
3.1. Flow Feature Extraction

Pattern recognition and data mining are core strengths of ML. Many techniques have been devel-
oped by the ML community that are readily applicable to spatiotemporal fluid data. We discuss
linear and nonlinear dimensionality reduction techniques, followed by clustering and classifica-
tion. We also consider accelerated measurement and computation strategies, as well as methods
to process experimental flow field data.
3.1.1. Dimensionality reduction: linear and nonlinear embeddings. A common approach

in fluid dynamics simulation and modeling is to define an orthogonal linear transformation from
physical coordinates into a modal basis. The POD provides such an orthogonal basis for com-
plex geometries based on empirical measurements. Sirovich (1987) introduced the snapshot POD,
which reduces the computation to a simple data-driven procedure involving a singular value de-
composition. Interestingly, in the same year, Sirovich used POD to generate a low-dimensional
feature space for the classification of human faces, which is a foundation for much of modern
computer vision (Sirovich & Kirby 1987).
POD is closely related to the algorithm of PCA, one of the fundamental algorithms of applied
statistics and ML, to describe correlations in high-dimensional data. We recall that the PCA can be
expressed as a two-layer neural network, called an autoencoder, to compress high-dimensional data
for a compact representation, as shown in Figure 5. This network embeds high-dimensional data
into a low-dimensional latent space and then decodes from the latent space back to the original
high-dimensional space. When the network nodes are linear and the encoder and decoder are
constrained to be transposes of one another, the autoencoder is closely related to the standard
POD/PCA decomposition (Baldi & Hornik 1989) (see also Figure 6). However, the structure
of the NN autoencoder is modular, and by using nonlinear activation units for the nodes, it is
possible to develop nonlinear embeddings, potentially providing more compact coordinates. This
observation led to the development of one of the first applications of deep NNs to reconstruct
the near-wall velocity field in a turbulent channel flow using wall pressure and shear (Milano &
Koumoutsakos 2002). More powerful autoencoders are available today in the ML community, and
this link deserves further exploration.
On the basis of the universal approximation theorem (Hornik et al. 1989), which states that
a sufficiently large NN can represent an arbitrarily complex input–output function, deep NNs
Flow snapshots
POD modes
Autoencoder modes
Figure 6
Unsupervised learning example: merging of two vortices (top), proper orthogonal decomposition (POD) modes (middle), and respective
modes from a linear autoencoder (bottom). Note that unlike POD modes, the autoencoder modes are not orthogonal and are not
ordered.

are increasingly used to obtain more effective nonlinear coordinates for complex flows. How-
ever, deep learning often implies the availability of large volumes of training data that far exceed
the parameters of the network. The resulting models are usually good for interpolation but may
Interpretability:
not be suitable for extrapolation when the new input data have different probability distributions the degree to which a
than the training data (see Equation 1). In many modern ML applications, such as image clas- model may be
sification, the training data are so vast that it is natural to expect that most future classification understood or
tasks will fall within an interpolation of the training data. For example, the ImageNet data set in interpreted by an
expert human
2012 (Krizhevsky et al. 2012) contained over 15 million labeled images, which sparked the current
movement in deep learning (LeCun et al. 2015). Despite the abundance of data from experiments
and simulations, the fluid mechanics community is still distanced from this working paradigm.
However, it may be possible in the coming years to curate large, labeled, and complete-enough
fluid databases to facilitate the deployment of such deep learning algorithms.
3.1.2. Clustering and classification. Clustering and classification are cornerstones of ML.
There are dozens of mature algorithms to choose from, depending on the size of the data and the
desired number of categories. The k-means algorithm has been successfully employed by Kaiser
et al. (2014) to develop a data-driven discretization of a high-dimensional phase space for the fluid
mixing layer. This low-dimensional representation, in terms of a small number of clusters, enabled
tractable Markov transition models of how the flow evolves in time from one state to another. Be-
cause the cluster centroids exist in the data space, it is possible to associate each cluster centroid
with a physical flow field, lending additional interpretability. Amsallem et al. (2012) used k-means
clustering to partition phase space into separate regions, in which local reduced-order bases were
constructed, resulting in improved stability and robustness to parameter variations.
Classification is also widely used in fluid dynamics to distinguish between various canonical
behaviors and dynamic regimes. Classification is a supervised learning approach where labeled
data are used to develop a model to sort new data into one of several categories. Recently, Colvert
et al. (2018) investigated the classification of wake topology (e.g., 2S, 2P + 2S, 2P + 4S) behind
a pitching airfoil from local vorticity measurements using NNs; extensions have compared per-
formance for various types of sensors (Alsalman et al. 2018). Wang & Hemati (2017) used the
k-nearest-neighbors algorithm to detect exotic wakes. Similarly, NNs have been combined with
dynamical systems models to detect flow disturbances and estimate their parameters (Hou et al.
2019). Related graph and network approaches in fluids by Nair & Taira (2015) have been used for
community detection in wake flows (Meena et al. 2018). Finally, one of the earliest examples of ML
classification in fluid dynamics by Bright et al. (2013) was based on sparse representation (Wright
et al. 2009).
3.1.3. Sparse and randomized methods. In parallel to ML, there have been great strides in
sparse optimization and randomized linear algebra. ML and sparse algorithms are synergistic in
that underlying low-dimensional representations facilitate sparse measurements (Manohar et al.
2018) and fast randomized computations (Halko et al. 2011). Decreasing the amount of data to
train and execute a model is important when a fast decision is required, as in control. In this
context, algorithms for the efficient acquisition and reconstruction of sparse signals, such as com-
pressed sensing (Donoho 2006), have already been leveraged for compact representations of wall-
bounded turbulence (Bourguignon et al. 2014) and for POD-based flow reconstruction (Bai et al.
2014).
Low-dimensional structure in data also facilitates accelerated computations via randomized
linear algebra (Halko et al. 2011, Mahoney 2011). If a matrix has low-rank structure, then there
are efficient matrix decomposition algorithms based on random sampling; this is closely related

to the idea of sparsity and the high-dimensional geometry of sparse vectors. The basic idea is
that if a large matrix has low-dimensional structure, then with high probability this structure
will be preserved after projecting the columns or rows onto a random low-dimensional subspace,
Generalizability:
the ability of a model facilitating efficient downstream computations. These so-called randomized numerical methods
to generalize to new have the potential to transform computational linear algebra, providing accurate matrix decom-
examples including positions at a fraction of the cost of deterministic methods. For example, randomized linear
unseen data; Newton’s algebra may be used to efficiently compute the singular value decomposition, which is used to
second law, F = ma, is
compute PCA (Rokhlin et al. 2009, Halko et al. 2011).
highly generalizable
3.1.4. Superresolution and flow cleansing. Much of ML is focused on imaging science, pro-
viding robust approaches to improve resolution and remove noise and corruption based on sta-
tistical inference. These superresolution and denoising algorithms have the potential to improve
the quality of both simulations and experiments in fluids.
Superresolution involves the inference of a high-resolution image from low-resolution mea-

surements, leveraging the statistical structure of high-resolution training data. Several approaches
have been developed for superresolution, for example, based on a library of examples (Freeman
et al. 2002), sparse representation in a library (Yang et al. 2010), and most recently CNNs (Dong
et al. 2014). Experimental flow field measurements from PIV (Adrian 1991, Willert & Gharib
1991) provide a compelling application where there is a tension between local flow resolution and
the size of the imaging domain. Superresolution could leverage expensive and high-resolution data
on smaller domains to improve the resolution on a larger imaging domain. Large-eddy simulations
(LES) (Germano et al. 1991, Meneveau & Katz 2000) may also benefit from superresolution to in-
fer the high-resolution structure inside a low-resolution cell that is required to compute boundary
conditions. Recently, Fukami et al. (2018) developed a CNN-based superresolution algorithm and
demonstrated its effectiveness on turbulent flow reconstruction, showing that the energy spectrum
is accurately preserved. One drawback of superresolution is that it is often extremely costly com-
putationally, making it useful for applications where high-resolution imaging may be prohibitively
expensive; however, improved NN-based approaches may drive the cost down significantly. We
note also that Xie et al. (2018) recently employed GANs for superresolution.
The processing of experimental PIV and particle tracking has also been one of the first applica-
tions of ML. NNs have been used for fast PIV (Knaak et al. 1997) and PTV (Labonté 1999), with
impressive demonstrations for three-dimensional Lagrangian particle tracking (Ouellette et al.
2006). More recently, deep CNNs have been used to construct velocity fields from PIV image
pairs (Lee et al. 2017). Related approaches have also been used to detect spurious vectors in PIV
data (Liang et al. 2003) to remove outliers and fill in corrupt pixels.
3.2. Modeling Flow Dynamics

A central goal of modeling is to balance efficiency and accuracy. When modeling physical systems,
interpretability and generalizability are also critical considerations.
3.2.1. Linear models through nonlinear embeddings: dynamic mode decomposition and
Koopman analysis. Many classical techniques in system identification may be considered ML,
as they are data-driven models that generalize beyond the training data. Dynamic mode decompo-
sition (DMD) (Schmid 2010, Kutz et al. 2016) is a modern approach to extract spatiotemporal co-
herent structures from time series data of fluid flows, resulting in a low-dimensional linear model
for the evolution of these dominant coherent structures. DMD is based on data-driven regression
and is equally valid for time-resolved experimental and numerical data. DMD is closely related to

the Koopman operator (Rowley et al. 2009, Mezic 2013), which is an infinite-dimensional linear
operator that describes how all measurement functions of the system evolve in time. Because the
DMD algorithm is based on linear flow field measurements (i.e., direct measurements of the fluid
velocity or vorticity field), the resulting models may not be able to capture nonlinear transients.
Recently, there has been a concerted effort to identify a coordinate system where the nonlinear
dynamics appears linear. The extended DMD (Williams et al. 2015) and variational approach of
conformation dynamics (Noé & Nüske 2013, Nüske et al. 2016) enrich the model with nonlinear
measurements, leveraging kernel methods (Williams et al. 2015) and dictionary learning (Li et al.
2017). These special nonlinear measurements are generally challenging to represent, and deep
learning architectures are now used to identify nonlinear Koopman coordinate systems where
the dynamics appear linear (Takeishi et al. 2017, Lusch et al. 2018, Mardt et al. 2018, Wehmeyer
& Noé 2018). The VAMPnet architecture (Mardt et al. 2018, Wehmeyer & Noé 2018) uses a
time-lagged autoencoder and a custom variational score to identify Koopman coordinates on an
impressive protein folding example. Based on the performance of VAMPnet, fluid dynamics may
benefit from neighboring fields, such as molecular dynamics, which have similar modeling issues,
including stochasticity, coarse-grained dynamics, and separation of timescales.
3.2.2. Neural network modeling. Over the last three decades, NNs have been used to model
dynamical systems and fluid mechanics problems. Early examples include the use of NNs to learn
the solutions of ordinary and partial differential equations (Dissanayake & Phan-Thien 1994,
Gonzalez-Garcia et al. 1998, Lagaris et al. 1998). We note that the potential of these works has
not been fully explored, and in recent years there have been further advances (Chen et al. 2018,
Raissi & Karniadakis 2018), including discrete and continuous-in-time networks. We note also
the possibility of using these methods to uncover latent variables and reduce the number of para-
metric studies often associated with partial differential equations (Raissi et al. 2019). NNs are also
frequently employed in nonlinear system identification techniques such as NARMAX, which are
often used to model fluid systems (Glaz et al. 2010, Semeraro et al. 2016). In fluid mechanics, NNs
were widely used to model heat transfer ( Jambunathan et al. 1996), turbomachinery (Pierret &
Van den Braembussche 1999), turbulent flows (Milano & Koumoutsakos 2002), and other prob-
lems in aeronautics (Faller & Schreck 1996).
RNNs with LSTMs (Hochreiter & Schmidhuber 1997) have been revolutionary for speech
recognition, and they are considered one of the landmark successes of AI. They are currently be-
ing used to model dynamical systems and for data-driven predictions of extreme events (Vlachas
et al. 2018, Wan et al. 2018). An interesting finding of these studies is that combining data-driven
and reduced-order models is a potent method that outperforms each of its components on several
studies. GANs (Goodfellow et al. 2014) are also being used to capture physics (Wu et al. 2018).
GANs have potential to aid in the modeling and simulation of turbulence (Kim et al. 2018), al-
though this field is nascent.
Despite the promise and widespread use of NNs in dynamical systems, several challenges re-
main. NNs are fundamentally interpolative, and so the function is well approximated only in the
span (or under the probability distribution) of the sampled data used to train them. Thus, caution
should be exercised when using NN models for an extrapolation task. In many computer vision
and speech recognition examples, the training data are so vast that nearly all future tasks may
be viewed as an interpolation on the training data, although this scale of training has not been
achieved to date in fluid mechanics. Similarly, NN models are prone to overfitting, and care must
be taken to cross-validate models on a sufficiently chosen test set; best practices are discussed by
Goodfellow et al. (2016). Finally, it is important to explicitly incorporate partially known physics,
such as symmetries, constraints, and conserved quantities.

3.2.3. Parsimonious nonlinear models. Parsimony is a recurring theme in mathematical

physics, from Hamilton’s principle of least action to the apparent simplicity of many governing
equations. In contrast to the raw representational power of NNs, ML algorithms are also being
employed to identify minimal models that balance predictive accuracy with model complexity, pre-
venting overfitting and promoting interpretability and generalizability. Genetic programming was
recently used to discover conservation laws and governing equations (Schmidt & Lipson 2009).
Sparse regression in a library of candidate models has also been proposed to identify dynami-
cal systems (Brunton et al. 2016) and partial differential equations (Rudy et al. 2017, Schaeffer
2017). Loiseau & Brunton (2018) identified sparse reduced-order models of several flow systems,
enforcing energy conservation as a constraint. In both genetic programming and sparse identi-
fication, a Pareto analysis is used to identify models that have the best trade-off between model
complexity, measured in number of terms, and predictive accuracy. In cases where the physics is
known, this approach typically discovers the correct governing equations, providing exceptional
generalizability compared with other leading algorithms in ML.
3.2.4. Closure models with machine learning. The use of ML to develop turbulence closures
is an active area of research (Duraisamy et al. 2019). The extreme separation of spatiotemporal
scales in turbulent flows makes it exceedingly costly to resolve all scales in simulation, and even
with Moore’s law, we are decades away from resolving all scales in relevant configurations (e.g.,
aircraft, submarines, etc.). It is common to truncate small scales and model their effect on the
large scales with a closure model. Common approaches include Reynolds-averaged Navier–Stokes
(RANS) and LES. However, these models may require careful tuning to match data from fully
resolved simulations or experiments.
ML has been used to identify and model discrepancies in the Reynolds stress tensor between a
RANS model and high-fidelity simulations (Ling & Templeton 2015, Parish & Duraisamy 2016,
Ling et al. 2016b, Xiao et al. 2016, Singh et al. 2017, Wang et al. 2017). Ling & Templeton (2015)
compared SVMs, Adaboost decision trees, and random forests to classify and predict regions of
high uncertainty in the Reynolds stress tensor. Wang et al. (2017) used random forests to build
a supervised model for the discrepancy in the Reynolds stress tensor. Xiao et al. (2016) lever-
aged sparse online velocity measurements in a Bayesian framework to infer these discrepancies. In
related work, Parish & Duraisamy (2016) developed the field inversion and ML modeling frame-
work that builds corrective models based on inverse modeling. This framework was later used by
Singh et al. (2017) to develop an NN enhanced correction to the Spalart–Allmaras RANS model,
with excellent performance. A key result by Ling et al. (2016b) employed the first deep network
architecture with many hidden layers to model the anisotropic Reynolds stress tensor, as shown
in Figure 7. Their novel architecture incorporates a multiplicative layer to embed Galilean in-
variance into the tensor predictions. This provides an innovative and simple approach to embed
known physical symmetries and invariances into the learning architecture (Ling et al. 2016a),
which we believe will be essential in future efforts that combine learning for physics. For LES
closures, Maulik et al. (2019) have employed artificial NNs to predict the turbulence source term
from coarsely resolved quantities.
3.2.5. Challenges of machine learning for dynamical systems. Applying ML to model phys-
ical dynamical systems poses several unique challenges and opportunities. Model interpretability
and generalizability are essential cornerstones in physics. A well-crafted model will yield hypothe-
ses for phenomena that have not been observed before. This principle is for example exhibited in
the parsimonious formulation of classical mechanics in Newton’s second law.

Tensor input
layer
T (n)
Merge output
layer
b
a b
Input Invariant
layer Output input layer Final
layer λ1, ..., λ 5 hidden
Hidden layers Hidden layers layer
g (n)
Figure 7
Comparison of standard neural network architecture (a) with modified neural network for identifying Galilean invariant Reynolds
stress models (b). Abbreviations: b, anisotropy tensor; g(n) , scalar coefficients weighing the basis tensors; T(n) , isotropic basis tensors;
λ1 , . . . , λ5 , five tensor invariants. Figure adapted with permission from Ling et al. (2016b).
High-dimensional systems, such as those encountered in unsteady fluid dynamics, have the
challenges of multiscale dynamics, sensitivity to noise and disturbances, latent variables, and tran-
sients, all of which require careful attention when applying ML techniques. In ML for dynamics,
we distinguish two tasks: discovering unknown physics and improving models by incorporating
known physics. Many learning architectures cannot readily incorporate physical constraints in
the form of symmetries, boundary conditions, and global conservation laws. This is a critical area
for continued development, and several recent works have presented generalizable physics mod-
els (Battaglia et al. 2018).
4. FLOW OPTIMIZATION AND CONTROL USING

MACHINE LEARNING
Learning algorithms are well suited to flow optimization and control problems involving black-
box or multimodal cost functions. These algorithms are iterative and often require several orders
of magnitude more cost function evaluations than gradient-based algorithms (Bewley et al. 2001).
Moreover, they do not offer guarantees of convergence, and we suggest that they be avoided when
techniques such as adjoint methods are applicable. At the same time, techniques such as RL have
been shown to outperform even optimal flow control strategies (Novati et al. 2019). Indeed, there
are several classes of flow control and optimization problems where learning algorithms may be
the method of choice, as described below.

OPTIMIZATION AND CONTROL: BOUNDARIES ERASED BY FAST COMPUTERS

Optimization and control are intimately related, and the boundaries are becoming even less distinct with increas-
ingly fast computers, as summarized by Tsiotras & Mesbahi (2017, p. 195):
Interestingly, the distinction between optimization and control is largely semantic and (alas!) implementation-dependent.
If one has the capability of solving optimization problems fast enough on the fly to close the loop, then one has (in principle)
a feedback control law... Not surprisingly then, the same algorithm can be viewed as solving an optimization or a control
problem, based solely on the capabilities of the available hardware. With the continued advent of faster and more capable
computer hardware architectures, the boundary between optimization and control will become even more blurred. How-
ever, when optimization is embedded in the implementation of feedback control, the classical problems of control such
as robustness to model uncertainty, time delays, and process and measurement noise become of paramount importance,
particularly for high-performance aerospace systems.
In contrast to flow modeling, learning algorithms for optimization and control interact with the
data sampling process in several ways. First, in line with the modeling efforts described in earlier
sections, ML can be applied to develop explicit surrogate models that relate the cost function and
the control/optimization parameters. Surrogate models such as NNs can then be amenable to even
gradient-based methods, although they often get stuck in local minima. Multifidelity algorithms
(Perdikaris et al. 2016) can also be employed to combine surrogates with the cost function of the
complete problem. As the learning progresses, new data are requested as guided by the results of
the optimization. Alternatively, the optimization or control problem may be described in terms of
learning probability distributions of parameters that minimize the cost function. These probability
distributions are constructed from cost function samples obtained during the optimization process.
Furthermore, the high-dimensional and nonconvex optimization procedures that are currently
employed to train nonlinear LMs are well suited to the high-dimensional, nonlinear optimization
problems in flow control.
We remark that the lines between optimization and control are becoming blurred by the avail-
ability of powerful computers (see the sidebar titled Optimization and Control: Boundaries Erased
by Fast Computers). However, the range of critical spatiotemporal scales and the nonlinearity of
the underlying processes will likely render real-time optimization for flow control a challenge for
decades to come.
4.1. Stochastic Flow Optimization: Learning Probability Distributions

Stochastic optimization includes ES and GAs, which were originally developed based on bio-
inspired principles. However, in recent years these algorithms have been placed in a learning
framework (Kern et al. 2004).
Stochastic optimization has found widespread use in engineering design, in particular as
many engineering problems involve black-box-type cost functions. A much-abbreviated list of
applications includes aerodynamic shape optimization (Giannakoglou et al. 2006), uninhabited
aerial vehicles (UAVs) (Hamdaoui et al. 2010), shape and motion optimization in artificial
swimmers (Gazzola et al. 2012, Van Rees et al. 2015), and improved power extraction in crossflow
turbines (Strom et al. 2017). We refer readers to the review article by Skinner & Zare-Behtash
(2018) for an extensive comparison of gradient-based and stochastic optimization algorithms for
aerodynamics.
These algorithms involve large numbers of iterations, and they can benefit from massively
parallel computer architectures. Advances in automation have also facilitated their application
in experimental (Strom et al. 2017, Martin & Gharib 2018) and industrial settings (Bueche et al.
2002). We note that stochastic optimization algorithms are well suited to address the experimental
and industrial challenges associated with uncertainty, such as unexpected system behavior, partial
descriptions of the system and its environment, and exogenous disturbances. Hansen et al. (2009)
proposed an approach to enhance the capabilities of evolutionary algorithms for online optimiza-
tion of a combustor test rig.
Stochastic flow optimization will continue to benefit from advances in computer hardware
and experimental techniques. At the same time, convergence proofs, explainability, and reliability
are outstanding issues that need to be taken into consideration when deploying such algorithms in
fluid mechanics problems. Hybrid algorithms that combine in a problem-specific manner stochas-
tic techniques and gradient-based methods may offer the best strategy for flow control problems.
4.2. Flow Control with Machine Learning

Feedback flow control modifies the behavior of a fluid dynamic system through actuation that is
informed by sensor measurements. Feedback is necessary to stabilize an unstable system, attenu-

ate sensor noise, and compensate for external disturbances and model uncertainty. Challenges of
flow control include a high-dimensional state, nonlinearity, latent variables, and time delays. ML
algorithms have been used extensively in control, system identification, and sensor placement.
4.2.1. Neural networks for control. NNs have received significant attention for system identi-
fication (see Section 3) and control, including applications in aerodynamics (Phan et al. 1995). The
application of NNs to turbulence flow control was pioneered by Lee et al. (1997). The skin-friction
drag of a turbulent boundary layer was reduced using local wall-normal blowing and suction based
on few skin-friction sensors. A sensor-based control law was learned from a known optimal full-
information controller, with little loss in overall performance. Furthermore, a single-layer network
was optimized for skin-friction drag reduction without incorporating any prior knowledge of the
actuation commands. Both strategies led to a conceptually simple local opposition control. Sev-
eral other studies employ NNs, e.g., for phasor control (Rabault et al. 2019) or even frequency
cross-talk. The need to optimize many parameters is the price for the theoretical advantage of ap-
proximating arbitrary nonlinear control laws. NN control may require exorbitant computational
or experimental resources for configurations with complex high-dimensional nonlinearities and
many sensors and actuators. At the same time, the training time of NNs has been improved by
several orders of magnitude since these early applications, which warrant further investigation
into their potential for flow control.
4.2.2. Genetic algorithms for control. GAs have been deployed to solve several flow control
problems. They require that the structure of the control law be prespecified and contain only a
few adjustable parameters. An example of the use of GA for control design in fluids was used for
experimental mixing optimization of the backward-facing step (Benard et al. 2016). As with NN
control, the learning time increases with the number of parameters, making it challenging or even
prohibitive for controllers with nonlinearities (e.g., a constant-linear-quadratic law), signal history
(e.g., a Kalman filter), or multiple sensors and actuators.
Genetic programming has been used extensively in active control for engineering appli-
cations (Dracopoulos 1997, Fleming & Purshouse 2002) and in recent years in several flow
control plants. This includes the learning of multifrequency open-loop actuation, multi-input
sensor feedback, and distributed control. We refer readers to Duriez et al. (2016) for an in-depth
description of the method and to Noack (2018) for an overview of the plants. We remark that
most control laws have been obtained within 1,000 test evaluations, each requiring only a few
seconds in a wind tunnel.
4.3. Flow Control via Reinforcement Learning

In recent years, RL has advanced beyond the realm of games and has become a fundamental
mode of problem solving in a growing number of domains, including to reproduce the dynamics
of hydrological systems (Loucks et al. 2005), actively control the oscillatory laminar flow around
bluff bodies (Guéniat et al. 2016), study the individual (Gazzola et al. 2014) or collective mo-
tion of fish (Gazzola et al. 2016, Novati et al. 2017, Verma et al. 2018), maximize the range of
simulated (Reddy et al. 2016) and robotic (Reddy et al. 2018) gliders, optimize the kinematic
motion of UAVs (Kim et al. 2004, Tedrake et al. 2009), and optimize the motion of microswim-
mers (Colabrese et al. 2017, 2018). Figure 8 provides a schematic of RL with compelling examples
related to fluid mechanics.
Deep reinforcement learning scheme

Receive reward
a
A GEN T
Gaussian policy
πw
m(St ) Mean action Take action

STA TE S t E NVI RONM E NT
σ(St ) Standard
deviation
of action
Deep neural w
network Parameters
Observe state
Application to collective fish motion

b
Figure 8
Deep reinforcement learning schematic (a) and application to the study of the collective motion of fish via the Navier–Stokes equations
(b). Panel b adapted from Verma et al. (2018).

Fluid mechanics knowledge is essential for applications of RL, as success or failure hinges on
properly selecting states, actions, and rewards that reflect the governing mechanisms of the flow
problem. Natural organisms and their sensors, such as the visual system in a bird or the lateral
line in a fish, can guide the choice of states. As sensor technologies progress at a rapid pace, the
algorithmic challenge may be that of optimal sensor placement (Papadimitriou & Papadimitriou
2015, Manohar et al. 2018). The actions reflect the flow actuation device and may involve body
deformation or wing flapping. Rewards may include energetic factors, such as the cost of trans-
port or proximity to the center of a fish school to avoid predation. The computational cost of
RL remains a challenge to its widespread adoption, but we believe this deficiency can be me-
diated by the parallelism inherent in RL. There is growing interest in methods designed to be
transferable from low-accuracy (e.g., two-dimensional) to high-accuracy (e.g., three-dimensional)
simulations (Verma et al. 2018) or from simulations to related real-world applications (Richter
et al. 2016, Bousmalis et al. 2017).
5. DISCUSSION AND OUTLOOK

This review presents ML algorithms for the perspective of fluid mechanics. The interface of the
two fields has a long history and has attracted a renewed interest in the last few years. This
review addresses applications of ML in problems of flow modeling, optimization, and control
in experiments and simulations. It highlights some successes of ML in critical fluid mechan-
ics tasks, such as dimensionality reduction, feature extraction, PIV processing, superresolution,
reduced-order modeling, turbulence closure, shape optimization, and flow control. It discusses
lessons learned from these efforts and justifies the current interest in light of the technological
advances of our times. Our goal is to provide a deeper understanding of ML and its context in
fluid mechanics. ML comprises data-driven optimization and applied regression techniques that
are well suited for high-dimensional, nonlinear problems, such as those encountered in fluid dy-
namics; fluid mechanics expertise will be necessary to formulate these optimization and regression
problems.
ML algorithms present an arsenal of tools, largely unexplored in fluid mechanics research, that
can augment existing modes of inquiry. Fluid mechanics knowledge and centuries-old conserva-
tion laws remain relevant in the era of big data. Such knowledge can help frame more precise
questions and assist in reducing the large computational cost often associated with the application
of ML algorithms in flow control and optimization. The exploration and visualization of high-
dimensional search spaces can be simplified by ML and increasingly abundant high-performance
computing resources.
In the near future, experience with ML algorithms will help frame new questions in fluid me-
chanics, extending decades-old linearized models and linear approaches to the nonlinear regime.
The transition to the nonlinear realm of ML is facilitated by the abundance of open source soft-
ware and methods and the prevalent openness of the ML community. In the long term, ML will
undoubtedly offer a fresh look into old problems of fluid mechanics under the light of data. Inter-
preting the ML solutions, and refining the problem statement, will again require fluid mechanics
expertise.
A word of caution is necessary to balance the current excitement about data-driven research
and the (almost magical) powers of ML. After all, an ML algorithm will always provide some kind
of answer to any question that is based on its training data—data that may not even be relevant
to the question at hand. Properly formulating the question and selecting the data, the LM, and
its training are all critical components of the learning process. Applying ML algorithms to fluid

mechanics faces numerous outstanding challenges (and opportunities!). Although many fields of
ML are concerned with raw predictive performance, applications in fluid mechanics often require
models that are explainable and generalizable and have guarantees.
Reproducibility:
the process of Although deep learning will undoubtedly become a critical tool in several aspects of flow mod-
documenting eling, not all ML is deep learning. It is important to consider several factors when choosing meth-
procedures and ods, including the quality and quantity of data, the desired inputs and outputs, the cost function
archiving code and to be optimized, whether the task involves interpolation or extrapolation, and how important it
data so that others can
is for models to be explainable. It is also important to cross-validate ML models; otherwise, re-
fully reproduce
scientific results sults may be prone to overfitting. It is also important to develop and adapt ML algorithms that
are not only physics informed but also physics consistent, a major outstanding challenge in AI.
This review concludes with a call for action in the fluid mechanics community to further em-
brace open and reproducible research products and standards. Reproducibility is a cornerstone
of science, and several frameworks are currently developed to render this into a systematic sci-
entific process (Barber 2015). It is increasingly possible to document procedures, archive code,
and host data so that others can reproduce results. Data are essential for ML; thus, creating
and curating benchmark data sets and software will spur interest among researchers in related
fields, driving progress. These fluid benchmarks are more challenging than the traditional im-
age data sets encountered in ML: Fluid data are multimodal and multifidelity, they have high
resolution in some dimensions and are sparse in others, many tasks balance multiple objectives,
and foremost, data come from dynamical systems, where many tasks do not admit postmortem
analysis.
We are entering a new and exciting era in fluid mechanics research. Centuries of theoretical
developments based on first principles are now merging with data-driven analysis. This fusion
will provide solutions to many long-sought problems in fluid dynamics, first and foremost the
enhanced understanding of its governing mechanisms.
SUMMARY POINTS
1. Machine learning (ML) entails powerful information-processing algorithms that are rel-
evant for modeling, optimization, and control of fluids. Effective problem solvers will
have expertise in ML and in-depth knowledge of fluid mechanics.
2. Fluid mechanics has been traditionally concerned with big data. For decades it has used
ML to understand, predict, optimize, and control flows. Currently, ML capabilities are
advancing at an incredible rate, and fluid mechanics is beginning to tap into the full
potential of these powerful methods.
3. Many tasks in fluid mechanics, such as reduced-order modeling, shape optimization, and
feedback control, may be posed as optimization and regression tasks. ML can improve
optimization performance and reduce convergence time. ML is also used for dimension-
ality reduction and identifying low-dimensional manifolds and discrete flow regimes,
which benefit understanding.
4. Flow control strategies have been traditionally based on the precise sequence from un-
derstanding to modeling and then to control. The ML paradigm suggests more flexibility
and iterates between data-driven and first principle approaches.

FUTURE ISSUES
1. ML algorithms often come without guarantees for performance, robustness, or conver-
gence, even for well-defined tasks. How can interpretability, generalizability, and explain-
ability of the results be achieved?
2. Incorporating and enforcing known flow physics is a challenge and opportunity for
ML algorithms. Can we hybridize data-driven and first principle approaches in fluid
mechanics?
3. There are many possibilities to discover new physical mechanisms, symmetries, con-
straints, and invariances from fluids data.

4. Data-driven modeling may be a potent alternative in revisiting existing empirical laws
in fluid mechanics.
5. ML encourages open sharing of data and software. Can this assist the development of
frameworks for reproducible and open science in fluid mechanics?
6. Fluids researchers will benefit from interfacing with the ML community, where the latest
advances are reported at peer-reviewed conferences.
DISCLOSURE STATEMENT
The authors are not aware of any biases that might be perceived as affecting the objectivity of this
review.
ACKNOWLEDGMENTS
S.L.B. acknowledges funding from the Army Research Office (ARO W911NF-17-1-0306,
W911NF-17-1-0422) and the Air Force Office of Scientific Research (AFOSR FA9550-18-1-
0200). B.R.N. acknowledges funding from LIMSI-CNRS, Université Paris Sud (SMEMaG),
the French National Research Agency (ANR-11-IDEX-0003-02, ANR-17-ASTR-0022), and the
German Research Foundation (CRC880, SE 2504/2-1, SE 2504/3-1). P.K. acknowledges funding
from an ERC Advanced Investigator Award (FMCoBe, No. 34117), the Swiss National Science
Foundation, and the Swiss National Supercomputing Centre. The authors are grateful for discus-
sions with Nathan Kutz (University of Washington), Jean-Christophe Loiseau (ENSAM Paris-
Tech, Paris), François Lusseyran (LIMSI-CNRS, Paris), Guido Novati (ETH Zurich), Luc Pastur
(ENSTA ParisTech, Paris), and Pantelis Vlachas (ETH Zurich).
LITERATURE CITED
Adrian RJ. 1991. Particle-imaging techniques for experimental fluid mechanics. Annu. Rev. Fluid Mech. 23:261–
304
Alsalman M, Colvert B, Kanso E. 2018. Training bioinspired sensors to classify flows. Bioinspiration Biomim.
14:016009
Amsallem D, Zahr MJ, Farhat C. 2012. Nonlinear model order reduction based on local reduced-order bases.
Int. J. Numer. Meth. Eng. 92:891–916
Bai Z, Wimalajeewa T, Berger Z, Wang G, Glauser M, Varshney PK. 2014. Low-dimensional approach for
reconstruction of airfoil data via compressive sensing. AIAA J. 53:920–33
Baldi P, Hornik K. 1989. Neural networks and principal component analysis: learning from examples without
local minima. Neural Netw. 2:53–58

Barber D. 2012. Bayesian Reasoning and Machine Learning. Cambridge, UK: Cambridge Univ. Press
Reproducible science: a Barber RF, Candes EJ. 2015. Controlling the false discovery rate via knock-offs. Ann. Stat. 43:2055–
framework. 85
Battaglia PW, Hamrick JB, Bapst V, Sanchez-Gonzalez A, Zambaldi V, et al. 2018. Relational inductive biases,
deep learning, and graph networks. arXiv:1806.01261 [cs.LG]
Bellman R. 1952. On the theory of dynamic programming. PNAS 38:716–19
Benard N, Pons-Prats J, Periaux J, Bugeda G, Braud P, et al. 2016. Turbulent separated shear flow control by
surface plasma actuator: experimental optimization by genetic algorithm approach. Exp. Fluids 57:22
Bewley TR, Moin P, Temam R. 2001. DNS-based predictive control of turbulence: an optimal benchmark for
feedback algorithms. J. Fluid Mech. 447:179–225
Bishop CM, James GD. 1993. Analysis of multiphase flows using dual-energy gamma densitometry and neural
networks. Nucl. Instrum. Methods Phys. Res. 327:580–93
Theoretical analysis of Bölcskei H, Grohs P, Kutyniok G, Petersen P. 2019. Optimal approximation with sparsely connected
the approximation deep neural networks. SIAM J. Math. Data Sci. 1:8–45
properties of deep Bourguignon JL, Tropp JA, Sharma AS, McKeon BJ. 2014. Compact representation of wall-bounded turbu-
neural networks. lence using compressive sampling. Phys. Fluids 26:015109
Bousmalis K, Irpan A, Wohlhart P, Bai Y, Kelcey M, et al. 2017. Using simulation and domain adaptation to
improve efficiency of deep robotic grasping. arXiv:1709.07857 [cs.LG]
Breiman L. 2001. Random forests. Mach. Learn. 45:5–32
Bright I, Lin G, Kutz JN. 2013. Compressive sensing based machine learning strategies for characterizing the
flow around a cylinder with limited pressure measurements. Phys. Fluids 25:127102
Brunton SL, Kutz JN. 2019. Data-Driven Science and Engineering: Machine Learning, Dynamical Systems, and
Control. New York: Cambridge Univ. Press
Brunton SL, Noack BR. 2015. Closed-loop turbulence control: progress and challenges. Appl. Mech. Rev.
67:050801
Brunton SL, Proctor JL, Kutz JN. 2016. Discovering governing equations from data by sparse identification
of nonlinear dynamical systems. PNAS 113:3932–37
Bueche D, Stoll P, Dornberger R, Koumoutsakos P. 2002. Multi-objective evolutionary algorithm for the
optimization of noisy combustion problems. IEEE Trans. Syst. Man Cybern. C 32:460–73
Chen TQ, Rubanova Y, Bettencourt J, Duvenaud DK. 2018. Neural ordinary differential equations. In Ad-
vances in Neural Information Processing Systems 31, ed. S Bengio, H Wallach, H Larochelle, K Grauman,
N Cesa-Bianchi, R Garnett, pp. 6571–83. Red Hook, NY: Curran Assoc.
Cherkassky V, Mulier FM. 2007. Learning from Data: Concepts, Theory, and Methods. Hoboken, NJ: John Wiley
& Sons
Colabrese S, Gustavsson K, Celani A, Biferale L. 2017. Flow navigation by smart microswimmers via rein-
forcement learning. Phys. Rev. Lett. 118:158004
Colabrese S, Gustavsson K, Celani A, Biferale L. 2018. Smart inertial particles. Phys. Rev. Fluids 3:084301
Colvert B, Alsalman M, Kanso E. 2018. Classifying vortex wakes using neural networks. Bioinspiration Biomim.
13:025003
Dissanayake M, Phan-Thien N. 1994. Neural-network-based approximations for solving partial differential
equations. Comm. Numer. Meth. Eng. 10:195–201
Dong C, Loy CC, He K, Tang X. 2014. Learning a deep convolutional network for image super-resolution. In
Computer Vision—ECCV 2014, ed. D Fleet, T Pajdla, B Schiele, T Tuytelaars, pp. 184–99. Cham, Switz.:
Springer
Donoho DL. 2006. Compressed sensing. IEEE Trans. Inf. Theory 52:1289–306
Dracopoulos DC. 1997. Evolutionary Learning Algorithms for Neural Adaptive Control. London: Springer-Verlag
Duraisamy K, Iaccarino G, Xiao H. 2019. Turbulence modeling in the age of data. Annu. Rev. Fluid Mech.
51:357–77
Duriez T, Brunton SL, Noack BR. 2016. Machine Learning Control: Taming Nonlinear Dynamics and Turbulence.
Cham, Switz.: Springer
Faller WE, Schreck SJ. 1996. Neural networks: applications and opportunities in aeronautics. Prog. Aerosp. Sci.
32:433–56

Fleming PJ, Purshouse RC. 2002. Evolutionary algorithms in control systems engineering: a survey. Control
Eng. Pract. 10:1223–41
Freeman WT, Jones TR, Pasztor EC. 2002. Example-based super-resolution. IEEE Comput. Graph. 22:56–65
Fukami K, Fukagata K, Taira K. 2018. Super-resolution reconstruction of turbulent flows with machine learn-
ing. arXiv:1811.11328 [physics.flu-dyn]
Gardner E. 1988. The space of interactions in neural network models. J. Phys. A 21:257
Gazzola M, Hejazialhosseini B, Koumoutsakos P. 2014. Reinforcement learning and wavelet adapted vortex
methods for simulations of self-propelled swimmers. SIAM J. Sci. Comput. 36:B622–39
Gazzola M, Tchieu A, Alexeev D, De Brauer A, Koumoutsakos P. 2016. Learning to school in the presence of
hydrodynamic interactions. J. Fluid Mech. 789:726–49
Gazzola M, Van Rees WM, Koumoutsakos P. 2012. C-start: optimal start of larval fish. J. Fluid Mech. 698:5–18
Germano M, Piomelli U, Moin P, Cabot WH. 1991. A dynamic subgrid-scale eddy viscosity model. Phys.
Fluids 3:1760–65
Giannakoglou K, Papadimitriou D, Kampolis I. 2006. Aerodynamic shape design using evolutionary algo-
rithms and new gradient-assisted metamodels. Comput. Methods Appl. Mech. Eng. 195:6312–29
Glaz B, Liu L, Friedmann PP. 2010. Reduced-order nonlinear unsteady aerodynamic modeling using a
surrogate-based recurrence framework. AIAA J. 48:2418–29
Gonzalez-Garcia R, Rico-Martinez R, Kevrekidis I. 1998. Identification of distributed parameter systems: a
neural net based approach. Comput. Chem. Eng. 22:S965–68
Goodfellow I, Bengio Y, Courville A. 2016. Deep Learning. Cambridge, MA: MIT Press
Goodfellow I, Pouget-Abadie J, Mirza M, Xu B, Warde-Farley D, et al. 2014. Generative adversarial Powerful deep learning
nets. In Advances in Neural Information Processing Systems 27, ed. Z Ghahramani, M Welling, architecture that learns
C Cortes, ND Lawrence, KQ Weinberger, pp. 2672–80. Red Hook, NY: Curran Assoc. through a game
Grant I, Pan X. 1995. An investigation of the performance of multi layer, neural networks applied to the between a network that
can generate new data
analysis of PIV images. Exp. Fluids 19:159–66
and a network that is an
Graves A, Fernández S, Schmidhuber J. 2007. Multi-dimensional recurrent neural networks. In Artificial Neural
expert classifier.
Networks—ICANN 2007, ed. JM de Sa, LA Alexandre, W Duch, D Mandic, pp. 549–58. Berlin: Springer
Grossberg S. 1976. Adaptive pattern classification and universal recoding, I: parallel development and coding
of neural feature detectors. Biol. Cybernet. 23:121–34
Grossberg S. 1988. Nonlinear neural networks: Principles, mechanisms, and architectures. Neural Netw. 1:17–
61
Guéniat F, Mathelin L, Hussaini MY. 2016. A statistical learning strategy for closed-loop control of fluid flows.
Theor. Comput. Fluid Dyn. 30:497–510
Halko N, Martinsson PG, Tropp JA. 2011. Finding structure with randomness: probabilistic algorithms for
constructing approximate matrix decompositions. SIAM Rev. 53:217–88
Hamdaoui M, Chaskalovic J, Doncieux S, Sagaut P. 2010. Using multiobjective evolutionary algorithms and
data-mining methods to optimize ornithopters’ kinematics. J. Aircraft 47:1504–16
Hansen N, Müller SD, Koumoutsakos P. 2003. Reducing the time complexity of the derandomized evolution
strategy with covariance matrix adaptation (CMA-ES). Evol. Comput. 11:1–18
Hansen N, Niederberger AS, Guzzella L, Koumoutsakos P. 2009. A method for handling uncertainty in evo-
lutionary optimization with an application to feedback control of combustion. IEEE Trans. Evol. Comput.
13:180–97
Hastie T, Tibshirani R, Friedman J, Hastie T, Friedman J, Tibshirani R. 2009. The Elements of Statistical
Learning, Vol. 2. New York: Springer
Hinton GE, Sejnowski TJ. 1986. Learning and relearning in Boltzmann machines. In Parallel Distributed Pro- Regularization of
recurrent neural
cessing: Explorations in the Microstructure of Cognition, Vol. 1, ed. DE Rumelhart, J McClelland, pp. 282–317.
networks and major
Cambridge, MA: MIT Press
contributor to the
Hochreiter S, Schmidhuber J. 1997. Long short-term memory. Neural Comput. 9:1735–80
success of Google
Holland JH. 1975. Adaptation in Natural and Artificial Systems: An Introductory Analysis with Applications to Biol- Translate.
ogy, Control, and Artificial Intelligence. Ann Arbor: Univ. Mich. Press
Hopfield JJ. 1982. Neural networks and physical systems with emergent collective computational abilities.
PNAS 79:2554–58

Hornik K, Stinchcombe M, White H. 1989. Multilayer feedforward networks are universal approximators.
Neural Netw. 2:359–66
Hou W, Darakananda D, Eldredge J. 2019. Machine learning based detection of flow disturbances using surface
pressure measurements. Paper presented at AIAA Atmospheric Flight Mechanics Conference 2019, San
Diego, Calif., AIAA Pap. 2019-1148
Jambunathan K, Hartle S, Ashforth-Frost S, Fontama V. 1996. Evaluating convective heat transfer coefficients
using neural networks. Int. J. Heat Mass Transf. 39:2329–32
Kaiser E, Noack BR, Cordier L, Spohn A, Segond M, et al. 2014. Cluster-based reduced-order modelling of
a mixing layer. J. Fluid Mech. 754:365–414
Kern S, Müller SD, Hansen N, Büche D, Ocenasek J, Koumoutsakos P. 2004. Learning probability distribu-
tions in continuous evolutionary algorithms—a comparative review. Nat. Comput. 3:77–112

Kim B, Azevedo VC, Thuerey N, Kim T, Gross M, Solenthaler B. 2018. Deep fluids: a generative network
for parameterized fluid simulations. arXiv:1806.02071 [cs.LG]
Kim HJ, Jordan MI, Sastry S, Ng AY. 2004. Autonomous helicopter flight via reinforcement learning. In
Advances in Neural Information Processing Systems 17, ed. B Scholkopf, Y Vis, JC Platt, pp. 799–806.
Cambridge, MA: MIT Press
Knaak M, Rothlubbers C, Orglmeister R. 1997. A Hopfield neural network for flow field computation based on
particle image velocimetry/particle tracking velocimetry image sequences. IEEE Int. Conf. Neural Netw.
1:48–52
Kohonen T. 1995. Self-Organizing Maps. Berlin: Springer-Verlag
Kolmogorov A. 1941. The local structure of turbulence in incompressible viscous fluid for very large Reynolds
number. Cr. Acad. Sci. USSR 30:301–5
Koza JR. 1992. Genetic Programming: On the Programming of Computers by Means of Natural Selection. Boston:
MIT Press
Krizhevsky A, Sutskever I, Hinton GE. 2012. Imagenet classification with deep convolutional neural net-
works. In Advances in Neural Information Processing Systems 25, ed. F Pereira, CJC Burges, L Bottou,
KQ Weinberger, pp. 1097–105. Red Hook, NY: Curran Assoc.
Kutz JN, Brunton SL, Brunton BW, Proctor JL. 2016. Dynamic Mode Decomposition: Data-Driven Modeling of
Complex Systems. Philadelphia: SIAM
Labonté G. 1999. A new neural network for particle-tracking velocimetry. Exp. Fluids 26:340–46
Lagaris IE, Likas A, Fotiadis DI. 1998. Artificial neural networks for solving ordinary and partial differential
equations. IEEE Trans. Neural Netw. 9:987–1000
LeCun Y, Bengio Y, Hinton G. 2015. Deep learning. Nature 521:436–44
Lee C, Kim J, Babcock D, Goodman R. 1997. Application of neural networks to turbulence control for drag
reduction. Phys. Fluids 9:1740–47
Lee Y, Yang H, Yin Z. 2017. PIV-DCNN: cascaded deep convolutional neural networks for particle image
velocimetry. Exp. Fluids 58:171
Li Q, Dietrich F, Bollt EM, Kevrekidis IG. 2017. Extended dynamic mode decomposition with dictionary
learning: a data-driven adaptive spectral decomposition of the Koopman operator. Chaos 27:103111
Liang D, Jiang C, Li Y. 2003. Cellular neural network to detect spurious vectors in PIV data. Exp. Fluids
34:52–62
Ling J, Jones R, Templeton J. 2016a. Machine learning strategies for systems with invariance properties.
J. Comput. Phys. 318:22–35
Ling J, Kurzawski A, Templeton J. 2016b. Reynolds averaged turbulence modelling using deep neural net-
works with embedded invariance. J. Fluid Mech. 807:155–66
Ling J, Templeton J. 2015. Evaluation of machine learning algorithms for prediction of regions of high
Reynolds averaged Navier Stokes uncertainty. Phys. Fluids 27:085103
Loiseau JC, Brunton SL. 2018. Constrained sparse Galerkin regression. J. Fluid Mech. 838:42–67
Loucks D, van Beek E, Stedinger J, Dijkman J, Villars M. 2005. Water Resources Systems Planning and Manage-
ment: An Introduction to Methods, Vol. 2. Cham, Switz.: Springer
Lumley J. 1970. Stochastic Tools in Turbulence. New York: Academic

Lusch B, Kutz JN, Brunton SL. 2018. Deep learning for universal linear embeddings of nonlinear dynamics.
Nat. Commun. 9:4950
Mahoney MW. 2011. Randomized algorithms for matrices and data. Found. Trends Mach. Learn. 3:123–224
Manohar K, Brunton BW, Kutz JN, Brunton SL. 2018. Data-driven sparse sensor placement for reconstruc-
tion: demonstrating the benefits of exploiting known patterns. IEEE Control Syst. Mag. 38:63–86
Mardt A, Pasquali L, Wu H, Noé F. 2018. VAMPnets for deep learning of molecular kinetics. Nat. Commun.
9:5
Martin N, Gharib M. 2018. Experimental trajectory optimization of a flapping fin propulsor using an evolu-
tionary strategy. Bioinspiration Biomim. 14:016010
Maulik R, San O, Rasheed A, Vedula P. 2019. Subgrid modelling for two-dimensional turbulence using neural
networks. J. Fluid Mech. 858:122–44

Meena MG, Nair AG, Taira K. 2018. Network community-based model reduction for vortical flows. Phys. Rev.
E 97:063103
Mehta UB, Kutler P. 1984. Computational aerodynamics and artificial intelligence. NASA Tech. Rep. NASA-TM-
85994, Ames Res. Cent., Moffett Field, CA

Meijering E. 2002. A chronology of interpolation: from ancient astronomy to modern signal and image pro-
cessing. Proc. IEEE 90:319–42
Meneveau C, Katz J. 2000. Scale-invariance and turbulence models for large-eddy simulation. Annu. Rev. Fluid
Mech. 32:1–32
Mezic I. 2013. Analysis of fluid flows via spectral properties of the Koopman operator. Annu. Rev. Fluid Mech.
45:357–78
Milano M, Koumoutsakos P. 2002. Neural network modeling for near wall turbulent flow. J. Comput. Phys.
182:1–26
Minsky M, Papert SA. 1969. Perceptrons: An Introduction to Computational Geometry. Cambridge, MA: MIT
Press
Mnih V, Kavukcuoglu K, Silver D, Rusu AA, Veness J, et al. 2015. Human-level control through deep rein-
forcement learning. Nature 518:529–33
Nair AG, Taira K. 2015. Network-theoretic approach to sparsified discrete vortex dynamics. J. Fluid Mech.
768:549–71
Noack BR. 2018. Closed-loop turbulence control—from human to machine learning (and retour). In Fluid-
Structure-Sound Interactions and Control: Proceedings of the 4th Symposium on Fluid-Structure-Sound Interac-
tions and Control, ed. Y Zhou, M Kimura, G Peng, AD Lucey, L Hung, pp. 23–32. Singapore: Springer
Noé F, Nüske F. 2013. A variational approach to modeling slow processes in stochastic dynamical systems.
SIAM Multiscale Model Simul. 11:635–55
Novati G, Mahadevan L, Koumoutsakos P. 2019. Controlled gliding and perching through deep-
reinforcement-learning. Phys. Rev. Fluids 4:093902
Novati G, Verma S, Alexeev D, Rossinelli D, Van Rees WM, Koumoutsakos P. 2017. Synchronisation through
learning for two self-propelled swimmers. Bioinspiration Biomim. 12:aa6311
Nüske F, Schneider R, Vitalini F, Noé F. 2016. Variational tensor approach for approximating the rare-event
kinetics of macromolecular systems. J. Chem. Phys. 144:054105
Ollivier Y, Arnold L, Auger A, Hansen N. 2017. Information-geometric optimization algorithms: a unifying
picture via invariance principles. J. Mach. Learn. Res. 18:564–628
Ostermeier A, Gawelczyk A, Hansen N. 1994. Step-size adaptation based on non-local use of selection infor-
mation. In International Conference on Parallel Problem Solving from Nature 1994 Oct 9, pp. 189–98. Berlin:
Springer
Ouellette NT, Xu H, Bodenschatz E. 2006. A quantitative study of three-dimensional Lagrangian particle
tracking algorithms. Exp. Fluids 40:301–13
Papadimitriou DI, Papadimitriou C. 2015. Optimal sensor placement for the estimation of turbulent model
parameters in CFD. Int. J. Uncert. Quant. 5:545–68
Parish EJ, Duraisamy K. 2016. A paradigm for data-driven predictive modeling using field inversion and ma-
chine learning. J. Comput. Phys. 305:758–74

Pathak J, Hunt B, Girvan M, Lu Z, Ott E. 2018. Model-free prediction of large spatiotemporally chaotic
systems from data: a reservoir computing approach. Phys. Rev. Lett. 120:024102
Pelikan M, Ocenasek J, Trebst S, Troyer M, Alet F. 2004. Computational complexity and simulation of rare
events of Ising spin glasses. In Genetic and Evolutionary Computation—GECCO 2004, pp. 36–47. Berlin:
Springer
Perdikaris P, Venturi D, Karniadakis G. 2016. Multifidelity information fusion algorithms for high-
dimensional systems and massive data sets. SIAM J. Sci. Comput. 38:B521–38
Perlman E, Burns R, Li Y, Meneveau C. 2007. Data exploration of turbulence simulations using a database
cluster. In Proceedings of the 2007 ACM/IEEE Conference on Supercomputing, p. 23. New York: ACM
Phan MQ, Juang JN, Hyland DC. 1995. On neural networks in identification and control of dynamic sys-
tems. In Wave Motion, Intelligent Structures and Nonlinear Mechanics, ed. A Guran, DJ Inman, pp. 194–225.
Singapore: World Sci.
Pierret S, Van den Braembussche R. 1999. Turbomachinery blade design using a Navier-Stokes solver and
artificial neural network. J. Turbomach. 121:326–32
Pollard A, Castillo L, Danaila L, Glauser M. 2016. Whither Turbulence and Big Data in the 21st Century? Cham.,
Switz.: Springer
Rabault J, Kuchta M, Jensen A, Réglade U, Cerardi N. 2019. Artificial neural networks trained through deep
reinforcement learning discover control strategies for active flow control. J. Fluid Mech. 865:281–302
Raissi M, Karniadakis GE. 2018. Hidden physics models: machine learning of nonlinear partial differential
equations. J. Comput. Phys. 357:125–41
Raissi M, Perdikaris P, Karniadakis G. 2019. Physics-informed neural networks: a deep learning framework
for solving forward and inverse problems involving nonlinear partial differential equations. J. Comput.
Phys. 378:686–707
Rechenberg I. 1964. Kybernetische lösungsansteuerung einer experimentellen forschungsaufgabe. In Annual
Conference of the WGLR at Berlin in September, Vol. 35, p. 33
Rechenberg I. 1973. Evolutionsstrategie: Optimierung technischer Systeme nach Prinzipien der Biologischen Evolution.
Stuttgart, Ger.: Frommann-Holzboog
Reddy G, Celani A, Sejnowski TJ, Vergassola M. 2016. Learning to soar in turbulent environments. PNAS
113:E4877–84
Reddy G, Wong-Ng J, Celani A, Sejnowski TJ, Vergassola M. 2018. Glider soaring via reinforcement learning
in the field. Nature 562:236–39
Richter SR, Vineet V, Roth S, Koltun V. 2016. Playing for data: ground truth from computer games. In
European Conference on Computer Vision 2016, ed. B Leibe, J Matas, N Sebe, M Welling, pp. 102–18.
Cham, Switz.: Springer
Rico-Martinez R, Krischer K, Kevrekidis IG, Kube MC, Hudson JL. 1992. Discrete- vs. continuous-time
nonlinear signal processing of Cu electrodissolution data. Chem. Eng. Commun. 118:25–48
Rokhlin V, Szlam A, Tygert M. 2009. A randomized algorithm for principal component analysis. SIAM J.
Matrix Anal. Appl. 31:1100–24
First example of a Rosenblatt F. 1958. The perceptron: a probabilistic model for information storage and organization
simple binary network in the brain. Psychol. Rev. 65:386–408
with learning Rowley CW, Dawson S. 2016. Model reduction for flow analysis and control. Annu. Rev. Fluid Mech. 49:387–
capabilities. 417
Rowley CW, Mezić I, Bagheri S, Schlatter P, Henningson D. 2009. Spectral analysis of nonlinear flows.
J. Fluid Mech. 645:115–27
Rudy SH, Brunton SL, Proctor JL, Kutz JN. 2017. Data-driven discovery of partial differential equations. Sci.
Adv. 3:e1602614
Rumelhart DE, Hinton GE, Williams RJ. 1986. Learning representations by back-propagating errors. Nature
323:533–36
Salimans T, Ho J, Chen X, Sutskever I. 2017. Evolution strategies as a scalable alternative to reinforcement
learning. arXiv:1703.03864 [stat.ML]
Schaeffer H. 2017. Learning partial differential equations via data discovery and sparse optimization. Proc. R.
Soc. A 473:20160446

Schaul T, Horgan D, Gregor K, Silver D. 2015. Universal value function approximators. In Proceedings of the
32nd International Conference on Machine Learning, ed. F Bach, DM Blei, pp. 1312–20. Red Hook, NY:
Curran Assoc.
Schmid PJ. 2010. Dynamic mode decomposition for numerical and experimental data. J. Fluid Mech. 656:5–28
Schmidt M, Lipson H. 2009. Distilling free-form natural laws from experimental data. Science 324:81–85
Schölkopf B, Smola AJ. 2002. Learning with Kernels: Support Vector Machines, Regularization, Optimization, and
Beyond. Cambridge, MA: MIT Press
Schwefel HP. 1977. Numerische Optimierung von Computer-Modellen Mittels der Evolutionsstrategie. Basel, Switz.:
Birkhäuser
Semeraro O, Lusseyran F, Pastur L, Jordan P. 2016. Qualitative dynamics of wavepackets in turbulent jets.
arXiv:1608.06750 [physics.flu-dyn]
Silver D, Huang A, Maddison CJ, Guez A, Sifre L, et al. 2016. Mastering the game of Go with deep neural
networks and tree search. Nature 529:484–89
Singh AP, Medida S, Duraisamy K. 2017. Machine-learning-augmented predictive modeling of turbulent
separated flows over airfoils. AIAA J. 55:2215–27

Sirovich L. 1987. Turbulence and the dynamics of coherent structures, parts I–III. Q. Appl. Math. 45:561–90
Sirovich L, Kirby M. 1987. A low-dimensional procedure for the characterization of human faces. J. Opt. Soc.
Am. A 4:519–24
Skinner SN, Zare-Behtash H. 2018. State-of-the-art in aerodynamic shape optimisation methods. Appl. Soft.
Comput. 62:933–62
Strom B, Brunton SL, Polagye B. 2017. Intracycle angular velocity control of cross-flow turbines. Nat. Energy
2:17103
The classic book on
Sutton RS, Barto AG. 2018. Reinforcement Learning: An Introduction. Cambridge, MA: MIT Press.
reinforcement learning.
2nd ed.
Taira K, Brunton SL, Dawson S, Rowley CW, Colonius T, et al. 2017. Modal analysis of fluid flows: an
overview. AIAA J. 55:4013–41
Takeishi N, Kawahara Y, Yairi T. 2017. Learning Koopman invariant subspaces for dynamic mode decom-
position. In Advances in Neural Information Processing Systems 30, ed. I Guyon, UV Luxburg, S Bengio,
H Wallach, R Fergus, et al., pp. 1130–40. Red Hook, NY: Curran Assoc.
Tedrake R, Jackowski Z, Cory R, Roberts JW, Hoburg W. 2009. Learning to fly like a bird. Paper presented at
the 14th International Symposium on Robotics Research, Lucerne, Switz.
Teo C, Lim K, Hong G, Yeo M. 1991. A neural net approach in analysing photographs in PIV. IEEE Trans.
Syst. Man Cybern. 3:1535–38
Tesauro G. 1992. Practical issues in temporal difference learning. Mach. Learn. 8:257–77
A monograph linking
Theodoridis S. 2015. Machine Learning: A Bayesian and Optimization Perspective. San Diego, CA:
machine learning and
Academic
Bayesian inference
Tsiotras P, Mesbahi M. 2017. Toward an algorithmic control theory. J. Guid. Control Dyn. 40:194–96
algorithms.
Van Rees WM, Gazzola M, Koumoutsakos P. 2015. Optimal morphokinematics for undulatory swimmers at
intermediate reynolds numbers. J. Fluid Mech. 775:178–88
Verma S, Novati G, Koumoutsakos P. 2018. Efficient collective swimming by harnessing vortices through
deep reinforcement learning. PNAS 115:5849–54
Vlachas PR, Byeon W, Wan ZY, Sapsis TP, Koumoutsakos P. 2018. Data-driven forecasting of high-
dimensional chaotic systems with long short-term memory networks. Proc. R. Soc. A 474:20170844
Wan ZY, Vlachas P, Koumoutsakos P, Sapsis T. 2018. Data-assisted reduced-order modeling of extreme events
in complex dynamical systems. PLOS ONE 13:e0197704
Wang JX, Wu JL, Xiao H. 2017. Physics-informed machine learning approach for reconstructing Reynolds
stress modeling discrepancies based on DNS data. Phys. Rev. Fluids 2:034603
Wang M, Hemati MS. 2017. Detecting exotic wakes with hydrodynamic sensors. arXiv:1711.10576
Wehmeyer C, Noé F. 2018. Time-lagged autoencoders: deep learning of slow collective variables for molecular
kinetics. J. Chem. Phys. 148:241703
Wiener N. 1965. Cybernetics or Control and Communication in the Animal and the Machine, Vol. 25. Cambridge,
MA: MIT Press

Willert CE, Gharib M. 1991. Digital particle image velocimetry. Exp. Fluids 10:181–93
Williams MO, Rowley CW, Kevrekidis IG. 2015. A kernel approach to data-driven Koopman spectral analysis.
J. Comput. Dyn. 2:247–65
Wright J, Yang A, Ganesh A, Sastry S, Ma Y. 2009. Robust face recognition via sparse representation. IEEE
Trans. Pattern Anal. Mach. Intell. 31:210–27
Wu H, Mardt A, Pasquali L, Noe F. 2018. Deep generative Markov State Models. Adv. Neural Inf. Process. Syst.
31:3975–84
Wu X, Moin P. 2008. A direct numerical simulation study on the mean velocity characteristics in turbulent
pipe flow. J. Fluid Mech. 608:81–112
Xiao H, Wu JL, Wang JX, Sun R, Roy C. 2016. Quantifying and reducing model-form uncertainties
in Reynolds-averaged Navier–Stokes simulations: a data-driven, physics-informed Bayesian approach.

J. Comp. Phys. 324:115–36
Xie Y, Franz E, Chu M, Thuerey N. 2018. tempoGAN: a temporally coherent, volumetric GAN for super-
resolution fluid flow. ACM Trans. Graph. 37:95
Yang J, Wright J, Huang TS, Ma Y. 2010. Image super-resolution via sparse representation. IEEE Trans. Image
Process. 19:2861–73

FL52_FrontMatter ARI 13 November 2019 13:18
Annual Review of
Contents Fluid Mechanics
Volume 52, 2020
Anatol Roshko, 1923–2017

Dimitri Papamoschou and Morteza Gharib p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 1
David J. Benney: Nonlinear Wave and Instability Processes

in Fluid Flows
T.R. Akylas p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p21
Ocean Wave Interactions with Sea Ice: A Reappraisal

Vernon A. Squire p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p37
Particles, Drops, and Bubbles Moving Across Sharp Interfaces
and Stratified Layers
Jacques Magnaudet and Matthieu J. Mercier p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p61
Convective Phenomena in Mushy Layers
Daniel M. Anderson and Peter Guba p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p93
Shear Thickening of Concentrated Suspensions: Recent Developments
and Relation to Other Phenomena
Jeffrey F. Morris p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 121
Subglacial Plumes
Ian J. Hewitt p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 145
Modeling Turbulent Flows in Porous Media
Brian D. Wood, Xiaoliang He, and Sourabh V. Apte p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 171
Acoustic Tweezers for Particle and Fluid Micromanipulation
M. Baudoin and J.-L. Thomas p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 205
Liquid-State Dewetting of Pulsed-Laser-Heated Nanoscale Metal
Films and Other Geometries
Lou Kondic, Alejandro G. González, Javier A. Diez,
Jason D. Fowlkes, and Philip Rack p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 235
Capillarity in Soft Porous Solids
Jonghyun Ha and Ho-Young Kim p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 263
Statics and Dynamics of Soft Wetting
Bruno Andreotti and Jacco H. Snoeijer p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 285
Turbulence with Large Thermal and Compositional
Density Variations
Daniel Livescu p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 309
v
FL52_FrontMatter ARI 13 November 2019 13:18
Patterns in Wall-Bounded Shear Flows

Laurette S. Tuckerman, Matthew Chantry, and Dwight Barkley p p p p p p p p p p p p p p p p p p p p p p p 343
Super-Resolution Imaging in Fluid Mechanics Using New
Illumination Approaches
Minami Yoda p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 369
Aeroacoustics of Silent Owl Flight
Justin W. Jaworski and N. Peake p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 395
Immersed Methods for Fluid–Structure Interaction

Boyce E. Griffith and Neelesh A. Patankar p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 421
Advances in Bioconvection
Martin A. Bees p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 449

Machine Learning for Fluid Mechanics
Steven L. Brunton, Bernd R. Noack, and Petros Koumoutsakos p p p p p p p p p p p p p p p p p p p p p p p p p p 477
Electroconvection near Electrochemical Interfaces: Experiments,
Modeling, and Computation
Ali Mani and Karen May Wang p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 509
Chemo-Hydrodynamic Patterns and Instabilities
A. De Wit p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 531
Indexes
Cumulative Index of Contributing Authors, Volumes 1–52 p p p p p p p p p p p p p p p p p p p p p p p p p p p p 557

Cumulative Index of Article Titles, Volumes 1–52 p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 568
Errata
An online log of corrections to Annual Review of Fluid Mechanics articles may be found
at http://www.annualreviews.org/errata/fluid
vi Contents

Machine Learning For Fluid Mechanics

Uploaded by

Copyright:

Available Formats

Machine Learning For Fluid Mechanics

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Machine Learning For Fluid Mechanics

Uploaded by

Copyright:

Available Formats

FL52CH19_Koumoutsakos ARjats.

cls December 7, 2019 16:29

Annual Review of Fluid Mechanics

Steven L. Brunton,1 Bernd R. Noack,2,3

and Petros Koumoutsakos4

Annu. Rev. Fluid Mech. 2020. 52:477–508 Keywords

The Annual Review of Fluid Mechanics is online at Abstract

Supervised Semisupervised Unsupervised

Optimization Reinforcement Generative Dimensionality

Support vector Linear Linear control Q-learning Generative k-means POD/PCA

478 Brunton • Noack • Koumoutsakos

www.annualreviews.org • Machine Learning for Fluid Mechanics 479

480 Brunton • Noack • Koumoutsakos

1.2. Challenges and Opportunities for Machine Learning in Fluid Dynamics

LEARNING FLUID MECHANICS: FROM LIVING ORGANISMS TO MACHINES

www.annualreviews.org • Machine Learning for Fluid Mechanics 481

2. MACHINE LEARNING FUNDAMENTALS

482 Brunton • Noack • Koumoutsakos

2.1. Supervised Learning

L[y, φ(x, y, w)] = ||y − φ(x, y, w)||2 . 2.

putational cost associated with the learning methods.

www.annualreviews.org • Machine Learning for Fluid Mechanics 483

ht–1 ht ht+1 ht+1 ht ht+1

x t–1 xt x t+1 x t–1 xt x t+1

2.2. Unsupervised Learning

2.2.1. Dimensionality reduction I: proper orthogonal decomposition, principal compo-

484 Brunton • Noack • Koumoutsakos

PCA/POD Deep encoder Deep decoder

www.annualreviews.org • Machine Learning for Fluid Mechanics 485

2.2.3. Clustering and vector quantization. Clustering is an unsupervised learning technique

L[φ(x, w)] = ||x − φ(x, w)||2 . 4.

2.3. Semisupervised Learning

486 Brunton • Noack • Koumoutsakos

2.3.2. Reinforcement learning. RL is a mathematical framework for problem solving (Sutton

www.annualreviews.org • Machine Learning for Fluid Mechanics 487

2.4. Stochastic Optimization: A Learning Algorithms Perspective

netic programming generalized parameter optimization to function optimization, initially coded

2.5. Important Topics We Have Not Covered: Bayesian Inference

3. FLOW MODELING WITH MACHINE LEARNING

3.1. Flow Feature Extraction

3.1.1. Dimensionality reduction: linear and nonlinear embeddings. A common approach

490 Brunton • Noack • Koumoutsakos

www.annualreviews.org • Machine Learning for Fluid Mechanics 491

Superresolution involves the inference of a high-resolution image from low-resolution mea-

3.2. Modeling Flow Dynamics

492 Brunton • Noack • Koumoutsakos

www.annualreviews.org • Machine Learning for Fluid Mechanics 493

3.2.3. Parsimonious nonlinear models. Parsimony is a recurring theme in mathematical

generalizability compared with other leading algorithms in ML.

494 Brunton • Noack • Koumoutsakos

4. FLOW OPTIMIZATION AND CONTROL USING

www.annualreviews.org • Machine Learning for Fluid Mechanics 495

OPTIMIZATION AND CONTROL: BOUNDARIES ERASED BY FAST COMPUTERS

4.1. Stochastic Flow Optimization: Learning Probability Distributions

4.2. Flow Control with Machine Learning

informed by sensor measurements. Feedback is necessary to stabilize an unstable system, attenu-

4.3. Flow Control via Reinforcement Learning

related to fluid mechanics.

Deep reinforcement learning scheme