EURO Journal on Computational Optimization (2020) 8:141–172

https://doi.org/10.1007/s13675-020-00121-0

ORIGINAL PAPER

A distributionally robust optimization approach

for two-stage facility location problems

Arash Gourtani1 · Tri-Dung Nguyen2 · Huifu Xu3

Received: 15 January 2016 / Accepted: 9 January 2020 / Published online: 4 February 2020
© The Author(s) 2020

Abstract
In this paper, we consider a facility location problem where customer demand consti-
tutes considerable uncertainty, and where complete information on the distribution of
the uncertainty is unavailable. We formulate the optimal decision problem as a two-
stage stochastic mixed integer programming problem: an optimal selection of facility
locations in the first stage and an optimal decision on the operation of each facility
in the second stage. A distributionally robust optimization framework is proposed to
hedge risks arising from incomplete information on the distribution of the uncertainty.
Specifically, by exploiting the moment information, we construct a set of distribu-
tions which contains the true distribution and where the optimal decision is based
on the worst distribution from the set. We then develop two numerical schemes for
solving the distributionally robust facility location problem: a semi-infinite program-
ming approach which exploits moments of certain reference random variables and a
semi-definite programming approach which utilizes the mean and correlation of the
underlying random variables describing the demand uncertainty. In the semi-infinite
programming approach, we apply the well-known linear decision rule approach to
the robust dual problem and then approximate the semi-infinite constraints through
the conditional value at risk measure. We provide numerical tests to demonstrate the
computation and properties of the robust solutions.

B Tri-Dung Nguyen
[email protected]
Arash Gourtani
[email protected]
Huifu Xu
[email protected]

1 School of Mathematics, University of Southampton, Southampton SO17 1BJ, UK

2 Schools of Mathematics and Management, University of Southampton, Southampton SO17 1BJ,
UK
3 Department of Systems Engineering and Engineering Management,
The Chinese University of Hong Kong, Hong Kong, China

123
142 A. Gourtani et al.

Keywords Distributionally robust optimization · Facility location problem ·

Semi-definite programming · Semi-infinite programming

Mathematics Subject Classification 90B80

1 Introduction

The classic discrete facility location problem (FLP) involves selecting a subset of
facility locations within a finite set of available locations and assigning customers to
the selected facilities with the aim to minimize the combined facility setup cost and
transportation cost. In the most basic form, the discrete FLPs consist of allocating p
facilities to a given list of candidate locations. In these so-called p-median problems,
the fixed setup cost of all candidate sites is assumed to be equal. The objective function
is only to minimize the total service cost to the customers, i.e., the transportation cost.
Under the non-homogeneity of the facilities’ setup cost, the p-median problem can be
extended to uncapacitated facility location problems (UFLP) in which the setup cost
is also added to the objective function. The UFLP assumes that facilities can serve an
unlimited amount of demand. However, in many practical problems, facilities have
capacity constraints and this leads to an important family of FLPs called capacitated
facility location problems (CFLP) in which the cheapest-assignment criterion is not
sufficient for the optimality of the solution. Note that in the p-median problem, the
number of facilities to be installed is fixed to p whereas, in UFLP and CFLP, no
such constraints are posed. The p-median, UFLP, and CFLP have been subjects of
extensive research, and interested readers might refer to Daskin (2011), Daskin (2008)
and ReVelle and Eiselt (2005) for some comprehensive reviews.
The aforementioned models share certain characteristics such as single-period plan-
ning horizon, single product and facility type, and deterministic parameters (i.e.,
demands, supplies, and costs). However, the deterministic assumption is one of the
major drawbacks in coping with many real-world problems. The strategic decisions
on facility setup are often capital-intensive, non-repetitive, and span a long time hori-
zon. The decision has to be made at present and hence is subject to risks arising from
uncertainties in demands and operations of the established facilities. Hedging the risk,
therefore, becomes a vital component of the decision-making process. The facility
location problem under uncertainty has attracted considerable attention recently; see,
for examples, reviews in Owen and Daskin (1998), Snyder (2006) and Lim et al. (2010).
Two major frameworks used to model uncertainty in the facility location problems are
stochastic optimization and robust optimization.
In the first framework, stochastic optimization has long been a well-known mathe-
matical method for finding optimal decisions under uncertainty. A key assumption in
this approach is that the decision maker has complete information on the distribution
of the uncertainty, through either empirical data or subjective judgment. However,
in some circumstances, this might turn out to be difficult if not impossible when a
strategic decision has to be made well in advance of the realization of the uncertainty.
In the second approach, the robust optimization framework, no assumption is made
on the probability distribution of the uncertainty. The traditional proposed measure of

123
A distributionally robust optimization approach for... 143

robustness is the minmax cost approach in which the cost associated with the worst
case scenario is minimized. Some of the examples of the minmax approach can be
found in Averbakh and Berman (1997), Averbakh and Berman (2000), Conde (2007)
and Snyder and Daskin (2006).
A feasible way to address the issue of distributional uncertainty in stochastic opti-
mization is to use the available data to construct a set of distributions which contains
the true distribution of the uncertainty and make an optimal decision on the basis of
the worst distribution from the set. This approach is known as distributionally robust
optimization which was proposed by Scarf et al. (1958) and has now been extensively
studied over the past few decades. [For some recent developments see Delage and Ye
(2010), Esfahani and Kuhn (2018), Wiesemann et al. (2014), Xu et al. (2018), Lu et al.
(2015), Santiváñez and Carlo (2018) and references therein.] How to construct the set
of distributions depends on the available information and there is no unified framework
for this. For examples, if there are some empirical data which allow one to construct
a nominal empirical distribution, then one may use the Kantorovich/Wasserstein met-
ric or φ-divergence to construct a ball of distributions (Esfahani and Kuhn (2018)
and Love and Bayraksan (2015)); if the information comes with statistical quantities
such as mean and variance, then one may use moment type conditions (Delage and Ye
(2010), Wiesemann et al. (2014) and Xu et al. (2018)). In this paper, we use the moment
approach. Specifically, we propose a distributionally robust optimization model for the
capacitated facility location problem to deal with future demand uncertainty. We then
propose two numerical schemes—namely a semi-definite program and a semi-infinite
program—to solve the distributionally robust optimization model. Finally, we provide
numerical results for medium size instances taken from the literature.
The remainder of this paper is organized as follows.
1. In Sect. 2, we formally describe the deterministic model, a two-stage stochastic
model and a distributionally robust formulation of the stochastic model.
2. We then proceed with discussions on numerical schemes in the following two
consecutive sections for solving the robust model depending on the availability of
information on the distribution of demands: a semi-definite programming (SDP)
scheme in Sect. 3 and a semi-infinite programming (SIP) scheme in Sect. 4.
3. In Sect. 5, we report numerical test results of the two schemes and draw some
conclusions in Sect. 6.

2 Facility location models

We first introduce the classic deterministic capacitated facility location (D-FLP) prob-
lem. By taking into account future uncertainty, we extend it to a two-stage stochastic
facility location (S-FLP) modeling framework, which then forms the basis for devel-
oping a distributionally robust facility location (R-FLP) model.

123
144 A. Gourtani et al.

2.1 Deterministic facility location model

There is a vast body of literature on the deterministic facility location problem. A good
review of the related literature is carried out by Owen and Daskin (1998) and Daskin
(2011). Suppose there are up to n facilities to be opened in a set of possible locations
I = {1, . . . , n}, indexed by i. Let J = {1, . . . , m} be the set of customers, indexed by
j. The location decision variable z i is defined as:

1 if the facility i is opened,
zi =
0 otherwise,

and continuous assignment variable xi j determines the service quantity or the trans-
portation quantity as we refer to it in this work, assigned from facility i to customer
j. Each facility i ∈ I has a fixed opening cost of bi and, if opened, has a service
capacity si that can be used by one or more customers. Likewise, each customer j ∈ J
is characterized by a demand d j that needs to be satisfied by one or more facilities. In
the deterministic framework, it is assumed that the customer demand and the capacity
of the facilities are known. The unit service cost of facility i for serving customer j is
denoted by ci j and assumed to be known; for example, transportation costs are pro-
portional to the distances between the facilities and the customers. Note that real-life
transportation problems are likely to be unbalanced; that is, the total demand might
exceed the total supply (Goyal 1984). To avoid the unbalance in the transportation
problem, we assume that an external supplier will serve the excess demand w j of
customer j. The unit service cost of this external supply facility, denoted by C j , is
assumed to be sufficiently large, C j ≥ maxi∈I ci j , so that the external supply is
invoked only when the total supply from all facilities fails to satisfy the demand.
The objective of the facility location problem is to minimize the total fixed cost
of opening the facilities and the future transportation costs while satisfying the cus-
tomer demand and supply capacity constraints. The deterministic FLP is formulated
as follows:
  
(D-FLP) min bi z i + ci j xi j + Cjwj
z,x,w
i i, j j

s.t. xi j + w j ≥ d j , ∀ j ∈ J , (2.1a)
i

xi j ≤ z i si , ∀ i ∈ I , (2.1b)
j
xi j ≥ 0, ∀ j ∈ J , i ∈ I , (2.1c)
w j ≥ 0, ∀ j ∈ J , (2.1d)
z i ∈ {0, 1}, ∀ i ∈ I . (2.1e)

Balance constraint (2.1a) ensures that the demands of all customers are met. Con-
straint (2.1b) prevents the service level assigned to each facility from exceeding its
capacity and also ensures that the customers cannot be served by un-built facilities

123
A distributionally robust optimization approach for... 145

(i.e., when z i = 0, xi j must be equal to zero). Finally, constraints (2.1c)–(2.1e) enforce

the nonnegativity of the service quantities and the binary nature of the facility allocat-
ing decisions, respectively. Variations of deterministic facility location formulations
similar to the D-FLP have been well studied in the literature. We refer the readers to
Daskin (2011) for more details about these.

2.2 Two-stage stochastic model

The facility location problems involve uncertainties that stem from unpredictability of
demand, supply and service costs. Since the facility location decisions are irreversible
and capital intensive, it is vital to take into account the future uncertainties when the
facility location decisions are made. Louveaux (1986) first introduced a two-stage
stochastic program with recourse for solving simple plant location problems and p-
median problems where uncertainties in demand, production and transportation costs
are considered. Wu et al. (2015) develop a distributionally robust model for the
uncapaciated facility location problem under demand uncertainty and derive some
nice reformulation of the stochastic model. We follow this approach but extend it to
the capacitated cases. There are several differences between our model and that in
Wu et al. (2015) which are described below. First of all, Wu et al.’s model is for the
uncapacitated case, which means each facility can serve as many customers as it wishes
as long as the total transportation distances are minimized as part of the objective
function. Their decision variables are which facilities to fully serve which customers
while our decision variables are to what extend each facility, if opened, serves each
customer. When the facilities have unlimited capacity, the demand uncertainty does
not play as much of an important role as when there is capacity limitation. This is
because, despite uncertain demand, each customer is still being fully served by the
nearest facility. In our case, it is possible that the demand exceeds the nearby facilities’
capacities and hence we have to buy and transport from one or more expensive sources.
To extend the deterministic model described in Sect. 2.1 to a stochastic setting,
customer demand is assumed to be stochastic with a known probability distribution.
Instead of having a deterministic demand vector d, we use the notation d(ξ ) for the
stochastic demands that depend on a random vector ξ . For convenience in notation, we
use d(ξ ) and ξ interchangeably; that is, both d j (ξ ) and ξ j refer to the stochastic demand
of customer j. The objective of the two-stage problem is to minimize the sum of fixed
investment cost of allocating the facilities and the expected future transportation costs.
The resulting mathematical model is given as follows,

  
(S-FLP) min i bi z i + E g(z, ξ )
z (2.2)
s.t. z i ∈ {0, 1}, ∀ i ∈ I ,

where g(z, ξ ) is the optimal value of the second-stage transportation problem

 
g(z, ξ ) := min ci j xi j + Cjwj,
x,w
i, j j

123
146 A. Gourtani et al.


s.t. xi j + w j ≥ d j (ξ ), ∀ j ∈ J , (2.3a)
i

xi j ≤ z i si , ∀ i ∈ I , (2.3b)
j
xi j ≥ 0, ∀ i ∈ I , j ∈ J , (2.3c)
w j ≥ 0, ∀ j ∈ J , (2.3d)

and the expectation is taken w.r.t. the distribution of random vector ξ . In this formula-
tion, the decision on z in the first stage determines the location of new facilities to be
built, before the realization of the uncertain demand d(ξ ), while in the second stage a
decision is made to specify allocation of transportation resources after the demand is
realized. The optimal value g(z, ξ ) of the second-stage problem therefore depends on
z and ξ .

2.3 Distributionally robust facility location model

One of the major difficulties which often arises in facility location problems is the lack
of complete information on the probability distribution of future customer demands
at different locations. We consider a setting where there might be limited information
on the probability distribution P of the random parameters ξ . Suppose that we are
able to construct a set of probability distributions, denoted by P, which contains
the true probability distribution P. In order to hedge the risk against ambiguity of
the true distribution, we may consider a robust model where the optimal decision
on the location of the new facilities to be built and the operation of all facilities is
based on the worst distribution from P. In the literature of robust optimization, P
is called the ambiguity set which reflects the fact that there is an ambiguity in the
true probability distribution in this setup. The corresponding distributionally robust
optimization problem can be formulated as
 
(R-FLP) min bT z + sup E P g(z, ξ ))
z P∈P (2.4)
s.t. z ∈ {0, 1}n .

One of the key elements in distributionally robust optimization is construction of

the ambiguity set. Over the past few decades, various statistical methods have been
proposed among which the method using moment information of the underlying prob-
ability distribution seems to be popular; see Delage and Ye (2010) and references
therein. Another widely adopted approach is a mixture distribution which uses a con-
vex combination of some observed and/or predicted distributions. In this paper, we
will consider the moment approach which seems to be relevant in the problem setting.
We will proceed our discussions through two distinct mathematical formulations: In
the first formulation, we assume the mean value and covariance of ξ are known and
consequently we reformulate problem (2.4) as a semi-definite program (SDP). In the
second formulation, we weaken the assumption by merely assuming the mean value
of ξ is known and subsequently reformulate problem (2.4) as a semi-infinite program

123
A distributionally robust optimization approach for... 147

(SIP). We then develop appropriate numerical procedures for solving the resulting
optimization problems.

3 A semi-definite programming approach

In this section, we first formulate the (R-FLP) problem as an integer semi-definite

program in Sect. 3.1. The resulting model, however, has a large number of SDP con-
straints in addition to the inherent binary variables for the facility location decisions.
We propose to resolve the integrality issue by using a generic genetic algorithm in
Sect. 3.2 and the large number of SDP constraints by using a row generation algo-
rithm in Sect. 3.3.

3.1 SDP formulation

We first investigate (R-FLP) with some moment information of ξ .

Let  denote the range of ξ and P() the set of all probability measures over .
We consider the following ambiguity set,
 
P = P ∈ P() : E P [ξ ] = μ, E P [ξ ξ T ] = Q , (3.1)

where μ and Q denote the first and second moments of ξ . In practice, the true
moments may be unknown. In the literature of distributionally robust optimization,
one often specifies a range for these statistical quantities; see for instance Delage and
Ye (2010). Assume here that both μ and Q can be approximated using empirical data.
Quantification of the difference between the ambiguity using true moments and the
one using sample average approximation is well documented in Sun and Xu (2015)
and Zhang et al. (2016). Its impact on the optimal value and optimal solutions can also
be found in these papers.
Let
 
H (z) = sup E P g(z, ξ ) . (3.2)
P∈P

Problem (3.2) is related to the classical problem of moments. Here, instead of finding a
feasible distribution P ∈ P, we want to find one which maximizes the expected value
of g(z, ξ ). For a discussion on the background of the problem of moments, interested
readers are referred to Landau (1987). Let M + denote the set of all nonnegative finite
measures on measurable space (, B). Then

123
148 A. Gourtani et al.

H (z) := sup g(z, ξ )P(dξ )

P∈M + 

s.t. ξi ξ j P(dξ ) = Q i j , ∀i, j = 1, . . . , m,


(3.3)
ξi P(dξ ) = μi , ∀i = 1, . . . , m,


P(dξ ) = 1,


where Q i j denotes the (i j)th component of Q, and μ j the jth component of μ.

Let Sm×m denote the space of m by m real matrices. Let Y ∈ Sm×m , y ∈ Rm and
y0 ∈ R denote the dual variables associated with the moment constraints in (3.3). We
can then write the Lagrangian dual problem of (3.3) as follows,

H D (z) := min Q • Y + μT y + y0
Y,y,y0 (3.4)
s.t. ξ T Yξ + ξ T y + y0 ≥ g(z, ξ ), ∀ξ ∈ Rm .

Here and later on A • B denotes the Frobenius inner product of matrices A and B.
This kind of dual formulation is well-known; see for instance Zymler et al. (2013) and
(Shapiro et al. 2009, Chapter 6) for general moment problems. It is easy to prove that
H (z) ≤ H D (z); see for instance Bertsimas and Sethuraman (2000).
For strong duality results to hold, we make the following assumption
Assumption 3.1 The linear matrix inequality Q − μμT  0 holds, where A  0
means A is positive definite.
Here and later on we write A  0 for matrix A being positive semi-definite. Note
that, for (μ, Q) to be valid first and second moments of some random variable, it is
necessary to have condition {Q − μμT  0} satisfied. Indeed, for any vector v, we
have

v T (Q − μμT )v = v T E (ξ − E[ξ ])(ξ − E[ξ ])T v

= E v T (ξ − E[ξ ])(ξ − E[ξ ])T v

 2
= E v T (ξ − E[ξ ]) ≥ 0,

since v T (ξ − E[ξ ]) is a scalar. Thus, the covariance matrix is positive definite unless
the uncertain sources ξ are linearly dependent. Assumption 3.1 is slightly stronger
as we replace positive semi-definiteness with positive definiteness, i.e., we explicitly
assume that there is no linear dependency among the sources of uncertainty. This
is needed for reasons of technicality in proving the strong duality result, which is
formally stated in the following proposition.
Proposition 3.1 Under Assumption 3.1, H (z) = H D (z).

123
A distributionally robust optimization approach for... 149

 
Proof Let us define H = (M, v) | M = M T , M  vv T . For any fixed (M, v) ∈ H,
there exists a symmetric matrix W such that W 2 = (M − vv T ). We can then construct
a random variable ξ = W X + v such that ξ has a mean value of v and a covariance
matrix of (M − vv T ), where X ∈ IRn is a random vector which follows the standard
multivariate normal distribution. It is easy to verify that ξ satisfies the following:

ξi ξ j P(dξ ) = Mi j , ∀i, j = 1, . . . , m,


ξi P(dξ ) = vi , ∀i = 1, . . . , m, (3.5)


P(dξ ) = 1.


In other words, using any choice of (M, v) ∈ H to replace (Q, μ) in the right-hand
side of (3.5) would lead to a feasible ξ . In addition, we can show that H is an open
set. Therefore, under Assumption 3.1, i.e., (Q, μ) ∈ H, we also have (Q, μ) to belong
to the interior of H. As a result, there exists a neighborhood B small enough around
(Q, μ) such that if we replace the right-hand side of (3.5) by any (M, v) ∈ B, the
system of equations (3.5) is still feasible (for some different ξ ). This is the sufficient
condition for having strong duality result to hold, i.e., H (z) = H D (z), as stated in
(Shapiro 2001, Proposition 3.4). 

From the strong duality result, problem (2.4) can be equivalently written as

min bT z + Q • Y + μT y + y0
z,Y,y,y0
s.t. ξ T Yξ + ξ T y + y0 ≥ g(z, ξ ), ∀ξ ∈ , (3.6)
z ∈ {0, 1}n .

Let us now write down the dual of the transportation problem described in prob-
lem (2.3),
 
g D (z, ξ ) = max j αjξj − i βi (si z i )
α,β
s.t. α j − βi ≤ ci j , ∀ i ∈ I , j ∈ J , (3.7)
αj ≤ Cj, ∀ j ∈ J,
α j , βi ≥ 0, ∀ i ∈ I , j ∈ J ,

where α j , j ∈ J , are dual variables associated with demand constraints (2.3a) and
βi , i ∈ I , are dual variables associated with supply constraints (2.3b). Observe that
problem (2.3) satisfies the Mangasarian-Fromovitz constraint qualification (MFCQ).
Thus the Lagrange multipliers of the problem are bounded and there exists a positive
number C such that the problem above is equivalent to the following,
 
(DTP) g D (z, ξ ) = max α, β j αjξj − i βi (si z i )
s.t. constraints of problem (3.7),
βi ≤ C , ∀i ∈ I .

123
150 A. Gourtani et al.

To ease the notation, let γ α := α, γ β := β, and γ := (γ α , γ β ). Let  denote the

feasible set of (DTP). It is easy to observe that  is a polyhedral in IR|I |+|J | where |I |
and |J | denote the cardinality of the index sets I and J , respectively.
It is easy to observe that  is a bounded polyhedral. In addition, the finiteness
number of vertices of  follows from Balinski and Russakoff (1984).
Let {γ 1 , . . . , γ N } denote the set of vertices. Using the notation introduced above,
we can rewrite (DTP) in a neater form,

g D (z, ξ ) = max γ (z) ξ T γ α − γ β T (s ◦ z)

(3.8)
s.t. γ ∈ ,

where (s ◦ z) denotes an m-dimensional vector with components si z i for i ∈ I . Com-

bining (3.6) and (DTP), we can recast the robust facility location problem (2.4) as a
semi-definite program through the following proposition.
Proposition 3.2 Let P be defined as in (3.1) and =IRm . Under Assumption 3.1, the
two-stage distributionally robust facility location problem (2.4) can be reformulated
as the following semi-definite optimization problem:
(R-SDP) min bT z + Q • Y + μT y + y0
z,Y,y,y0  
y0 + γ β T (s ◦ z) 21 (y − γ α )T
s.t. 1  0, ∀γ ∈ {γ 1 , . . . , γ N }
2 (y − γ α ) Y
z ∈ {0, 1}n .
(3.9)

Proof It follows from Proposition 3.1 that, under Assumption 3.1, H (z) = H D (z) and
problem (2.4) can be reformulated as problem (3.6). Note that the reformulation still
involves g(z, ξ ). Since g(z, ξ ) and g D (z, ξ ) are primal and dual LPs of each other and
since both of them are feasible (i.e., by setting x = 0, ω j = d j , ∀ j ∈ J in the primal
and = fi = 0 in the dual), strong duality result holds and we have g(z, ξ ) = g D (z, ξ ).
Thus, we can replace the second-stage transportation problem through its dual and
rewrite the constraint of problem (3.6) as
 
ξ T Yξ + ξ T y + y0 ≥ max γ ∈  ξ T γ α − γ β T (s ◦ z) , ∀ξ ∈ Rm . (3.10)

Since  is bounded with a finite set of extreme points {γ 1 , . . . , γ N }, the maximizer

of the LP on the R.H.S of (3.10) occurs at one of the extreme points. Thus, (3.10) can
be equivalently rewritten as

ξ T Yξ + ξ T y + y0 ≥ ξ T γ α − γ β T (s ◦ z), ∀ξ ∈ Rm , ∀γ ∈ {γ 1 , . . . , γ N }.

A simple rearrangement yields

 
min ξ T Yξ + ξ T (y − γ α ) + y0 + γ β T (s ◦ z) ≥ 0, ∀γ ∈ {γ 1 , . . . , γ N }.
ξ
(3.11)

123
A distributionally robust optimization approach for... 151

We can show that inequality (3.11) holds if and only if

 
y0 + γ β T (s ◦ z) 21 (y − γ α )T
 0, ∀γ ∈ {γ 1 , . . . , γ N }. (3.12)
2 (y − γ α )
1
Y

Here, it is very clear that (3.12) implies (3.11). For the reverse, suppose that (3.12)
does not hold, i.e., there exists γ and (q0 , q) such that
  
  y0 + γ β T (s ◦ z) 1 (y − γ α )T q0
q0 q T 2 < 0.
2 (y − γ α )
1 q
Y

We then need to show that (3.11) does not hold either. If q0 = 0, then we can construct
ξ = q/q0 and obtain ξ T Yξ + ξ T (y − γ α ) + y0 + γ β T (s ◦ z) < 0 which means that
(3.11) does not hold. If q0 = 0, we have q T Yq < 0. We can then construct ξ = δq
with sufficiently large δ such that δ 2 q T Yq + δq T (y − γ α ) + y0 + γ β T (s ◦ z) < 0
which also means that (3.11) does not hold.
Finally, we can replace the constraint in (3.4) with the SDP constraint (3.12) and
obtain the SDP (3.9). 

Notice that the derivation from inequality (3.11) to inequality (3.12) requires
 ≡ IRm . In practice, there is often some information on the bounds of the uncertain
parameters. For example the customer demand cannot take negative values. In order to
handle the indefinite set of constraints that appears in problem (3.6) for this case, we
will approximate the indefinite constraint with a finite set of semi-definite constraints
as shown next. 
Suppose the support set is specified as  = j∈J  j , where  j = [ξ j , ξ̄ j ] for all
j ∈ J and ξ and ξ are some given lower and upper bounds. For reasons of technicality
in proving strong duality, we make an assumption that there exists a random vector
X with support set  such that E[X ] = 0 and E[X X T ] = I , where I ∈ IRm×m
is the identity matrix. In practice, the support set of ξ may not be necessarily box
structured. In such a case, we might consider a box within the support set of ξ such
that the probability of ξ taking values outside the box is negligible. It is possible
to investigate the difference between the ambiguity sets before and after cut and its
impact on the optimal value and optimal decisions although we have not attempted to
do so in this paper.
Under this new assumption and Assumption 3.1, we can show that the strong duality
result still holds and the proof is very similar to that of Proposition 3.1. The only
difference is in the way that we construct the random variable X (i.e., instead of
choosing a multivariate normal random variable, we choose a random vector X such
that E[X ] = 0 and E[X X T ] = I ). It is noted that the new assumption can be relaxed
further by a proper scaling of the random variables ξ .
Once we have derived the strong duality result, problem (3.4) becomes

H D (z) = min Q • Y + μT y + y0
Y,y,y0 (3.13)
s.t. ξ T Yξ + ξ T y + y0 ≥ g(z, ξ ), ∀ξ ∈ ,

123
152 A. Gourtani et al.

where the only difference compared to problem (3.4) is that we have replaced the
semi-infinite constraint {∀ξ ∈ IRn } with {∀ξ ∈ }. Inequality (3.11) now become

min ξ T Yξ + ξ T (y − γ α ) + y0 + γ β T (s ◦ z) ≥ 0, ∀γ ∈ {γ 1 , . . . , γ N },
ξ ≤ ξ ≤ ξ̄
(3.14)

which is equivalent to

φ(γ ) ≥ 0, ∀γ ∈ {γ 1 , . . . , γ N }, (3.15)

where φ(γ ) is the optimal value of the following program

   
  y0 + γ β T (s ◦ z) 1 (y − γ α )T 1
min 1 ξ 2
ξ
1
(y − γ α ) Y ξ
2 
  1
s.t. 1 ξ V j ≤ 0, ∀ j ∈ J,
ξ
⎡ ⎤
ξ j ξ j v Tj
and where V j = ⎣ ⎦ , with I j denoting an m × m matrix with all elements
vj Ij
being 0 except 1 at ( j, j), and v j is an m-dimensional vector with all components are
equal to zero except for the jth element which is equal to −(ξ j + ξ j )/2. The S-lemma
(Derinkuyu and Pınar 2006) provides a sufficient condition for the nonnegativity of
the quadratic objective function over the quadratic inequalities corresponding to the
bounds. In other words, for the conditions (3.15) to be satisfied for each γ ∈ , it
suffices that there exists h ≥ 0 such that
  
y0 + γ β T (s ◦ z) 21 (y − γ α )T
+ h j V j  0,
2 (y − γ α )
1
Y
j∈J

where h j denotes the jth component of vector h. Consequently, the SDP problem
(3.9) can be reformulated as

min bT z + Q • Y + μT y + y0
z,Y,y,y0 ,h
 
y0 + γ β T (s ◦ z) 21 (y − γ α )T
s.t.
2 (y − γ α )
1
 Y (3.16)
+ j∈J h j V j  0, ∀γ ∈ {γ 1 , . . . , γ N },
h ≥ 0,
z ∈ {0, 1}n .

Remark: Problem (3.16) is very similar to problem (3.9) except for the newly intro-
duced decision variable h. If we restrict h = 0, then problem (3.16) is exactly the same

123
A distributionally robust optimization approach for... 153

as problem (3.9). Each h > 0 essentially enlarges the feasible domain of (Y, y, y0 ) in
problem (3.9) to a larger feasible domain of (Y, y, y0 ) in problem (3.16).

3.2 Genetic algorithm for finding z

From a computational perspective, problems (3.9) and (3.16) are complex to solve due
to both the presence of the binary variable z and the potentially large number of SDP
constraints.
The issue of having binary variables z is unavoidable as this is an intrinsic part of the
facility location problems, even in the deterministic case. For practical purposes, we
utilize a genetic algorithm (GA) to search for the optimal facility location variables.
To this end, we rewrite Problem (3.16) in a compact form on decision variable z as
follows,

min bT z + H D (z), (3.17)

z∈{0,1}n

where H D (z) is defined in Formulation (3.4) and has a SDP reformulation similar to
Formulation (3.16) except that z is not a decision variable and the fixed term bT z is
taken out of the objective function.
As long as we are able to evaluate the fitness function bT z + H D (z) for each z,
the optimization problem (3.17) can be embedded in a generic genetic algorithm.1
In this case, we view the decision variable z as genomes and the genetic algorithm
keeps updating them through operations such as mutation and crossover to find better
solutions. The genetic algorithm is a probabilistic search method that mimics the
biological model of natural selection. It applies the principle of “survival of the fittest”
to a population of potential solutions to produce progressively better solutions over
the generations. We refer the interested readers to Davis (1991) for more details about
the genetic algorithm.

3.3 Row generation algorithm for evaluating the fitness function

The fitness function is the optimal value of a SDP problem of a similar form as in
(3.16) with N semi-definite constraints, where N is the number of vertices of the DTP
polyhedra , that is the feasible space of the DTP. The challenge here is that N grows
exponentially as the problem size (n, m) increases (Balinski and Russakoff 1984). We
resolve this issue by utilizing a row generation (RG) algorithm. The RG algorithm
starts by including only a subset of constraints and solves the restricted problem. The
optimal solution obtained might violate some constraints of the original problem. The
next step is to identify such violated constraints which are then added to the restricted
problem in an iterative manner. This procedure is applied until there is no further
violating constraint. In that case, the optimal solution of the restricted problem is also
the optimal solution of the original problem.

1 Other local search technique can be used too.

123
154 A. Gourtani et al.

A key component of a row generation algorithm is to identify violating constraints.

Suppose,
 at iteration k, we solve
 problem (3.16) with the subset of SDP constraints and
(k) (k) (k) (k)
obtain Y , y , y0 , h . To find a violating constraint, we need to find γ ∈ 
such that

min ξ T Yξ + ξ T (y − γ α ) + y0 + γ β T (s ◦ z) < 0,
ξ ≤ ξ ≤ ξ̄

since this is the complementary condition of the feasibility constraint (3.14).

This can be done by solving

(k)
min ξ T Y(k) ξ + ξ T (y(k) − γ α ) + y0 + γ β T (s ◦ z) (3.18)
γ ∈,ξ ≤ ξ ≤ ξ̄

and checking whether the optimal value is less than zero, in which case we have
identified
 a violatingconstraint. Otherwise, we conclude that the relaxed solution
Y(k) , y(k) , y0(k) , h(k) is an optimal solution of the original SDP problem. Prob-
lem (3.18) is a non-convex quadratic optimization problem, i.e., still involving a
bilinear term in the objective function between decision variable ξ and γ . Despite
the NP-hardness nature of the problem, there are several numerical schemes for solv-
ing practical problems globally such as in Chen and Burer (2012), Bonami et al.
(2018), Gondzio and Yildirim (2018) and Xia et al. (2019). In our numerical scheme,
we develop a simple iterative approach to alternatively fix γ to solve the (convex)
quadratic program on ξ and then fix the newly found ξ to solve the linear program
on γ . While this approach  does not provide us a definite conclusion on the feasibility
(k)
of Y(k) , y(k) , y0 , h(k) if the local optimal (ξ, γ ) results in a nonnegative objective
value, it is still effective in identifying violating constraints when this results in a nega-
tive objective value. Here, we note that a global optimization procedure is only needed
in the very last iteration of the row generation algorithm to verify the feasibility of the
relaxed optimal solution.

4 A semi-infinite approach

In this section, we consider the case when the uncertainty set is defined through the
first moment2

P = {P ∈ P : E P [ξ ] = μ} , (4.1)

where μ is the true mean value of the random demand ξ .

We first formulate the (R-FLP) problem as an integer semi-infinite program (SIP)
in Sect. 4.1. Similar to the SDP formulation, the resulting model still has the inherent
2 It is possible to include the second moment of ξ and the mathematical derivation on strong duality results
still holds as shown in Sect. 3. The numerical scheme to follow would still be applicable (except the problem
size is larger). We exclude the second moment in this work for clarity.

123
A distributionally robust optimization approach for... 155

binary variables for the facility location decisions and hence we use the same genetic
algorithm as described in Sect. 3. The major additional challenge is that the resulting
model has an infinite number of constraints. We first utilize a linear decision rule
approximation approach in Sect. 4.2 to simplify the SIP model by restricting the
second-stage decision variables as linear functions of the uncertainties. We then utilize
the conditional value at risk approximations over the sets of infinite constraints. Finally,
we use sample approximation to approximate the semi-infinite programs, i.e., both the
original SIP and the linear decision rule approximation, and the CVaR formulation in
Sect. 4.4.

4.1 Semi-infinite formulation

Let us reconsider the inner maximization problem associated with the robust problem
(2.4)
 
H (z) := sup E P g(z, ξ ) .
P∈P

We can derive the dual formulation with respect to the moment condition as

H D (z) = min λ0 + λT μ
λ0 ,λ (4.2)
s.t. g(z, ξ ) ≤ λ0 + ξ T λ, ∀ξ ∈ ,

where  ⊂ Rm is the support set of ξ , λ ∈ Rm and λ0 ∈ R are the dual variables

associated with the moment constraints and the normalization constraint, respectively.
Since the support set  is infinite, the dual problem (4.2) is a linear semi-infinite
programming problem (SIP). It is important to note that (4.2) is a deterministic semi-
infinite programming problem. If  is structured, e.g., polyhedral or semi-algebraic
and g is linear or quadratic w.r.t. ξ , then through the well known S-lemma, the semi-
infinite system can be represented as an SDP; see for instance Zymler et al. (2013).
Here, we don’t assume any special structure as such. To avoid duality gap, we assume
that the regularity conditions specified in Shapiro et al. (2009) hold. Specifically, we
assume that the dual problem has a non-empty and bounded set of optimal solutions
and also the support set  is convex and compact. The second-stage maximization
problem in (2.4) can
be replaced by its dual as follows,

(R-SIP) min bT z + λ0 + λT μ
z,λ0 ,λ
s.t. g(z, ξ ) ≤ λ0 + ξ T λ, ∀ξ ∈ , (4.3)
z ∈ {0, 1}n ,

or equivalently

123
156 A. Gourtani et al.

min b T z + λ0 + λ T μ
z,x(·),w(·),λ0 ,λ
s.t. c• x(ξ ) + C•w(ξ
 ) ≤ λ0 + ξ T λ, ∀ξ ∈ ,
(4.4)
x(ξ ), w(ξ ) ∈ G(z, ξ ), ∀ξ ∈ ,
z ∈ {0, 1}n ,

where x(ξ ) ∈ Rn×m and w(ξ ) ∈ Rm are optimal transportation decisions for each
fixed z and for each realization of ξ , and where c is the matrix of transportation cost
coefficients and C is the cost vector of serving customers from the external source.
Moreover, G(z, ξ ) is the feasible regions associated with the second-stage problem
(2.3).

4.2 Linear decision rule approximation

One of the main challenges in solving the semi-infinite problem above is the depen-
dence of the second-stage transportation variables x(ξ ) and w(ξ ) on random variable
ξ . These “adjustable” variables are often referred to as decision rules, and their pres-
ence could often complicate the solution procedure. Formally, a decision rule x(ξ )
can be defined as a vector-valued function, mapping the random variables ξ ∈ Rm
with support set  into the decisions. The decision rule problem can be interpreted as
identifying the best decision x(ξ ) ∈ ⊂ X once ξ is observed, where X denotes the
set of all the mapping from  to Rn×m and a subset of X .
A tractable approximation scheme to deal with the decision rules is to restrict their
feasible set to the ones that have a functionality affine relation with the uncertain
random variables (that are affine functions of the uncertain data). This approach was
proposed by Ben-Tal et al. (2004) and was extended in Shapiro and Nemirovski (2005)
and Kuhn et al. (2011) to develop tractable numerical procedure for stochastic pro-
gramming problems. Here, we take the initiative to apply the linear decision rule (LDR)
approximation to problem (4.4); that is, we impose the dependence of transportation
decisions on the random demand to follow linear functions

x(ξ ) = Xξ + x0 ,
w(ξ ) = Wξ + w0 ,

where X ∈ R(nm×m) , W ∈ Rm×m ∈ Rn×m , x0 ∈ Rn×m , and w0 ∈ Rm . Consequently,

problem (4.4) can be approximated as

(R-LDR) min b T z + λ0 + λ T μ
z,X,x0 ,W,w0 ,λ0 ,λ
s.t. c • (Xξ ) + x0 + C•(Wξ ) + w0 ≤ λ0
 +ξ λ, ∀ξ ∈ , 
T
(4.5)
Xξ + x0 , Wξ + w0 ∈ G(z, ξ ), ∀ξ ∈ ,
z ∈ {0, 1}n .

123
A distributionally robust optimization approach for... 157

Note that here we are slightly abusing the notation in this formulation: ξ should
be understood as a parameter rather than a random variable. Indeed, it represents
a realization of the random vector ξ . The optimal value of the LDR approximation
problem will generate an upper bound on the optimal value of original robust problem
(4.4).

4.3 Conditional value at risk approximation

Having defined the LDR formulation of the original robust semi-infinite problem, we
approximate the first semi-infinite constraint with Conditional Value at Risk (CVaR)
and then approximate the latter through Monte Carlo sampling to reduce the number
of constraints. One of the main advantages of using CVaR is that it converts the semi-
infinite number of constraints into a single constraint. A recent study by Anderson
et al. (2014) has shown promising performance of CVaR approximation in dealing
with semi-infinite problems. The CVaR approximation method has been extensively
used in stochastic programming for approximating the chance constraints, and we
refer the readers to Sun et al. (2014) and Hong et al. (2011) for more details.
In the case of our LDR problem, we start by considering CVaR approximation of
the first semi-infinite constraints in problem (4.5). To ease the notation, let

h(X, x0 , W, w0 , λ0 , λ, ξ ) := c • (Xξ ) + x0 + C•(Wξ ) + w0 − λ0 − ξ T λ,

and Q := (X, x0 , W, w0 , λ0 , λ). The semi-infinite constraint of (4.5) can be written

as

sup h(Q, ξ ) ≤ 0.
ξ ∈

Let ξ̃ be a continuous random vector with support . Then supξ ∈ h(Q, ξ ) can be
approximated by CVaR of h(Q, ξ̃ ), which is defined as

CVaRβ (h(Q, ξ̃ )) := min β (Q, η),

η∈R

where

1
β (Q, η) := η + h(Q, ξ̃ ) − η +
P̃(d ξ̃ ),
1−β ξ̃ ∈

(τ )+ = max(0, τ ), and P̃ denotes the distribution of ξ̃ . Consequently, the CVaR

approximation of the semi-infinite constraint in problem (4.5) can be expressed as
follows:
  
1
min η + E (h Q, ξ̃ ) − η +
≤ 0. (4.6)
η∈R 1−β

123
158 A. Gourtani et al.

It is important to distinguish the expectation E[·] here from the expectation E[·] in
(S-FLP). The former should be understood as a mathematical expectation taken w.r.t.
any distribution of any random variable ξ̃ with support set . In other words, here,
the ξ̃ does not have to be identical to the ξ in (S-FLP). For example, we may set ξ̃ as
a random variable with uniform distribution over . Of course, the selection of ξ̃ and
its distribution will affect the quantity of CVaR of h and the rate of approximation to
its essential supremum.
Under some mild conditions, we can show that the error arising from the approx-
imation scheme does not have significant impact on the optimal value; see Anderson
et al. (2014). By replacing the constraint in the original LDR problem, we can write
the CVaR approximation problem as

(R-CVaR) min b T z + λ0 + λ T μ
η,z,X,x0 ,W,w0 ,λ0 ,λ   
s.t. η + 1−β E h(X, x , W, w , λ0 , λ, ξ̃ ) − η
1 0 0 ≤ 0,
  +

Xξ + x0 , Wξ + w0 ∈ G(z, ξ ), ∀ξ ∈ ,
z ∈ {0, 1}n .
(4.7)

Under the Slater constraint qualification of the LDR problem (4.5), we can demon-
strate, similar to (Anderson et al. 2014, Theorem 4) that the optimal solution of the
CVaR approximation problem (4.7) converges to optimal solution of the LDR problem
as β → 1.

4.4 Discretization through sampling

One of the well-known solution approaches for semi-infinite programs is random

discretization. The basic idea is to construct a tractable sub-problem by considering
a randomly drawn finite subset of constraints and hence enlarging the solution set.
Calafiore and Campi (2005), Calafiore and Campi (2006) investigated this approach
and used Monte Carlo sampling (often referred to as sample average approximation)
to approximate the convex problems consisting of linear objectives and semi-infinite
constraints. They showed that the resulting randomized solution fails to satisfy only
a small proportion of the original constraints for a sufficiently large sample size. An
explicit bound on the measures of the original constraints that may be violated by the
randomized solution is derived. The approach has been shown numerically efficient,
and it has been widely applied to various stochastic and robust programs. We refer
interested readers to Campi and Garatti (2011), Shapiro (2003) and references therein.
In this paper, we apply the Monte Carlo sampling approach respectively to the
original semi-infinite problem (4.4), its LDR approximation (4.5) and the CVaR for-
mulation (4.7).
Let K = {1, . . . , K } denote the finite set of sample indices and ξ 1 , . . . , ξ K an
independent and identically distributed (i.i.d) sampling of ξ . We may construct a
discretized approximation of problem (4.4) through Monte Carlo sampling as follows,

123
A distributionally robust optimization approach for... 159

min b T z + λ0 + λ T μ
z,x(ξ k ),w(ξ k ):k∈K,λ0 ,λ

s.t. c • x(ξ k ) + C•w(ξ k ) ≤ λ0 + λT ξ k , ∀k ∈ K,

  (4.8)
x(ξ k ), w(ξ k ) ∈ G(z, ξ k ), ∀k ∈ K,

z ∈ {0, 1}n .

This kind of discretization scheme was recently applied to a distributionally robust

formulation of a two-stage stochastic unit commitment problem. From a mathemat-
ical perspective, it is justified in that under some mild conditions, one can show
the convergence of the optimal value of the discretized problem to its true coun-
terpart almost surely as the sample size increases, see details in (Xu et al. 2018,
Sect. 3.1).
Similarly the discretized LDR problem (4.5) can be formulated as

min b T z + λ0 + λ T μ
z,X,x0 ,W,w0 ,λ0 ,λ

s.t. c • (Xξ k ) + x0 + C•(Wξ k ) + w0 ≤ λ0 + λT ξ k , ∀k ∈ K,

  (4.9)
Xξ k + x0 , Wξ k + w0 ∈ G(z, ξ k ), ∀k ∈ K,

z ∈ {0, 1}n ,

and, finally, we apply the sample average approximation (SAA) scheme to CVaR
approximation problem (4.7) as follows,

min bT z + λ0 + λT μ
η,z,X,x0 ,W,w0 ,λ0 ,λ
  
1 K
s.t. η + (1−β)K k=1 h(X, x0 , W, w0 , λ0 , λ, ξ k ) − η ≤ 0,
+ (4.10)
 
Xξ k + x0 , Wξ k + w0 ∈ G(z, ξ k ), ∀k ∈ K,

z ∈ {0, 1}n .

Compared to (4.8), the CVaR approximation scheme allows one to take a few samples
at the tail rather than essential superemum of h and in that way smooth up or stabilize
the numerical computation; see Anderson et al. (2014). In the case of CVaR formula-
tion, we replace the CVaR constraint with the equivalent system of linear inequalities
by introducing additional positive dummy variables θ 1 , . . . θ k as follows

⎧ K
⎪
⎪η +
1
k=1 θ ≤ 0,
k
⎨ (1−β)K

⎪ h(X, x0 , W, w0 , λ0 , λ, ξ k ) − η ≤ θ k , ∀k ∈ K, (4.11)
⎪
⎩ k
θ ≥ 0, ∀k ∈ K.

123
160 A. Gourtani et al.

The substitution results in

min b T z + λ0 + λ T μ
θ,η,z,X,x0 ,W,w0 ,λ0 ,λ
K
s.t. η+ 1
(1−β)K k=1 θ ≤ 0,
k

h(X, x0 , W, w0 , λ0 , λ, ξ k ) − η
≤ θ k , ∀k ∈ K,
  (4.12)
Xξ k + x0 , Wξ k + w0 ∈ G(z, ξ k ), ∀k ∈ K,

θ k ≥ 0, ∀k ∈ K
z ∈ {0, 1}n .

The reformulation will effectively address the non-smoothness caused by the (.)+ oper-
ation but at the cost of introducing additional variables and constraints. It is worthwhile
to do that here as the latter formulation will result in an overall MILP.

5 Computational results

In this section, we report the numerical experiments performed to evaluate the pro-
posed methodologies. We have used MATLAB R2015b with CPLEX 12.6 for solving
transportation problems and quadratic optimization problems while SEDUMI 1.3 was
used for solving SDPs.
We first provide numerical results on a small-case study in Sect. 5.1 with three
facilities and four customers. The purpose of this part is to illustrate the performance
of various components of the SDP and SIP models such as the constraint generation
algorithm and the CVaR approximation. We also use this example to demonstrate how
the sample size affects the performance of the SIP model. We then report the numerical
results for medium test instances in Sect. 5.2. Here we report the computational times
for some test instances in the literature. We also study the robustness of the proposed
SDP and SIP solutions by varying the realized demand distributions. For convenience
in recapping these methods, Table 1 provides a quick reference on their abbreviations
and the key differences among them.

5.1 A small case study

We study a small-scale facility location problem to illustrate the quality of solutions

obtained from the proposed solution methods. We randomly generate a facility location
problem with 4 demand nodes and 3 potential locations to build facilities. The trans-
portation costs are assumed to be proportional to the distances between the customers
and the facilities. Figure 1 shows the network layout of this problem.
The transportation cost, fixed investment cost and the capacity of each potential
facility are given in Table 2.
We assume that customer demands are unknown prior to the construction of
facilities and we are only given the first moment information of the demand with

123
A distributionally robust optimization approach for... 161

Table 1 Abbreviations and references of methodologies used

Abbreviation Problem Methods Known information on demand

D-FLP (2.1a) Deterministic FLP Actual values (d)

S-FLP (2.2) Stochastic FLP Probability distribution (P)
R-SDP (3.16) Robust FLP with semi-definite First and second moments (μ, Q)
formulation
R-SIP (4.4) Robust FLP with semi-infinite
formulation
R-LDR (4.5) LDR approximation of R-SIP First moment (μ)
problem
R-CVaR (4.7) CVaR approximation of R-LDR
problem

1.2

Customer Locations
Potential Facility Locations
1

2
4

0.8

3
0.6

0.4

0.2 3

1
0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Fig. 1 The network of facility location problem

Table 2 Transportation and investment costs and capacity of the facilities

Supply points Demand points Supply capacity Fixed initial cost

dem 1 dem 2 dem 3 dem 4

sup1 14 12 21 25 200 2000

sup2 14 18 16 16 300 3200
sup3 17 10 14 19 254 3700
External source 27 27 27 27 ∞ 0

123
162 A. Gourtani et al.

Table 3 Worst expected cost associated with each possible facility location decision

Facility Possible FL decisions

Decision sol1 sol2 sol3 sol4 sol5 sol6 sol7 sol8

z1 0 0 0 0 1 1 1 1
z2 0 0 1 1 0 0 1 1
z3 0 1 0 1 0 1 0 1
Cost 100,132.1 56,012.3 47,851.8 13,723.6 64,639.2 21,349.9 13,829.1 15,507.8

μ = (150, 150, 100, 100). Moreover, in the SDP formulation of the problem, we are
also provided with the second moment information of the uncertain demand

⎛ ⎞
22669.34 22511.07 15038.4 15026.78
⎜ 22511.07 22551.57 15000.53 14988.58 ⎟
⎜
Q=⎝ ⎟
15038.4 15000.53 10045.48 10013.84 ⎠
15026.78 14988.58 10013.84 10031.04

Under the assumption of available first and second moments, we formulate the
problem as a robust SDP (R-SDP). Since the problem size is small, we first consider
all combinations of the possible facility location decisions (23 possible combinations
of z). For each potential solution, the full R-SDP problem (3.16) is constructed by
including all of the extreme points of the dual transportation polytopes [obtained by
using the signature method in Balinski and Rispoli (1993)]. The full R-SDP problem
is then solved for each potential solution and comparisons between the total cost of
each decision are provided in Table 3.
Note that the very high costs of certain facility location solutions, such as with
z = [0, 0, 0] or z = [0, 0, 1], are due to the high costs of using the external facility
when the total (random) demand exceeds the capacity of the built facilities. It can be
observed that solution number 4, i.e., z = (0, 1, 1) has the least cost and therefore is
optimal (and is highlighted in bold in Table 3).
In order to illustrate the performance of the proposed constraint generation algo-
rithm for the R-SDP formulation, we first fixed the facility decisions to z = (1, 1, 1). In
each run, the problem was initiated by randomly selecting one of the SDP constraints
corresponding to one of the extreme points of the dual transportation problem. We
recorded the objective value after each iteration of the constraint generation process.
Figure 2 summarizes the objective value after each iteration of the algorithm (by tak-
ing the average over 100 runs, each with a different starting point). It can be observed
that the convergence of the constraint generation algorithm to the optimal value takes
place after around 15 iterations.
After finding the optimal solution for a fixed facility location decision z, the next
step of the proposed solution method is to find the optimal z using GA. However, this
is not necessary for this small instance as it has only 8 possible solutions.

123
A distributionally robust optimization approach for... 163

4
x 10
1.551

1.55

1.549
Objective value (cost)

1.548

1.547

1.546

1.545
True optimal solution (full problem)
Average cost of 100 runs

1.544
0 2 4 6 8 10 12 14 16 18 20
Constraint generation iteration

Fig. 2 Convergence of the constraint generation algorithm solutions

Table 4 Deterministic versus

Models Optimal FLP decisions Total cost
robust solutions
z1 z2 z3

D-FLP 1 1 0 12,300.00
R-SDP 0 1 1 13,723.61
R-SIP 1 1 1 15,641.42
R-LDR 1 1 1 15,781.42

5.1.1 Robust SIP formulation

Let us assume that we are only given the first moment information μ and the second
moment Q is unknown. We implement the second proposed method and formulate the
FLP as a semi-infinite program. For assessing the quality of LDR approximation (R-
LDR) of the “true” robust SIP (R-SIP) solution, we limit the support set of the random
demand for each customer to 2000 values generated from the uniform distribution
U(0, 250). We then solve the full R-SIP and its R-LDR approximation over this support
set. As shown in Table 4, the R-LDR solution provides an upper bound approximation
with 0.9% deviation from the original R-SIP solution. For comparison purposes, we
also solve the deterministic version of the instance by assuming that the demand values
are known and given by μ. The solution from the deterministic problem (D-FLP) is
then benchmarked against the robust solution in Table 4.
The deterministic solution is to install just enough capacity to meet the predicted
(assumed to be known) demand values by locating facilities 1 and 2. In other words,
the D-FLP solution provides no flexibility for possible variation in future demand.
The robust solutions, on the other hand, offers to install the facilities with a higher
total capacity at a higher total cost (i.e., constructing facilities 2 and 3 in the R-SDP

123
164 A. Gourtani et al.

4
x 10
1.6

1.5

1.4
Objective value

1.3

1.2

1.1
CVaR approximation of LDR
LDR-SIP

1
0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
Beta value

Fig. 3 CVaR approximation of R-LDR problem for various β values

case and all facilities in the R-SIP case), to hedge against the risk of not meeting the
customer demand.
In the next step, we implement the proposed CVaR approximation (R-CVaR) of
the R-LDR solution. Using the same support set of 2000 values, we solved the R-
CVaR approximation with various β values. The results are presented in Fig. 3. It
can be observed that the CVaR solution approximates the R-LDR optimal solution
consistently and without any error for β ≥ 0.7.

5.1.2 Discretization through sampling

The complexity of the R-LDR and R-CVaR schemes increases as the number of scenar-
ios increases. As described in Sect. 4.4, one way to resolve this issue is to use sample
approximation (SA) for R-LDR and sample average approximation for R-CVaR. These
involve drawing i.i.d samples from the underlying distribution of the uncertainty. In
the case of the first instance, we have chosen the samples from the same support set
used to run the full R-SIP tests. To assess the quality of sample approximation, we
solve R-SIP, R-DLR and R-CVaR using various sample sizes. For each sample size,
we carried out 1000 independent runs. The β value for all of the CVaR instances was
set to 0.99. The normalized deviation of the sample approximations of all problems
from the true robust (full R-SIP) solution is shown in Fig. 4 for various choices of the
sample sizes.
It can be observed that, for each sample size, the mean deviation of R-CVaR and R-
LDR solutions from the true robust solution is very similar. They range from -0.4% to
0.9% of the true solution. (Here 0.9% deviation means that the approximate solution is
0.9% higher than the true value of the robust solution.) We can also see that the sample
approximation of R-CVaR and R-LDR converges to the full R-LDR solution as the
sample size increases. Furthermore, the sampling method applied to R-SIP provides
a very good approximation of the true solution even for small sample sizes.

123
A distributionally robust optimization approach for... 165

1
0.9
% Deviation from true R-SIP solution

% Deviation from true R-SIP solution

0.8

0.7
0.6

0.5
0.4

0.3

0.2
0.1

0 0
−0.1
R-SIP R-SIP
R-LDR −0.2 R-LDR
−0.3 Mean Deviation of Mean Deviation of
R-CVaR (SAA) R-LDR (SA)
−0.4
−0.5
90 100 200 500 800 1500 1800 1900 90 100 200 500 800 1500 1800 1900
Sample size Sample size
(a) SAA of R-CVaR (b) SA of R-LDR

−0.05
% Deviation from true R-SIP

−0.1

−0.15

−0.2

−0.25

−0.3

−0.35

−0.4

R-SIP
−0.45
Mean Deviation (SA)
90 100 200 500 800 1500 1800 1900
Sample size

(c) SA of R-SIP

Fig. 4 Normalized deviation of approximation problems from the true robust solution

25
CVAR
LDR
Average CPU time (seconds)

20 R-SIP

15

10

0
90 100 200 500 800 1500 1800 1900
Sample size

Fig. 5 Average computation time

The average computation times for 1000 runs of each sample size are shown in the
graph below (Fig. 5).

123
166 A. Gourtani et al.

5.2 Medium size test instances

In this section, we consider a set of larger facility location problems. Test instances
are selected from those existing in the literature. We have modified and used test prob-
lems presented by Díaz and Fernández (2002) for the single-source capacitated facility
location problem.3 The first 6 instances consist of networks of 10 potential facility
locations and 20 customer demand points. The network size is increased twice in the
last instance to observe how the computational time increases. We have used the given
demand values in test instances as the first moment of the customer demand distribu-
tion. The second moment matrix for the SDP formulation was randomly generated.4

5.2.1 SDP formulation

The test results for the medium size instances are presented in Table 5. The first two
columns show the test instances from Díaz and Fernández (2002) and their correspond-
ing network sizes. The third and fourth columns show the optimal solution z and the
corresponding SDP objective values. Columns 5–7 show the CPU computational time
while column 8 shows the optimal value of the corresponding deterministic facility
location problem as defined in Model 2.
We can see from column 7 that it took between 2–4 hours to run the instances with
20 customers and 10 facilities. This increases sharply to over 11 hours for instance
p7 and 22 hours for instance p18 when the number of customers and facilities are
increased to (30, 15) and (40, 20), respectively. The total CPU times are broken down
to the time to (iteratively) run the SDPs and to perform row generations (i.e., to identify
violating constraints). We can see that most of the computational time is on solving
the SDPs. In theory, the SDP models (3.9) and (3.16) can be solved in polynomial
times for fixed numbers of facilities and customers since the problem has a linear
objective function with a polynomial number of SDP constraints. While we have made
a contribution in transforming the two-stage stochastic, minimax, integer optimization
problem into a single-level SDP optimization problem, we acknowledge that there are
still drawbacks in the framework. Specifically, despite recent developments in SDP
programming, problems (3.9) and (3.16) do not scale well with the number of facilities
and the number of customers due to the fast growth in the number of SDP constraints.
For practical purposes with larger instances, if being able to find a reasonably robust
solution is more crucial than optimality, our framework is still applicable and what
we need to do is to set the time limits on the constraint generation algorithm and the
genetic algorithms.
The numerical scheme for solving the SDP contains several fine tuning parameters.
First of all, the stopping condition for the row generation algorithm is triggered either

3 Available at http://www-eio.upc.edu/~elena/sscplp/index.html. Here, we note that the facility location

model in Díaz and Fernández (2002) is slightly different from Model 2 in that the authors restrict one facility
to each customer, which is not appropriate in our case due to demand uncertainty. In addition, our model
includes an external source which can be used if the (uncertain) demand exceeds the supply. The serving
cost from the external source is set as C = 1.2 max i, jci j .
4 We use default Matlab function “gallery(’randcorr’,m)” to generate the correlation matrix and the standard
deviations is set equal to 0.04 times the first moments.

123
 Table 5 Robust facility location solutions

Test # Size (m, n) Optimal robust decisions (z) SDP solution CPU (s) D-FLP
SDP CG Total

p1 (20, 10) [1 1 0 1 1 0 1 1 1 0] 1814.62 14,251 1789 16,143 1810.86

A distributionally robust optimization approach for...

p2 (20, 10) [1 0 0 1 0 0 1 1 1 1] 3903.44 11,563 1673 13,337 3915.73

p3 (20, 10) [0 0 0 0 0 0 0 0 0 0] 4218.68 7028 737 7858 4218.6
p4 (20, 10) [0 0 0 1 0 0 0 1 0 0] 4746.66 7223 830 8136 4746.42
p5 (20, 10) [1 1 0 0 1 1 1 0 1 1] 4377.84 13,012 1685 14,797 4375.87
p6 (20, 10) [1 1 0 1 1 1 1 1 1 0] 2221.55 13,441 1725 15,266 2172.59
p7 (30, 15) [0 1 0 1 1 1 0 0 1 0 1 0 0 1 0] 4419.46 39,980 2262 42,343 4166.64
p18 (40, 20) [ 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0] 8822.93 81,427 2389 83,924 8624.18

123
167
168 A. Gourtani et al.

when a maximum of 100 SDP constraints are added or when the additional constraints
do not lead to more than 0.1% changes on the objective value in the 5 consecutive
iterations. Second, the number of iterations in the alternating procedure for finding
violating constraints is set at most 100 iterations. We implement the constraint genera-
tion and utilize the built-in Matlab GA algorithms with mostly default settings except
that the GA population size is set to 30 with stopping criteria of a maximum of 10
generations. The population size is the number of solutions under consideration at any
point by the GA algorithm to perform various operations in order to produce further
solutions. The parameters are set to balance between the run time and the quality of
the solutions. All the numerical tests are run in MATLAB R2015b utilizing SeDuMi
and Cplex LP solvers version 12.6.
For comparison purposes, we also include the deterministic solutions of the
instances in the table above on column 8. As expected, most of the robust solutions
have higher costs than deterministic facility decisions as a result of installing higher
total capacities, although the lower cost of D-FLP solutions comes at the expense
of non-flexibility of the facility decisions against the fluctuation in future customer
demand. The only exception is on instance p2 . Here, we note that the row generation
algorithm have some stopping criteria and the reported SDP solutions might not be
the “true optimal” solutions should we have enough computational power to solve the
big SDPs with all the constraints included.
In order to verify the performance of GA, we compare its solutions with the
exhaustive search technique where the objective values for every combination of z
are evaluated in order to find the optimal solutions. This exhaustive search is only
doable when n is small enough, i.e., n = 10 in the first 6 instances. We found that the
solution obtained from GA matched the optimal solution in the first 5 instances while
the optimality gap for instance p6 is quite small at 1.5%.5

5.2.2 SIP formulation

We now reconsider the problem instance p5 and assume that the only available infor-
mation on demand uncertainty is the first moment of the distribution μ. The R-SIP
framework is then used to construct and solve this problem. As before, we limit the
support set of the random variable ξ to 200 uniformly generated random values. For
comparison purposes, we summarize the full R-SIP solution to this problem along
with those from D-FLP, S-FLP and full R-SDP versions of the problem in Table 6
below.
It can be observed that R-SIP offers a more flexible solution by installing a higher
level of supply capacity. This, of course, comes at the expense of a higher total expected
cost. Also, as we include progressively less information on the uncertainty in our
models, the solution becomes more and more robust (higher levels of supply capacity
installed), which is intuitively sensible.

5 Despite the consistent performance of GA in finding the optimal solution in these instances, in theory an
optimal solution cannot be guaranteed. However, in practice, the “local” solutions obtained using GA are
often of high quality. For instance p3 , the robust solution coincides with the deterministic solution at z = 0.
This is because the facility building fixed costs dominate the serving costs from the external source.

123
A distributionally robust optimization approach for... 169

Table 6 FLP solutions for various methodologies

Model Available Optimal FLP decisions Total Installed

information cost capacity
z1 z2 z3 z4 z5 z6 z7 z8 z9 z 10

D-FLP d 1 1 0 0 0 1 1 1 1 1 4417.26 458

S-FLP P 1 1 0 1 1 1 1 0 1 0 4495.33 462
R-SDP (μ, Q) 1 1 0 0 1 1 1 1 1 0 4603.65 481
R-SIP μ 1 1 0 0 1 1 1 1 1 1 4938.95 511

1
0.5

% Deviation from true R-SIP solution

% Deviation from true R-SIP solution

0
−1

−2
−0.5
−3

−1 −4

−5
−1.5
−6

R-SIP −7 R-SIP
−2 R-LDR
R-LDR
Mean Deviation of −8 Mean Deviation of
R-CVaR (SAA) R-LDR (SA)
−2.5 −9
50 100 50 100
Sample size Sample size

(a) SAA of R-CVaR (b) SA of R-LDR

% Deviation from true R-SIP solution

−1

−2

−3

−4

−5

−6

−7

−8
R-SIP
Mean Deviation(SA)
−9
50 100
Sample size

(c) SA of R-SIP

Fig. 6 Deviation of approximation solutions from the true robust solution

Having solved the full R-SIP p5 , we consider sample approximation of the R-SIP,
R-LDR and R-CVaR. Various sample sizes are used for each problem, and we repeat
each test for 100 times. The normalized deviation from the full R-SIP solution is
computed and presented in Fig. 6.
It can be observed that, in all models, the sampling scheme results in approximation
of the true robust solution (R-SIP) with a very low error (with < 3% mean deviation).
Moreover, in the case of smaller sample sizes, the CVaR model performs much better
than the other two models and has a mean deviation of < 0.5%.

123
170 A. Gourtani et al.

6 Conclusion

In this paper, we consider a capacitated facility location problem with customer

demand uncertainty and propose a two-stage distributionally robust model for the
problem to tackle the issue of incomplete information on the true distribution of the
uncertainty. We construct the uncertainty set using the moments information associated
with the distribution of the random demands. Two numerical methods are proposed
based on the available moment information. Specifically, we first formulate the robust
problem as a semi-definite program on the basis of the first and the second moments
which is then solved by using a constraint generation algorithm. Moreover, we for-
mulate the robust problem as a semi-infinite program for the case that only the first
moment information is given. The semi-infinite program is then solved by approxima-
tion using a linear decision rule, CVaR and Monte Carlo sampling. Finally, we carry out
numerical tests for a small instance and also for some medium-sized instances taken
from the literature. In each case, the distributionally robust solutions offer the flexi-
bility in hedging against uncertainty compared to the deterministic and the stochastic
solutions.
In the future, it would be interesting to study the possibility of extending the results
and methodologies presented in this paper to include uncertainty in supply, multistage
problems, and the competition in the market. Also, we would like to explore the
problem structure to enhance the solution algorithms for a better performance in large-
scale instances. Another open direction is to apply the proposed methodologies and
numerical schemes to the practical problems with similar structure and characteristics
such as uncertain supply and demand. Some examples related to the energy industry
could include the wind farm site location problem and liquefied natural gas storage
facility location problems.

Acknowledgements Thanks are due to two anonymous reviewers for their valuable comments. The second
author acknowledges the funding support from the Engineering and Physical Sciences Research Council
(Grant EP/P021042/1).

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which
permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give
appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence,
and indicate if changes were made. The images or other third party material in this article are included
in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If
material is not included in the article’s Creative Commons licence and your intended use is not permitted
by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the
copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

References
Anderson E, Xu H, Zhang D (2014) Confidence levels for cvar risk measures and minimax limits. Business
Analytics Working Paper Series, University of Sydney, Australia
Averbakh I, Berman O (1997) Minimax regret p-center location on a network with demand uncertainty.
Locat Sci 5(4):247–254
Averbakh I, Berman O (2000) Minmax regret median location on a network under uncertainty. INFORMS
J Comput 12(2):104–110

123
A distributionally robust optimization approach for... 171

Balinski ML, Rispoli FJ (1993) Signature classes of transportation polytopes. Math Program 60(1–3):127–
144
Balinski ML, Russakoff A (1984) Faces of dual transportation polyhedra. In: Mathematical programming
at Oberwolfach II. Springer, Berlin, pp 1–8
Ben-Tal A, Goryashko A, Guslitzer E, Nemirovski A (2004) Adjustable robust solutions of uncertain linear
programs. Math Program 99(2):351–376
Bertsimas D, Sethuraman J (2000) Moment problems and semidefinite optimization. In: Handbook of
semidefinite programming. Springer, Berlin, pp 469–509
Bonami P, Günlük O, Linderoth J (2018) Globally solving nonconvex quadratic programming problems
with box constraints via integer programming methods. Math Program Comput 10(3):333–382
Calafiore G, Campi MC (2005) Uncertain convex programs: randomized solutions and confidence levels.
Math Program 102(1):25–46
Calafiore GC, Campi MC (2006) The scenario approach to robust control design. IEEE Trans Autom Control
51(5):742–753
Campi MC, Garatti S (2011) A sampling-and-discarding approach to chance-constrained optimization:
feasibility and optimality. J Optim Theory Appl 148(2):257–280
Chen J, Burer S (2012) Globally solving nonconvex quadratic programming problems via completely
positive programming. Math Program Comput 4(1):33–52
Conde E (2007) Minmax regret location-allocation problem on a network under uncertainty. Eur J Oper
Res 179(3):1025–1039
Daskin MS (2008) What you should know about location modeling. Naval Res Logist 55(4):283–294
Daskin MS (2011) Network and discrete location: models, algorithms, and applications. Wiley, London
Davis L (1991) Handbook of genetic algorithms. Van Nostrand Reinhold Company. ISBN-10: 0442001738
Delage E, Ye Y (2010) Distributionally robust optimization under moment uncertainty with application to
data-driven problems. Oper Res 58(3):595–612
Derinkuyu K, Pınar M (2006) On the s-procedure and some variants. Math Methods Oper Res 64(1):55–77
Díaz JA, Fernández E (2002) A branch-and-price algorithm for the single source capacitated plant location
problem. J Oper Res Soc 53(7):728–740
Esfahani PM, Kuhn D (2018) Data-driven distributionally robust optimization using the wasserstein metric:
performance guarantees and tractable reformulations. Math Program 171(1–2):115–166
Gondzio J, Yildirim EA (2018) Global solutions of nonconvex standard quadratic programs via mixed
integer linear programming reformulations. arXiv preprint arXiv:1810.02307
Goyal SK (1984) Improving vam for unbalanced transportation problems. J Oper Res Soc 35(12):1113–1114
Hong LJ, Yang Y, Zhang L (2011) Sequential convex approximations to joint chance constrained programs:
a monte carlo approach. Oper Res 59(3):617–630
Kuhn D, Wiesemann W, Georghiou A (2011) Primal and dual linear decision rules in stochastic and robust
optimization. Math Program 130(1):177–209
Landau HJ (1987) Moments in mathematics, vol 37. American Mathematical Society, Providence
Lim M, Daskin MS, Bassamboo A, Chopra S (2010) A facility reliability problem: formulation, properties,
and algorithm. Naval Res Logist 57(1):58–70
Louveaux FV (1986) Discrete stochastic location models. Ann Oper Res 6(2):21–34
Love D, Bayraksan G (2015) Phi-divergence constrained ambiguous stochastic programs for data-driven
optimization. Technical report, Department of Integrated Systems Engineering, The Ohio State Uni-
versity, Columbus, Ohio
Lu M, Ran L, Shen JM (2015) Reliable facility location design under uncertain correlated disruptions.
Manuf Serv Oper Manag 17(4):445–455
Owen SH, Daskin MS (1998) Strategic facility location: a review. Eur J Oper Res 111(3):423–447
ReVelle CS, Eiselt HA (2005) Location analysis: a synthesis and survey. Eur J Oper Res 165(1):1–19
Santiváñez J, Carlo HJ (2018) Reliable capacitated facility location problem with service levels. EURO J
Transp Logist 7(4):315–341
Scarf H, Arrow KJ, Karlin S (1958) A min-max solution of an inventory problem. Stud Math Theory
Inventory Prod 10:201–209
Shapiro A (2001) On duality theory of conic linear problems. In: Semi-infinite programming. Springer,
Berlin, pp 135–165
Shapiro A (2003) Monte carlo sampling methods. Handb Oper Res Manag Sci 10:353–425
Shapiro A, Nemirovski A (2005) On complexity of stochastic programming problems. In: Continuous
optimization. Springer, Berlin, pp 111–146

123
172 A. Gourtani et al.

Shapiro A, Dentcheva D, Ruszczyński AP (2009) Lectures on stochastic programming: modeling and theory,
vol 9. SIAM, Philadelphia
Snyder LV (2006) Facility location under uncertainty: a review. IIE Trans 38(7):547–564
Snyder LV, Daskin MS (2006) Stochastic p-robust location problems. IIE Trans 38(11):971–985
Sun H, Xu H (2015) Convergence analysis for distributionally robust optimization and equilibrium problems.
Math Oper Res 41(2):377–401
Sun H, Xu H, Wang Y (2014) Asymptotic analysis of sample average approximation for stochastic opti-
mization problems with joint chance constraints via conditional value at risk and difference of convex
functions. J Optim Theory Appl 161(1):257–284
Wiesemann W, Kuhn D, Sim M (2014) Distributionally robust convex optimization. Oper Res 62(6):1358–
1376
Wu C, Du D, Xu D (2015) An approximation algorithm for the two-stage distributionally robust facility
location problem. In: Advances in global optimization. Springer, Berlin, pp 99–107
Xia W, Vera JC, Zuluaga LF (2019) Globally solving nonconvex quadratic programs via linear integer
programming techniques. INFORMS J Comput. https://doi.org/10.1287/ijoc.2018.0883
Xu H, Liu Y, Sun H (2018) Distributionally robust optimization with matrix moment constraints: Lagrange
duality and cutting plane methods. Math Program 169(2):489–529
Zhang J, Xu H, Zhang L (2016) Quantitative stability analysis for distributionally robust optimization with
moment constraints. SIAM J Optim 26(3):1855–1882
Zymler S, Kuhn D, Rustem B (2013) Distributionally robust joint chance constraints with second-order
moment information. Math Program 137(1–2):167–198

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps
and institutional affiliations.

123