2012 ?

9 =ff [ 16 o [ 3 G
Sep., 2012 Operations Research Transactions Vol.16 No.3

\k ¡ p ¦ byt ∗

«­ª 1†
§¬¨ 1
Æ­© 2

¢ ~ |) M
,<#{7*Vt :p<#Zb
Æ!9?*X#

nV} |)
,<#a?RnV?!:}N6
Æ,!p
mXp2WT,m:pV_%<p| !IUBÆ|) V<ppm!M
.< apm
Qg apm
DC pm!)*pmX*f|) M
,Vki
Mg_^,
℄ V 7*|) M
, apm)*pm
¥v[i_ O221.5
q{[i_ 90C15, 90C26
A survey on probabilistically constrained
optimization problems

SUN Xiaoling1† BAI Xiaodi1 ZHENG Xiaojin2

Abstract We give a brief review on the probabilistically constrained optimization

problem which is an important class of stochastic programming with wide applications
in finance, management and engineering planning. We introduce the modeling of prob-
abilistic constraints and summarize some important solution methods including convex
approximation, DC approach, scenario approach and integer programming approach. We
also discuss some future research perspectives of the probabilistically constrained opti-
mization problem.
Keywords stochastic programming, probabilistic constraints, approximation meth-
ods, integer programming approach
Chinese Library Classification O221.5

2010 Mathematics Subject Classification 90C15, 90C26

`H 2012  6  18 `
* 5[N9 DG\xg5[D>N9 (Nos. 11101092, 10971034)
1. xMHggk 200433; School of Management, Fudan University, Shanghai 200433, China
2. +=Hge=gk 200092; School of Economics and Management, Tongji University,

† )hP# Corresponding author

Shanghai 200092, China
66 S!QY+SZ 16 o
0  |
L +;"Us5[y'x5
(P) min f (x)
s.t. ci (x, ξ) 6 0, i = 1, · · · , m,
x ∈ X,

$ f : ℜ → ℜ c (·, ξ) : ℜ → ℜ (i = 1, · · · , m) zt0Æ ξ ∈ R z2ÆO

n n s

X z ℜ UE9O f c _z-ÆO X z-9r;" (P) zd℄U- +;"

i
n

yv~agTUlSV . [hL +5[UagtA|~72Æ ξ 33

i
[1,8,28]

n+Z^_d ξ y=J' BU ys43%KJÆU4CU}

5`SWs8ILUA +Z^_Uolz5[ (P) 7&* ξ U+Z^_h~s
Z^_U >J' ξ, dyv~ ξ Unq0O5 ξ s>A >Je
(P) SSs`L V3~ ξ UBW03jW0O53 >Je (P) S
V$nA [ILNSQUL V5[tA|~7y=z+y℄UhO=0
>l5[7~5+y℄Uy=_GR$s8A 2Æ ξ U+Z^_Uolz5[
(P) 7oS 'L c (x, ξ) 6 0 (i = 1, · · · , m) e ξ UV0_-JEh5[UL
VdFU-tA|~rHzy℄UNoS 'Le ξ UV0_-J
i

tA|~733+y℄3JGhJ|UL V8 h.^"tA~

A
/=4<eC Charnes W 1958 >C!>~{( 61 A
L +2ÆU+Z^_ , ILzs^U{(B, 'L-J: (P)
[10]

72Æ c (x, ξ) 6 0 $/5{( P{c(x, ξ) 6 0} > 1 − α, $ c(x, ξ) =

(c (x, ξ), · · · , c (x, ξ)) , 0 < α < 1 zuEDJ|U{( ;"5
i
T
1 m

(PCP) min f (x)

s.t. P{c(x, ξ) 6 0} > 1 − α,
x ∈ X.

B1=0l;E3{( +;" (PCP) U|~

j 1 (,CK#;") p ξ = (ξ , · · · , ξ ) z n uEC4U6 yOIq0z
T

µ, Vo3k'5 Σ. d℄Uq0o3,CK#5[7e,CK# x = (x , · · · , x ) ,
1 n
T

I_ µ x > R ~oS,CK#U F y+R^U,C08&0 R. e

1 n
T

F yUs#  'Lz1~dB{( P(ξ x > R) > 1 − α, :uED T

α ∈ (0, 1) B,,CK# y+R R, $ q65uE0 (VaR)  IuE0

 U,CK#5[y'5
min xT Σx
s.t. P(ξ T x > R) > 1 − α,
x ∈ X,

I7 X zsKZ^_UI_ 4 VaR  UYa +5[z`>Ya +"

Usjh℄\ (K [2-3,5,7,15]).
3 G |) M
,<# 67

j ;") p784C n |i m  }|i i U|z M ,

2(
L84CE|i i Q } j Ur~z c . Fp }U`S ξ =
i
i = 1, · · · , n,
z6Ud=04vs^U{(, }U`S-Jz6
ij
(ξi , · · · , ξm )T
;"yM55
n X
X m
min cij xij
i=1 j=1
( n )
X
s. t. P xij > ξj , j = 1, · · · , m > 1 − α,
i=1
m
X
xij 6 Mi , i = 1, · · · , n,
j=1
xij > 0, i = 1, · · · , n, j = 1, · · · , m,

$ x > 0 zE|i i Q } j U$;"zI_{( ;"Us9

o 9GjuagjhT (K [12,14,25,31]). C2gRd|7UF /
ij

q4 7`SU+Z^_yv~{( M UI_6 +5[

(K [23]).
{( ;" (PCP) U;o
:y℄ Uq-_ ξ vE*s-WlÆ
2 (K [16-17,20,29]), {( P{c(x, ξ) 6 0} > 1 − α ^xUy℄ szq-
U
j 3 u'B1U 2 6{( ;"
min f (x) := 2x21 + 3x22 + 2x1 x2
s. t. P{ξ T x > 0.4} > 0.6,
x ∈ X = {x ∈ ℜ2 | x1 + x2 6 1, x21 + 2x22 6 1, 0 6 x1 , x2 6 1},

$ ξ ∈ ℜ z g6#Is-' 1 Z^/ 1 >;"Uy℄

2
S, yvt
> S zq-U
T 1 h7$ ξ Vt.
Qg 1 2 3 4 5 6
ξ1i 0 0.5 0.6 0.8 1 2
ξ2i 1 2 0.7 0.5 0 0.4
pi 1/6 1/6 1/6 1/6 1/6 1/6

9OIoXn{( L +;"Ujh^; -;`ol Pf

`ol DC ol (Æ)olW=0.e{( L +UsU7VpU9o;
"^℄+46 + (z{( +;"={Jg;e"U9~)r
JÆjh
68 S!QY+SZ 16 o

0.8

0.7

0.6

0.5

0.4

x2 0.3 S

0.2

0.1

−0.1
0 0.2 0.4 0.6 0.8 1
x1

v 1  3 Vz^
1 uSeYW
-`olU;oLz;" (PCP) Uq-y℄ U-;`5ILJK
u' m = 1 UP\: c(·, ξ) : ℜ → ℜ zt0ÆwK P{c(x, ξ) 6 0} > 1 − α WG
n

 P{c(x, ξ) > 0} 6 α. B1u'd"-Æ φ(x), vS P{c(x, ξ) > 0} 6 φ(x), Eh

φ(x) 6 α z{(Æ U-`
p ψ : ℜ → ℜ zqyqJ-ÆO-JdB_6
ψ(z) > ψ(0) = 1, ∀z > 0.

e^" t > 0 6# Z 

E[ψ(tZ)] > E[1[0,+∞) (tZ)] = P(tZ > 0) = P(Z > 0).

% Z = c(x, ξ), ~ t $/ t, jwSQ

−1

P{c(x, ξ) > 0} 6 E[ψ(t−1 c(x, ξ))], ∀t > 0.

% Ψ(x, t) := tE[ψ(t −1
c(x, ξ))]. wt> Ψ(x, t) 6 tα y0> P{c(x, ξ) > 0} 6 α. ^s
/
inf [Ψ(x, t) − tα] 6 0 (1.1)
t>0

y0> P{c(x, ξ) > 0} 6 α. yvm,jwN"UÆz-ÆEh (1.1) z{(

U-;`J|Z;" (PCP) U-;`;"5
min f (x)
s.t. inf [Ψ(x, t) − tα] 6 0,
t>0
x ∈ X.
3 G |) M
,<# 69

Shapiro  [26] 7+-;` (1.1) U  CVaR `

Nemirovski
`B1=0℄+ CVaR `p ψ(z) := [1 + z] , $ [a] = max{0, a}.
Bernstein + +

D[zÆ-J ψ(z) `_6Cr-` (1.1) dB\w

inf E[t + c(x, ξ)]+ − tα 6 0.
 
(1.2)
t>0

^x'LuE0dB (K [30]):
 
1 +
CVaR1−α (z) := inf t + E[z − t] .
t∈ℜ α

wt> (1.2) WG CVaR (c(x, ξ)) 6 0, (1.2) 65 CVaR `yv,3 CVaR
`z{( ULB-;` (K [11]).
1−α

 ξ vE gs-UP B: ξ vW{(V0 ξ , i = 1, · · · , N r;" (PCP) U i

CVaR `;"yvWGZW5dBU- +;"

(CVaR) min f (x)

N
1 X
s.t. t + yi 6 0,
αN i=1
yi > 0, yi > cj (x, ξ i ) − t, i = 1, · · · , N, j = 1, · · · , m,
x ∈ X, t ∈ R.

2 SeYW
DC
DC `olUL (K [18]) z4!Q!-;`\U DC Æ (-Æ.
3) `{(ÆzolC1xÆ 1 (z) U DC `% (0,+∞)

ψ(z, t) = t−1 [z + t]+ φ(z, t) := t−1 [z]+ , t > 0.

 DC Æ π(z, t) := ψ(z, t) − φ(z, t) z1xÆ 1 (0,+∞) (z) UsU`K/ 2–

/ 5.

6 6 ψ(z, t)

1(0,+∞) (z) 1(0,+∞) (z)

1 b 1

0 z- −t 0 z-

v 2 2y 1 (0,+∞) (z) v 3 CVaR a ψ(z, t)

70 S!QY+SZ 16 o

6 ψ(z, t) φ(z, t) 6

1(0,+∞) (z) π(z, t)

1 1

−t 0 z- −t 0 z-

v 4  DC  φ(z, t) v 5 DC  π(z, t)
% g (x, t) = E[t + c(x, ξ)]
1
+
, g (x) = g (x, 0) = E[c(x, ξ)] . p
2 1
+

p̃(x) = inf p̃(x, t) = inf {t−1 [g1 (x, t) − g2 (x)]}.

t>0 t>0

s^UFpByv,3;" (PCP) WGB q-;"

min f (x)
s. t. p̃(x) 6 α, (2.1)
x ∈ X.

V^7;sRU t = ε, j q-;"y~B DC );"^℄`

min f (x)
s. t. g1 (x, ε) − g2 (x) 6 εα,
x ∈ X.

9 [18] !>~b -+ol Monte Carlo olSVj ;"

3 ¤r^aYW
p6# ξ vEH gs-: P(ξ = ξ ) = p , i = 1, · · · , N , P p = 1. p i
i
N
i

u z c (x, ξ )  X jUsjWO j = 1, · · · , N . {( ;" (PCP) yvWG

i=1
j

Z'5B 2# 0-1 (Æ);" (K [31]):

j i

(MIP) min f (x)

s.t. c(x, ξ j ) 6 zj uj , j = 1, · · · , N,
N
X
pj zj 6 α,
j=1

x ∈ X, z ∈ {0, 1}N .

O f , c (·, ξ ) _z-ÆO X z-9rj ;"z-2#(Æ);"Ehy

j

v~4 :Us-^WlSVdO f zI_3iDÆ c zI_

i

i
3 G |) M
,<# 71

Æ X zI_Æ^xU-9r=0yv~FUir +fLd CPLEX WSV

(MIP). N(Æ)5[ (MIP) U :Æ;"UHA33$GEh~4
:Us-^WlUT()+~
p f (x) = x Qx + c x, c(x, ξ) = R − ξ Bx, $ R ∈ ℜ, ξ ∈ ℜ , B ∈ ℜ . Fp ξ
T T T m m×n

vEH gs-OW{(V0 ξ , i = 1, · · · , N . p l u s(z (ξ ) Bx  X jU

i b b i T

jBW i = 1, · · · , N . ~
_sVol ^) (SDP) ol=0yvd

i i

B! (MIP) \U2#(Æ)5[ (K [33]):

N
X N
X
(MIQP) min xT (Q − θi B T ξ i (ξ i )T B)x + cT x + θi (wi2 + φi − R2 yi )
i=1 i=1
s.t. (ξ i )T Bx = wi + zi − yi R, i = 1, · · · , N,
eT y 6 K, y ∈ {0, 1}N ,
lib yi 6 zi 6 Ryi , φi yi > zi2 , φi > 0, i = 1, · · · , N,
x ∈ X, R 6 wi 6 ubi , i = 1, · · · , N.

$ K = ⌊N α⌋, θ ∈ ℜ z2ÆO-J Q − P θ B ξ (ξ ) B  0. +*Uolyv~

N
+
N
i
T i i T

Z^2ÆO θ. 9 [33] !>s8(SVs^);"SQ “LB” 2Æ θ

i=1

Uol
O (PCP) 7UÆ 9# X _zI_O{( ÆU\w5 P{T x > ξ} > 1 − α
r~y℄ U T J|UT+Ww9 [25] !> (MIP) U8ENU2
#(Æ)5[9 [19] >C ;"U$2s T+Ww
4 nsYW
L1;SXnUSV{( U4ol29G7.+loly~A
s3 \wU{( ;"
• Pf3nol ([9,27]);
• nDqol ([24]);
• p T\ol ([12-14,21-22])
• 69w};" ([6,32]).
B1=0℄IoXnL8ol
4.1ofZX
Pfol/Q6OUHgn3Pf ξ i
(i = 1, · · · , N ) Je EhS
QdB` +;"
(SP) min f (x)
s.t. c(x, ξ j ) 6 0, j = 1, · · · , N,
x ∈ X.
72 S!QY+SZ 16 o
PfolU \;o (1) e ξ Us-.^"H5 (2) e f , c X
-_FpB (SP) qz- +;"EhOnÆ N +$Gr;" (SP) yv w
i

SV[hPfolUY\q$3D +*UndV1P4+*U 'L

(SP) ;"qz6UEhIL Vqz6Us9oU;"zd"s^U{(
B, (SP) UVqzÆ;" (PCP) Uy℄V9 [9] ,3nÆ N `o-J
2 1 2n 2
N> log + 2n + log ,
α δ α α

0=,v{( 1 − δ ∈ (0, 1) v (SP) UL Vz (PCP) Uy℄VÆPfolU

g+yK9 [27].
4.2}RmgZX
5{v>{(Æ P{c(x, ξ) 6 0} U
:=0yvu'~nDqUol
`{( ;"p ξ , · · · , ξ z6O ξ U N `ne^U ε ∈ (0, 1),
1 N

u'dBnDq`;"
(SAP) min f (x)
N
1 X
s.t. 1(−∞,0] (c(x, ξ j )) > 1 − ε,
N j=1
x ∈ X.

D[O ε = 0, ;" (SAP) jzPf`;" (SP). 9 [24] ,3O ε > α r (SAP) v

1Æb Q 1 U{(4qsÆ;"UL 0UBWO ε < α rs^U'L
BnDq;" (SAP) UL Vv$GU{(zÆ;"Uy℄V
5 dl` £ w
B>< L +M5 lUjd℄L +;"Ul fL_T7
s9oU;"zd"ZetA;"^℄M5)&e>FUX;"!>TUo
l2ÆU+Z^_z8  " U +M57E%>FUFPd"A +
;"72ÆU+Z^_75`><  veUs℄\ n& _U;"{
( O5L +M5Us8ol|~" Q9~$yvE`><
+9oAFsjj'UÆ+9Æt>sU=4U +Cd Nemirovski,
Nemhasuer, Shapio W{( ;"jhj_9oO
yv~{( A +Z^_2.yv~2Æ9#|(2ÆU+Z^_
35'_$8ol65& (robust) +olI;oU +nz^) iR
A)ZXUa"yv2uÆ9G[h~2Æ9#|(+Z^_q$3D
UY\C$:Z^#|U2Æ9#s3=v~UH31R'x2Æ9#
$zz5J|U& ;"yvWGZ+5iRA)3^)$2$g|
~" & +5[4qUL V338b4.gGU|~G0O[{
( q%*nU;" d"# ZZ^uED α?
3 G |) M
,<# 73

{( ;"Uol.77.ag9oUn& _U;"KVp

dOe6#Us-.^"CmUY?rn`3~uÆmO5 gs
-Un\zs8$F[UdJ|U(Æ)5[33$g 0-1 # 2#(Æ
 d"SV$ G5;"U`VÆ9 [4] !>s4P$oO ksVU`
SV{( ;"Uols^s/UjhoOzd"T#T+Ww D1p
>TU` Zl
U h x z
[1] iK:Q M
,Æ,pm [M].b wf?o 1997.
[2] Alexander G J, Baptista A M. A comparison of VaR and CVaR constraints on portfolio selec-
tion with the mean-variance model [J]. Management Science, 2004, 50: 1261-1273.
[3] Alexander G J, Baptista A M, Yan S. Mean-variance portfolio selection with ‘at-risk’ con-
straints and discrete distributions [J]. Journal of Banking & Finance, 2007, 31: 3761-3781.
[4] Bai X D, Sun J, Zheng X J, Sun X L. A penalty decomposition method for probabilistically
constrained convex programs [R]. Technical Report, School of Management, Fudan University,
2012. [2012-04-03] http://www.optimization-online.org/DB− FILE/2012/04/3448.pdf.
[5] Benati S, Rizzi R. A mixed integer linear programming formulation of the optimal mean/Value-
at-Risk portfolio problem [J]. European Journal of Operational Research, 2007, 176: 423-434.
[6] Beraldi P, Ruszczyński A. The probabilistic set-covering problem [J]. Operations Research,
2002, 50: 956-967.
[7] Bonami P, Lejeune M A. An exact solution approach for portfolio optimization problems under
stochastic and integer constraints [J]. Operations Research, 2009, 57: 650-670.
[8] Boyd S P, Vandenberghe L. Convex Optimization [M]. Cambridge: Cambridge University
Press, 2004.
[9] Calafiore G C, Campi M C. The scenario approach to robust control design [J]. IEEE Trans-
actions on Automatic Control, 2006, 51: 742-753.
[10] Charnes A, Cooper W W, Symonds G H. Cost horizons and certainty equivalents: an approach
to stochastic programming of heating oil [J]. Management Science, 1958, 4: 235-263.
[11] Chen W, Sim M, Sun J, Teo C P. From CVaR to uncertainty set: Implications in joint chance-
constrained optimization [J]. Operations Research, 2010, 58: 470-485.
[12] Cheon M S, Ahmed S, Al-Khayyal F. A branch-reduce-cut algorithm for the global optimization
of probabilistically constrained linear programs [J]. Mathematical Programming, 2006, 108:
617-634.
[13] Dentcheva D, Martinez G. Augmented Lagrangian method for probabilistic optimization [J].
Annals of Operations Research, 2011. DOI: 10.1007/s10479-011-0884-5.
[14] Dentcheva D, Prékopa A, Ruszczynski A. Concavity and efficient points of discrete distribu-
tions in probabilistic programming [J]. Mathematical Programming, 2000, 89: 55-77.
[15] Gaivoronski A A, Pflug G. Value at risk in portfolio optimization: Properties and computa-
tional approach [J]. Journal of Risk, 2005, 7: 1-31.
[16] Henrion R. Structural properties of linear probabilistic constraints [J]. Optimization, 2007, 56:
425-440.
[17] Henrion R, Strugarek C. Convexity of chance constraints with independent random variables
[J]. Computational Optimization and Applications, 2008, 41: 263-276.
[18] Hong L J, Yang Y, Zhang L. Sequential convex approximations to joint chance constrained
programs: A monte carlo approach [J]. Operations Research, 2011, 59: 617-630.
74 S!QY+SZ 16 o
[19] Küçükyavuz S. On mixing sets arising in chance-constrained programming [J]. Mathematical
Programming, 2012, 132: 31-56.
[20] Lagoa C M, Li X, Sznaier M. Probabilistically constrained linear programs and risk-adjusted
controller design [J]. SIAM Journal on Optimization, 2005, 15: 938-951.
[21] Lejeune M, Noyan N. Mathematical programming approaches for generating p-efficient points
[J]. European Journal of Operational Research, 2010, 207: 590-600.
[22] Lejeune M, Noyan N. Mathematical programming approaches for generating p-efficient points
[J]. European Journal of Operational Research, 2010, 207: 590-600.
[23] Lejeune M A, Ruszczynski A. An efficient trajectory method for probabilistic inventory production-
distribution problems [J]. Operations Research, 2007, 55: 378-394.
[24] Luedtke J, Ahmed S. A sample approximation approach for optimization with probabilistic
constraints [J]. SIAM Journal on Optimization, 2008, 19: 674-699.
[25] Luedtke J, Ahmed S, Nemhauser G L. An integer programming approach for linear programs
with probabilistic constraints [J]. Mathematical Programming, 2010, 122: 247-272.
[26] Nemirovski A, Shapiro A. Convex approximations of chance constrained programs [J]. SIAM
Journal on Optimization, 2006, 17: 969-996.
[27] Nemirovski A, Shapiro A. Scenario approximations of chance constraints [M] // Calafiore G,
Dabbene F, editors. Probabilistic and Randomized Methods for Design Under Uncertainty,
Berlin: Springer, 2006, 3-47.
[28] Nocedal J, Wright S J. Numerical Optimization [M]. Berlin: Springer, 1999.
[29] Prékopa A. Probabilistic programming [M] // Ruszczyński A, Shapiro A, editors. Stochastic
Programming, Handbooks in Operations Research and Management Science. Amsterdam:
Elsevier Science, 2003: 267-351.
[30] Rockafellar R T, Uryasev S. Optimization of conditional value-at-risk [J]. Journal of Risk,
2000, 2: 21-42.
[31] Ruszczyński A. Probabilistic programming with discrete distributions and precedence con-
strained knapsack polyhedra [J]. Mathematical Programming, 2002, 93: 195-215.
[32] Saxena A, Goyal V, Lejeune M A. MIP reformulations of the probabilistic set covering problem
[J]. Mathematical Programming, 2010, 121: 1-31.
[33] Zheng X J, Sun X L, Li D, Cui X T. Lagrangian decomposition and mixed-integer quadratic
programming reformulations for probabilistically constrained quadratic programs [J]. European
Journal of Operational Research, 2012, 221: 38-48.