ON EXISTENCE OF SOLUTIONS TO DEGENERATE NONLINEAR OPTIMIZATION PROBLEMS
Agnieszka Prusi´ nska
1and Alexey Tret’yakov
1,21
Institute of Mathematics and Physics
University of Podlasie, 3–go Maja 54, 08–110 Siedlce, Poland
2
System Research Institute
Polish Academy of Sciences, Newelska 6, 01–447 Warsaw, Poland
Abstract
We investigate the existence of the solution to the following problem min ϕ(x) subject to G(x) = 0,
where ϕ : X → R, G : X → Y and X, Y are Banach spaces. The question of existence is considered in a neighborhood of such point x
0that the Hessian of the Lagrange function is degenerate. There was obtained an approximation for the distance of solution x
∗to the initial point x
0.
Keywords: Lagrange function, necessary condition of optimality, p- regularity, contracting mapping, p-factor operator.
2000 Mathematics Subject Classification: Primary: 90C30, Secondary: 49M05, 47H10, 47A50, 47J05.
1. Introduction
As it is well known, one of the most important questions of applied mathe- matics, for example from the numerical point of view (but not only), is the problem of existence of solutions.
We investigate the conditions of existence of solutions for the following nonlinear optimization problem
(1) min ϕ(x) subject to G(x) = 0
in some neighborhood of an initial point x
0, where ϕ : X → R and G : X → Y are sufficiently smooth and X, Y are Banach spaces.
Based on the classical approach, the answer to this question is given by one of the modifications of the implicit function theorem. Nevertheless, there are many cases of nonlinear type problems, where this theorem cannot be applied, at least not directly.
The purpose of our paper is to investigate the above mentioned type of problems, where the Hessian of the Lagrange function is degenerate at the initial point.
Let us consider the Lagrange function of the following form L(x, λ) = ϕ(x) + hG(x), λi, λ ∈ Y
?and define the mapping F F (x, λ) =
ϕ
0(x) + G
0(x)
?· λ G(x)
,
where F : X × Y
?→ X
?× Y. As it is known, in accordance with necessary conditions, in an optimal point x
∗there exists λ
∗such that the following equality
(2) F (x
∗, λ
∗) = 0
holds (obviously under some regularity condition for G(x) at the point x
∗). If operator F
0(x
0, λ
0) is non-degenerate in some initial point z
0= (x
0, λ
0), then we can guarantee the existence of solutions for (2) in some neighborhood of z
0(see for instance [2]). If F
0(x
0, λ
0) is degenerate, then the question about the existence of solutions to (2) is open.
In this paper, we investigate a degenerate situation and obtain condi- tions for which there exists a solution to the problem (2) in some neighbor- hood of the initial point z
0.
Our approach is based on the notion of p-regularity introduced in 1982 by the second author [9]. Among many applications the structure of p-factor operator can be used in the construction of numerical methods for solving degenerate nonlinear operator equations and optimization problems.
We would like to present how to apply the p-regularity theory, also
known as factor-analysis of nonlinear mappings to develop methods for find-
ing solutions to related singular problems, in particular, we would like to
show how these ideas can be applied in some specific situations, namely in optimization problems.
2. Elements of p-regularity theory We start with some important notions of the p-regularity theory.
The problem (2) is called regular at the initial point z
0= (x
0, λ
0) if F
(x,λ)0(x
0, λ
0) is surjection. Otherwise, the problem (2) is called nonregular, (irregular, singular, degenerate) at the point z
0.
Let p be a natural number and let Z, W be Banach spaces. We consider a mapping F : Z → W which is continuously p-times Fr´echet differentiable on Z. Furthermore, we shall use the following notation
Ker
kF
(k)(z) = n
h ∈ Z : F
(k)(z)[h]
k= 0 o
for the k-kernel of the mapping F
(k)(z) (the zero locus of F
(k)(z)), where k = 1, . . . , p.
The set
M (z
0) = {z ∈ U : F (z) = F (z
0)}
is called the solution set for the mapping F in neighborhood U .
We call h a tangent vector to a set M ⊆ Z at z
0∈ M if there exist ε > 0 and a mapping r : [0, ε] → Z with the property that for t ∈ [0, ε] we have z
0+ th + r(t) ∈ M and kr(t)k = o(t).
The collection of all tangent vectors at z
0is called the tangent cone to M at z
0and it is denoted by T
1M (z
0). (For more detail we refer for instance to [3]).
For a linear operator Λ : Z → W, we denote by Λ
−1its right inverse, that is Λ
−1: W → 2
Z(Λ
−1is said to be a multimapping or a multivalued mapping from W into 2
Z) whose any element w ∈ W maps on its complete inverse image under the mapping Λ :
Λ
−1w = {z ∈ Z : Λz = w} . Furthermore, we will use the ”norm”
(3)
Λ
−1= sup
kwk=1
inf {kzk : Λz = w, z ∈ Z} .
Note that when Λ is one-to-one, kΛ
−1k can be considered as the usual norm
of the element Λ
−1in the space L(Z, W ).
Let U be a neighborhood of the point z
0∈ Z. Consider a sufficiently smooth nonlinear mapping F : U → W, such that ImF
0(z
0) 6= W. We construct the p-factor operator under the assumption that the space W is decomposed into the direct sum:
(4) W = W
1⊕ . . . ⊕ W
p,
where W
1= cl(ImF
0(z
0)) (the closure of the image of the first derivative of F evaluated at z
0), and the remaining spaces are defined as follows. Let V
1= W, V
2be closed complementary subspace to W
1(we are assuming that such a closed complement exists), and let P
V2: W → V
2be the projection operator onto V
2along W
1. Let W
2be the closed linear span of the image of the quadratic map P
V2F
(2)(z
0)[·]
2. More generally, define inductively,
W
i= cl(span ImP
ViF
(i)(z
0)[·]
i) ⊆ V
i, i = 2, . . . , p − 1,
where V
iis a choice of a closed complementary subspace for (W
1⊕. . .⊕W
i−1) with respect to W, i = 2, . . . , p and P
Vi: W → V
iis the projection operator onto V
ialong (W
1⊕ . . . ⊕ W
i−1) with respect to W, i = 2, . . . , p. Finally, W
p= V
p. The order p is chosen as the minimum number for which (4) holds.
Now, define the following mappings (see [4]),
f
i: U → W
i, f
i(z) = P
WiF (z), i = 1, . . . , p,
where P
Wi: W → W
iis the projection operator onto W
ialong (W
1⊕ . . . ⊕ W
i−1⊕ W
i+1⊕ . . . ⊕ W
p) with respect to W, i = 1, . . . , p.
Definition 1. The linear operator Ψ
p(h) ∈ L(Z, W
1⊕ . . . ⊕ W
p) is defined for h ∈ Z by
Ψ
p(h)z = f
10(z
0)z + f
200(z
0)[h]z + . . . + f
p(p)(z
0)[h]
p−1z, for z ∈ Z and is called the p-factor operator.
Sometimes it is convenient to use the following equivalent definition of the p-factor operator e Ψ
p(h) ∈ L(Z, W
1× . . . × W
p) for h ∈ Z,
Ψ e
p(h)z = (f
10(z
0)z, f
200(z
0)[h]z, . . . , f
p(p)(z
0)[h]
p−1z), for z ∈ Z.
Note that in the completely degenerate case, i.e., in the case that
F
(r)(z
0) = 0, r = 1, . . . , p−1, the p-factor operator is simply F
(p)(z
0)[h]
p−1.
Roughly speaking, we construct a decomposition of a ”non-regular part”
of the mapping F on partiall mappings f
i, in such a way that all of those mappings are completely degenerate up to the order i − 1, where i = 2, ..., p.
For our consideration we need the following generalization of the notion of the regular mapping.
Definition 2. We say that the mapping F is p-regular at z
0along h if ImΨ
p(h) = W.
Let us introduce the following auxiliary set H
p(z
0) = {z ∈ Z : Ψ
p(z)z = 0}.
Definition 3. The mapping F is p-regular at the point z
0if either it is p-regular along any h from the set H
p(z
0) \ {0} or H
p(z
0) = {0}.
The following theorem gives a description of a solution set in the degenerate case:
Theorem (Generalization of the Lyusternik theorem) [11]. Let Z and W be Banach spaces, and U be a neighborhood of z
0∈ Z. Assume that Φ : Z→W, Φ ∈ C
p(U ) is p-regular at z
0. Then T
1M (z
0) = H
p(z
0).
3. Regular case
For solving many nonlinear problems in the regular case (that is when F
0is non-degenerate at the initial point z
0) classical results can be used, such as Lyusternik Theorem, Implicit Function Theorem, Lagrange-Euler’s op- timality conditions. In this case, the tangent cone to a solution set of the equation (2) is equal to the zero locus of the first derivative of the mapping F, i.e., T
1M (z
0) = KerF
0(z
0).
Besides description of the solution set and formulation of optimality conditions, a very important problem is to give a guarantee of the existence of a solution in some neighborhood of an initial point.
We quote one modification of the theorem about the existence of solu- tions of equation (2) in the regular case (see e.g. [2, 8]).
Consider a mapping F : Z → W and the existence of such point z
∗that
F (z
∗) = 0. Throughout this section we assume that F (z
0) is regular, i.e.,
F
0(z
0)Z = W and that [F
0(z
0)]
−1is a multivalued mapping. Moreover, let
U
ε(z
0) = {z ∈ Z : kz − z
0k < ε} where 0 < ε < 1.
Theorem 1. Let F ∈ C
2(U
ε(z
0)), kF (z
0)k = η, k[F
0(z
0)]
−1k = δ and sup
z∈Uε(z0)
kF
00(z)k = c < ∞. In addition, if the following inequalities 1. δ · c · ε ≤
162. δ · η ≤
ε2hold, then the equation F (z) = 0 has a solution z
∗∈ U
ε(z
0).
4. Degenerate case
There are many interesting optimization problems which are of degenerate form. We shall prove a generalization of Theorem 1 in the degenerate case.
For this purpose we need three auxiliary lemmas. The first of these lemmas is a ”multivalued” generalization of the contraction mapping principle. By σ(A
1, A
2) we denote the deviation of the set A
1from the set A
2and by h(A
1, A
2) we mean the Hausdorff distance between sets A
1and A
2.
Lemma 1 (Contraction multimapping principle) [3]. Let Z be a complete metric space with distance ρ. Assume that we are given a multimapping
Φ : U
(z
0) → 2
Z,
on a ball U
(z
0) = {z : ρ(z, z
0) < } ( > 0) where the sets Φ(z) are non- empty and closed for any z ∈ U
(z
0). Further, assume that there exists a number θ, 0 < θ < 1, such that
1. h(Φ(z
1), Φ(z
2)) ≤ θρ(z
1, z
2) for any z
1, z
2∈ U
(z
0) 2. ρ(z
0, Φ(z
0)) < (1 − θ), where
ρ(z
0, Φ(z
0)) = inf
u∈Φ(z0)
ρ(z
0, u).
Then, for every number
1which satisfies the inequality ρ(z
0, Φ(z
0)) <
1< (1 − θ),
there exists an element z ∈ B = {ω : ρ(ω, z
0) ≤
1/(1 − θ)} such that
(5) z ∈ Φ(z).
Further, by the distance ρ we mean just a norm, that is ρ(z
1, z
2) = kz
1−z
2k.
Lemma 2 [3]. Let Z be a Banach space, and let M
1and M
2be linear manifolds in Z which are translations of a single subspace L. Then
h(M
1, M
2) = σ(M
1, M
2) = σ(M
2, M
1) = inf {kz
1− z
2k : z
1∈ M
1, z
2∈ M
2} . Lemma 3 [3]. Let Z and W be Banach spaces, and let Λ ∈ L(Z, W ). We set
C(Λ) = sup
kwk=1
inf {kzk : z ∈ Z, Λz = w} . If ImΛ = W, then C(Λ) < ∞.
Let us mention one of the consequences of the Mean Value Theorem, which is important for our further investigations. For the proof we refer the reader to [3, 7].
Lemma 4. Let f : U → W, where [a, b] ⊂ U ⊆ Z. Then kf (b) − f (a) − Λ(b − a)k ≤ sup
ξ∈[a,b]
kf
0(ξ) − Λk · ka − bk,
for any Λ ∈ L(Z, W ).
Lemma 5. Let Λ : Z → W be a nonlinear operator of the form Λ[z]
p= Λ
1[z], Λ
2[z]
2, . . . , Λ
p[z]
p= (w
1, . . . , w
p) = w, where kwk 6= 0 and Λ
r[z]
ris r-form for r = 1, 2, . . . , p. If
(6) sup
Λ
−1w kwk
= inf
kzk : Λ[z]
p= w kwk
= c < ∞, then
(7)
Λ
−1w ≤ cp
kw
1k + kw
2k
12+ . . . + kw
pk
1 p
.
P roof. For any w = (w
1, . . . , w
p) ∈ W there exists α > 0 such that
(8) kw
1k
α + . . . + kw
pk α
p= 1.
Without loss of generality, we can assume that kw
ik 6= 0, i = 1, . . . , p. Let us
denote
wα1= e w
1, . . . ,
wαpp= e w
p. Then from (8) we have k e w
1k + . . . + k e w
pk = 1.
Taking into account (6) we obtain k
αzk ≤ c. Hence kzk ≤ cα. Suppose that α > p
kw
1k + kw
2k
12+ . . . + kw
pk
1 p
. Then
kw1k
α
+ . . . +
kwαppk<
kw1kp
kw1k+...+kwpk
1 p
+ . . . +
kwpkpp
kw1k+...+kwpk
1 p
p
≤
pkwkw1k1k
+ . . . +
pkwkwpkpk
=
1p+ . . . +
1p= 1.
This contradicts our assumption (8). Hence α ≤ p
kw
1k + . . . + kw
pk
1 p
and finally (7).
Let us introduce the following additional notations and assumptions U
ε(z
0) = {z ∈ Z : kz − z
0k < ε} where 0 < ε < 1,
(9)
δ = kF (z
0)k 6= 0, (10)
Λ(h) ∈ L(Z, W
1× . . . × W
p) where Λ(h) =
f
10(z
0), f
200(z
0)[h], . . . , 1
(p − 1)! f
p(p)(z
0)[h]
p−1, (11)
Λ[h]
p=
f
10(z
0)[h], f
200(z
0)[h]
2, . . . , 1
(p − 1)! f
p(p)(z
0)[h]
p, (12)
Λ
−1w = {h ∈ Z : w = Λ[h]
p} , (13)
h = ˆ h
khk , where h ∈ Λ
−1[−F (z
0)], h 6= 0, (14)
Λ
i(h) = 1
(i − 1)! f
i(i)(z
0)[h]
i−1, i = 1, . . . , p, (15)
n
Λ
ˆ h o
−1≤ c
1, (16)
sup
z∈Uε(z0)
f
i(i+1)(z)
≤ c
2< ∞, i = 1, . . . , p, (17)
Λ
−1= sup
kwk=1
inf {kzk : w = Λ[z]
p, z ∈ Z} = c
3.
(18)
Remark. From the condition of p-regularity follows c
1< +∞ (see [9]). Let the assumptions (9)–(18) be satisfied. Then the following generalization of Theorem 1 holds
Theorem 2. Let F : Z → W, F ∈ C
p+1(U
ε(z
0)) be a p-regular mapping at z
0along some ˆ h . Moreover, assume the following inequalities to be satisfied
1. c
3p
2δ
1 p
≤
13ε, 2. δ < 1,
3.
43(4
p− 1)c
1c
2ε ≤
12.
Then the equation F (z) = 0 has a solution z
∗∈ U
ε(z
0).
P roof. Consider a multivalued mapping Φ
h: U
ε(z
0) → 2
Z, such that Φ
h(z) = z − {Λ(h)}
−1(f
1(z
0+ h + z), . . . , f
p(z
0+ h + z)), z ∈ U
ε(z
0).
Similarly as in the regular case, the assumptions of the contraction mul- timapping principle hold for Φ
h. Indeed, the sets Φ
h(z) are non-empty be- cause Λ(h) is a surjection for any z ∈ U
ε(z
0).
Moreover, for any w ∈ W
1× . . . × W
pthe sets {Λ(h)}
−1w are linear manifolds parallel to KerΛ(h), and hence the sets Φ
h(z) are closed for any z ∈ U
ε(z
0). We will now prove that
(19) h (Φ
h(u
1), Φ
h(u
2)) ≤ 1
2 ku
1− u
2k, for u
1, u
2∈ U
ε2
(z
0) such that ku
jk ≤
khkR, j = 1, 2, where R = max
R
i: R
i= max
1, 2
ε · c
2· 1 (i − 1)!
f
i(i)(z
0)
, i = 1, . . . , p
.
Let s
1= z
0+ h + u
1, s
2= z
0+ h + u
2. Then
h (Φ
h(u
1), Φ
h(u
2)) = inf {kz
1− z
2k : z
i∈ Φ
h(u
j), j = 1, 2}
= inf {kz
1− z
2k : Λ(h)z
j= Λ(h)u
j− (f
1(s
j), . . . , f
p(s
j)) , j = 1, 2}
≤ inf {kzk : Λ(h)z = Λ(h)(u
1− u
2)− (f
1(s
1) − f
1(s
2), . . . , f
p(s
1) −f
p(s
2))}
= inf
kzk : Λ
h khk
z = (Λ
1(h)(u
1− u
2) − f
1(s
1) + f
1(s
2),
. . . , 1
khk
p−1(Λ
p(h)(u
1− u
2) − f
p(s
1) + f
p(s
2))
≤ inf
kzk : z = n
Λ(ˆ h) o
−1(Λ
1(h)(u
1− u
2) − f
1(s
1) + f
1(s
2),
. . . , 1
khk
p−1(Λ
p(h)(u
1− u
2) − f
p(s
1) + f
p(s
2)))
≤ c
1· X
pi=1
1
khk
i−1kf
i(s
1) − f
i(s
2) − Λ
i(h)(s
1− s
2)k . Taking into account Lemma 4, we have
(20)
kf
i(s
1) − f
i(s
2) − Λ
i(h)(s
1− s
2)k
≤ sup
θ∈[0,1]
f
i0(s
2+ θ(s
1− s
2)) − Λ
i(h)
· ku
1− u
2k .
By a complete degeneration of f
iup to the order i we obtain the following Taylor expansion
f
i0(s
2+ θ(s
1− s
2)) = f
i0(z
0) + . . . + f
i(i)(z
0)
(i − 1)! [s
2− z
0+ θ(s
1− s
2)]
i−1+ ω(h, u
1, u
2, θ)
= f
i(i)(z
0)
(i − 1)! [s
2− z
0+ θ(s
1− s
2)]
i−1+ ω(h, u
1, u
2, θ), (21)
where
kω(h, u
1, u
2, θ)k ≤ sup
z∈Uε(z0)
f
i(i+1)(z)[h + u
2+ θ(s
1− s
2)]
i.
On account of R and ku
jk, j = 1, 2, we have kh + u
2+ θ(s
1− s
2)k ≤ 4khk.
Let F (z
0) = (w
1, . . . , w
p), where w
i∈ W
i, i = 1, . . . , p. Then from the
assumption and the definition of norm in Z we have kw
1k + . . . + kw
pk ≤ δ.
Taking into account Lemma 5 and assumptions 3 and 2 we have khk ≤ (1 + ∆)kΛ
−1(−F (z
0))k ≤ (1 + ∆)c
3p
kw
1k+kw
2k
12+ . . . + kw
pk
1 p
≤ (1 + ∆)c
3p
2δ
1p≤
ε2,
where 0 < ∆ <
12. Hence, from previous formulas we get (22) kω(h, u
1, u
2, θ)k ≤ c
2kh + u
2+ θ(s
1− s
2)k
i≤ 4
ic
2ε
2 khk
i−1. Moreover,
(23)
f
i(i)(z
0)[h + u
2+ θ(s
1− s
2)]
i−1=
i−1
X
k=0 i−1
k
f
i(i)(z
0)[h]
i−1−k[u
2+ θ(s
1− s
2)]
k= f
i(i)(z
0)[h]
i−1+
i−1
X
k=1 i−1
k
f
i(i)(z
0)[h]
i−1−k[u
2+ θ(s
1− s
2)]
k,
where
(24) ku
2+ θ(s
1− s
2)k ≤ 3khk/R ≤ 3khk/R
i. Taking into account the choice of R
i,
(25)
i−1
X
k=1 i−1
k
f
i(i)(z
0)[h]
i−1−k[u
2+ θ(s
1− s
2)]
k≤
f
i(i)(z
0) ·
i−1
X
k=1 i−1
k
khk
i−1−k(3khk)
k/R
ki≤
f
i(i)(z
0)
· khk
i−1· 4
i−1/R
i≤ 4
i(i − 1)! ε
2 c
2khk
i−1. Now, inserting (21)–(25) into (20) we obtain
kf
i(s
1) − f
i(s
2) − Λ
i(h)(s
1− s
2)k ≤ 4
iεc
2khk
i−1· ku
1− u
2k.
Hence
h(Φ
h(u
1), Φ
h(u
2)) ≤ c
1· X
p i=11
khk
i−14
ic
2εkhk
i−1ku
1− u
2k
= 4
3 (4
p− 1) c
1c
2εku
1− u
2k ≤ 1
2 ku
1− u
2k so this proves (19).
Let us take z
1, and an arbitrary element of Φ
h(z
0), such that z
1∈ z
0− Λ
−1(F (z
0)) and kz
1− z
0k ≤ (1 + ∆)kΛ
−1(−F (z
0)k, where 0 < ∆ <
12. Thus we have kΦ
h(z
0) − z
0k ≤ kz
1− z
0k ≤ (1 + ∆)c
3p
2δ
1
p
≤
12ε. From the above and from (19) we obtain that for the mapping Φ
hall the assumptions of the contraction multimapping principle hold and hence there exists an element z
∗such that z
∗∈ Φ
h(z
∗). It means that
0 ∈ [Λ(h)]
−1(f
1(z
0+ h + z
∗), . . . , f
p(z
0+ h + z
∗))
and hence (f
1(z
0+ h + z
∗), . . . , f
p(z
0+ h + z
∗)) = 0. Then for i = 1, . . . , p, f
i(z
0+ h + z
∗) = 0 which is equivalent to F (z
0+ h + z
∗) = 0. It follows that z
0+ h + z
∗is the solution of (1).
We conclude the discussion by the following simple examples, which serve to illustrate how to apply Theorem 2 to degenerate optimization problems.
Example 1. Consider the optimization problem
(26) min ϕ(x
1, x
2) subject to G(x
1, x
2) = 0, where ϕ : R
2→ R, G : R
2→ R and ϕ(x
1, x
2) = (x
1−
13·103
)
2+ x
42and G(x
1, x
2) = x
1x
2with the solution x
∗= (
12·103
, 0). Consider also the La- grange function of the form
L(x
1, x
2, λ) = (x
1−
13·103
)
2+ x
42+ λx
1x
2and the gradient of this function
F (x
1, x
2, λ) = ∇L(x
1, x
2, λ) = 2(x
1−
3·1013) + λx
2, 4x
32+ λx
1, x
1x
2T.
For the following initial point (x
01, x
02, λ
0) = (0, 0, 0), we obtain
F (0, 0, 0) =
−
3·102 30 0
, F
0(0, 0, 0) =
2 0 0 0 0 0 0 0 0
and hence F is singular at (0, 0, 0). Moreover the mapping F is 2-regular at the given initial point along ˆ h = (1, 0, 0)
Tbecause the 2-factor operator Λ(ˆ h) =
2 0 0
0 0
12·103
0
2·10130
is a surjection. Then for h = (
13·103
, 0, 0)
Twe obtain Λ[h]
2=
3·1023, 0, 0
Tand we calculate δ =
3·1023, c
1= 1, c
2= 24, c
3=
12·10√152. Applying Theorem 2 we obtain that in U
ε(0, 0, 0) where ε =
1
103
, there exists a solution of the equation F (x
1, x
2, λ) = 0 and, as it is easy to see, there exists a solution to the optimization problem (26) under consideration, namely x
∗= (
2·1013, 0) ∈ U
ε(0, 0).
Example 2. Similarly, let us consider
(27) min(x
21+ x
22) subject to x
1x
2= 1 7200 , where the solution is x
∗= (
160√ 2
,
160√
2
). The Lagrange function for this problem is as follows
L(x
1, x
2, λ) = x
21+ x
22+ λx
1x
2− λ
72001, and its gradient has the form
F (x
1, x
2, λ) = ∇L(x
1, x
2, λ) =
2x
1+ λx
2, 2x
2+ λx
1, x
1x
2− 1 7200
T.
For the initial point (x
01, x
02, λ
0) = (0, 0, −2) we get
F (0, 0, −2) =
0 0
−
72001
, F
0(0, 0, −2) =
2 −2 0
−2 2 0
0 0 0
and hence F is a singular mapping at (0, 0, −2). As it is easy to see F is 2-regular at this point along ˆ h = (
√12
,
√12
, 0)
Tand the 2-factor operator
Λ(ˆ h) =
2 −2
√22−2 2
√22√2 2
√2
2
0
is nonsingular. For h = (
1201,
1201, 0)
Twe get
Λ[h]
2= 0, 0,
72001 T. On account of (9)–(18) we obtain c
1=
√12
, c
2= 1, c
3=
14. Applying Theorem 2 we conclude that in U
ε(0, 0, −2), where ε =
√402there exists a solution of equation (3), and hence there exists also a solution to the optimization problem (27), namely x
∗= (
160√ 2
,
160√ 2