검색
Article Search

JMB Journal of Microbiolog and Biotechnology

OPEN ACCESS eISSN 0454-8124
pISSN 1225-6951
QR Code

Articles

Kyungpook Mathematical Journal 2018; 58(3): 559-571

Published online September 30, 2018

Copyright © Kyungpook Mathematical Journal.

Stability and Constant Boundary-Value Problems of f-Harmonic Maps with Potential

Bouazza Kacimi, and Ahmed Mohammed Cherif*

Department of Mathematics, University Mustapha Stambouli, Mascara, 29000, Algeria
e-mail : kacimibouazza@yahoo.fr and med_cherif_ahmed@yahoo.fr

Received: October 3, 2017; Accepted: July 18, 2018

In this paper, we give some results on the stability of f-harmonic maps with potential from or into spheres and any Riemannian manifold. We study the constant boundary-value problems of such maps defined on a specific Cartan-Hadamard manifolds, and obtain a Liouville-type theorem. It can also be applied to the static Landau-Lifshitz equations. We also prove a Liouville theorem for f-harmonic maps with finite f-energy or slowly divergent f-energy.

Keywords: f-harmonic maps with potential, stability, boundary-value.

We give some definitions.

(1) Let (M,g) be a Riemannian manifold. The divergence of (0, p)-tensor α on M is defined by

(divMα)(X1,,Xp-1)=(eiMα)(ei,X1,,Xp-1),

where ∇M is the Levi-Civita connection with respect to g, X1, …, Xp−1 ∈ Γ(TM), and {ei} is an orthonormal frame. Given a smooth function λ on M, the gradient of λ is defined by

g(gradMλ,X)=X(λ),

the Hessian of λ is defined by

(HessMλ)(X,Y)=g(XMgrad λ,Y),

where X, Y ∈ Γ(TM), the Laplacian of γ is defined by

ΔM(λ)=tracegHessMλ,

(see [11]).

(2) Let ϕ : (M,g) → (N, h) be a smooth map between two Riemannian manifolds, τ (ϕ) the tension field of ϕ (see [1, 2, 6]), f a smooth positive function on M, and let H be a smooth function on N, the (f,H)-tension field of ϕ is given by

τf,H(ϕ)=fτ(ϕ)+dϕ(gradMf)+(gradNH)ϕ,

where gradM (resp. gradN) denotes the gradient operator with respect to g (resp. h). Then ϕ is called f-harmonic with potential H if the (f,H)-tension field vanishes, i.e. τf,H(ϕ) = 0 (for more details on the concept of f-harmonic maps with potential H see [7]). The notion of f-harmonic with potential H is a generalization of harmonic maps with potential H if f ≡ 1, f-harmonic maps if H = 0 and the usual harmonic maps if f ≡ 1 and H = 0. We define the index form for f-harmonic maps with potential H by

If,Hϕ(v,w)=Mh(Jf,Hϕ(v),w)vM,

for all v,w ∈ Γ(ϕ−1TN), where

Jf,Hϕ(v)=-ftracegRN(v,dϕ)dϕ-tracegϕfϕv-(vNgradNH)ϕ,

RN is the curvature tensor of (N, h), ∇N is the Levi-Civita connection of (N, h), ∇ϕ denote the pull-back connection on ϕ−1TN, and vM is the volume form of (M,g) (see [1, 11]). If ϕ be a f-harmonic map with potential H and for any vector field v along ϕ, the index form satisfies If,Hϕ(v,v)0, then ϕ is called a stable f-harmonic map with potential H. Note that, the definition of stable f-harmonic maps with potential H is a generalization of stable harmonic maps if f = 1 on M and H = 0 on N (see [4, 16]).

For the smooth map ϕ : (M,g) → (N, h), S. Ouakkas et al. introduced in [12] the f-stress energy tensor Sf of ϕ associated to the f-energy functional

Ef(ϕ)=Mef(ϕ)vg,

is given by

Sf(ϕ)=ef(ϕ)g-fϕ*h,

where ef(ϕ)=12fdϕ2 is the f-energy density of ϕ. For any vector field X on M (see [12]), we have

divMSf(ϕ)(X)=-h(τf(ϕ),dϕ(X))+12X(f)dϕ2,

where τf (ϕ) = (ϕ) + (gradM f). If ϕ is a f-harmonic map with potential H, it follows that

divMSf(ϕ)(X)=h((gradNH)ϕ,dϕ(X))+12X(f)dϕ2.

Theorem 2.1

Let ϕ be a stable f-harmonic map with potential H from sphere ( , g) (n > 2) to Riemannian manifold (N, h), where f is a smooth positive function onsatisfying, (·)) ≥ 0, and H is a smooth function on N. Then, ϕ is constant.

Proof

Choose a normal orthonormal frame {ei} at point x0 in . Set

λ(x)=<α,x>n+1,

for all , where α ∈ ℝn+1 and let . Note that

v=<α,ei>n+1ei,XSnv=-λX,for all XΓ(TSn),traceg(Sn)2v=eiSneiSnv-eiSneiSnv=-v,

where is the Levi-Civita connection on with respect to the standard metric g of the sphere (see [16]). At point x0, we have

eiϕfeiϕdϕ(v)=gradSnfϕdϕ(v)+feiϕeiϕdϕ(v),

the first term of (2.1) is given by

gradSnfϕdϕ(v)=vϕdϕ(gradSnf)+dϕ([gradSnf,v])=vϕdϕ(gradSnf)+dϕ(gradSnfSnv)-dϕ(vSngradSnf),

the seconde term of (2.1) is given by

feiϕeiϕdϕ(v)=feiϕvϕdϕ(ei)+feiϕdϕ([ei,v])=fRN(dϕ(ei),dϕ(v))dϕ(ei)+fvϕeiϕdϕ(ei)+fdϕ([ei,[ei,v]])+2f[ei,v]ϕdϕ(ei),

from the definition of tension field, we get

feiϕeiϕdϕ(v)=-fRN(dϕ(v),dϕ(ei))dϕ(ei)+fvϕτ(ϕ)+fvϕdϕ(eiSnei)+fdϕ(eiSneiSnv)-fdϕ(eiSnvSnei)+2f[ei,v]ϕdϕ(ei)=-fRN(dϕ(v),dϕ(ei))dϕ(ei)+vϕfτ(ϕ)-v(f)τ(ϕ)+fvϕdϕ(eiSnei)+fdϕ(eiSneiSnv)-fdϕ(eiSnvSnei)+2f[ei,v]ϕdϕ(ei),

by equations (2.1), (2.2), (2.4), and the f-harmonicity with potential H condition of ϕ, we have (2.5)

eiϕfeiϕdϕ(v)=dϕ(gradSnfSnv)-dϕ(vSngradSnf)-fRN(dϕ(v),dϕ(ei))dϕ(ei)-vϕ(gradNH)ϕ-v(f)τ(ϕ)+fdϕ(vSneiSnei)+fdϕ(eiSneiSnv)-fdϕ(eiSnvSnei)+2feiSnvϕdϕ(ei),

by the definition of Ricci tensor, we get

eiϕfeiϕdϕ(v)=dϕ(gradSnfSnv)-dϕ(vSngradSnf)-fRN(dϕ(v),dϕ(ei))dϕ(ei)-vϕ(gradNH)ϕ-v(f)τ(ϕ)+fdϕ(RicciSnv)+fdϕ(trace(Sn)2v)+2feiSnvϕdϕ(ei),

from the property XSnv=-λX, we obtain

eiϕfeiϕdϕ(v)=-λdϕ(gradSnf)-dϕ(vSngradSnf)-fRN(dϕ(v),dϕ(ei))dϕ(ei)-vϕ(gradNH)ϕ-v(f)τ(ϕ)+fdϕ(RicciSnv)+fdϕ(trace(Sn)2v)-2λfτ(ϕ).

From the definition of Jacobi operator (1.7) and equation (2.7) we have

Jϕf(dϕ(v))=λdϕ(gradSnf)+dϕ(vSngradSnf)+v(f)τ(ϕ)-fdϕ(RicciSnv)-fdϕ(trace(Sn)2v)+2λfτ(ϕ),

since and (see [1, 16]), we conclude

h(Jϕf(dϕ(v)),dϕ(v))=λh(dϕ(gradSnf),dϕ(v))+h(dϕ(vSngradSnf),dϕ(v))+v(f)h(τ(ϕ),dϕ(v))-(n-2)fh(dϕ(v),dϕ(v))+2λfh(τ(ϕ),dϕ(v)),

by (2.9) and the f-harmonicity with potential H condition of ϕ, it follows that

traceαh(Jϕf(dϕ(v)),dϕ(v))=h(dϕ(ejSngradSnf),dϕ(ej))+h(τ(ϕ),dϕ(gradSnf))-(n-2)fdϕ2,

note that

h(τ(ϕ),dϕ(gradSnf))=h(eiϕdϕ(ei),dϕ(gradSnf))=divSnη-h(dϕ(ei),eiϕdϕ(gradSnf)),

with η(X) = h((X), . We obtain

traceαh(Jϕf(dϕ(v)),dϕ(v))=-h((dϕ)(ej,gradSnf),dϕ(ej))+divSnη-(n-2)fdϕ2,

since , from the stable f-harmonic with potential H condition, and equation (2.11), we get

0traceαIfϕ(dϕ(v),dϕ(v))+Snh((dϕ)(ej,gradSnf),dϕ(ej))vSn=-(n-2)Snfdϕ2vSn0.

Consequently, || = 0, that is ϕ is constant, because n > 2.

If f = 1 on , we get the following result:

Corollary 2.2

([14]) Any stable harmonic map ϕ from sphereto Riemannian manifold (N, h) is constant.

Corollary 2.3

([3]) Any stable harmonic map with potential from sphereto Riemannian manifold (N, h) is constant.

Using the similar technique we have:

Theorem 2.4

Let (M,g) be a compact Riemannian manifold, anda stable f-harmonic map with potential H, where f is a smooth positive function on M, and H is a smooth function onsatisfying. Then, ϕ is constant.

Proof

Choose a normal orthonormal frame {ei} at point x0 in M. When the same data of previous proof, we have

eiϕfeiϕ(vϕ)=gradMfϕ(vϕ)+feiϕeiϕ(vϕ),

the first term of (2.12) is given by

gradMfϕ(vϕ)=-(λϕ)dϕ(gradMf),

the seconde term of (2.12) is given by

feiϕeiϕ(vϕ)=-feiϕ(λϕ)dϕ(ei)=-fdϕ(gradM(λϕ))-(λϕ)fτ(ϕ),

by the definition of gradient operator, we get

-fdϕ(gradM(λϕ))=-f<dϕ(ei),vϕ>dϕ(ei),

substituting the formulas (2.13), (2.14), (2.15) into (2.12) gives

eiϕfeiϕ(vϕ)=-(λϕ)dϕ(gradMf)-f<dϕ(ei),vϕ>dϕ(ei)-(λϕ)fτ(ϕ),

from the f-harmonicity with potential H condition of ϕ, and equation (2.16), we have

<eiϕfeiϕ(vϕ),vϕ>=-f<dϕ(ei),vϕ><dϕ(ei),vϕ>+(λϕ)<(gradSnH)ϕ,vϕ>,

since the sphere has constant curvature, we obtain

<fRSn(vϕ,dϕ(ei))dϕ(ei),vϕ>=fdϕ2<vϕ,vϕ>-f<dϕ(ei),vϕ><dϕ(ei),vϕ>,

by the definition of Jacobi operator and equations (2.17), (2.18), we get

<Jϕf(vϕ),vϕ>=2f<dϕ(ei),vϕ><dϕ(ei),vϕ>-fdϕ2<vϕ,vϕ>-(λϕ)<(gradSnH)ϕ,vϕ>-<(vϕSngradSnH)ϕ,vϕ>,

so that

traceα<Jf,Hϕ(vϕ),vϕ>=(2-n)fdϕ2-(ΔSnH)ϕ,

and then

traceαIf,Hϕ(vϕ,vϕ)=(2-n)Mfdϕ2vM-M[(ΔSnH)ϕ]vM

Hence Theorem 2.4 follows from (2.20) and the stable f-harmonicity with potential H condition of ϕ with n > 2 and .

From Theorem 2.4, we deduce:

Corollary 2.5

([14]) Let (M,g) be a compact Riemannian manifold. When n > 2, any stable harmonic mapmust be constant.

Corollary 2.6

([5]) Let (M,g) be a compact Riemannian manifold. When n > 2, any stable f-harmonicmust be constant, where f is a smooth positive function on M.

Let ϕ: (M,g) → (N, h) be a smooth map. For any fixed x0M, by r(x) we denote the distance function from x0 to x, and by BR(x0) the geodesic ball with radius R and center x0. We say that the f-energy of ϕ is divergent slowly if there exists a positive function ψ(t) with R0dttψ(t)=(R0>0), such that

limRBR(x0)ef(ϕ)(x)ψ(r(x))<,

(see [7]). The next lemma is very useful in the sequel.

Lemma 3.1

Let ϕ: (Mm, g) → (Nn, h) be a smooth map, DM a compact domain such thatD is a smooth hypersurface in M. Let n denotes the unit normal vector ofD. Let X be any vector field in M with compact support. Then

Def(ϕ)g(X,n)=Dfh(dϕ(X),dϕ(n))+DdivMSf(ϕ)(X)+DSf(ϕ),X.

Here 〈,〉 denote the inner product on T*MT*M.

Proof

Choosing a local orthonormal frame field {ei} on M, and define X(ei,ej)=g(XMei,ej), then

divM(ef(ϕ)X)=g(eiM(ef(ϕ)X),ei)=g(eiM(ef(ϕ)X),ei)+ef(ϕ)g(eiMX,ei)=XMef(ϕ)+ef(ϕ)g(eiMX,ei),

and

XMef(ϕ)=12XM(fh(dϕ(ei),dϕ(ei)))=12X(f)dϕ2+fh((Xdϕ)ei,dϕ(ei))=12X(f)dϕ2+fh((eidϕ)X,dϕ(ei))=12X(f)dϕ2+h(eiϕdϕ(X),fdϕ(ei))-fh(dϕ(eiMX),dϕ(ei))=12X(f)dϕ2+eiMh(dϕ(X),fdϕ(ei))-h(dϕ(X),(ei(fdϕ))ei)-fh(dϕ(eiMX),dϕ(ei))=12X(f)dϕ2+divM(fh(dϕ(X),dϕ(ei))ei)-h(dϕ(X),τf(ϕ))-fX,ϕ*h.

Hence we obtain

divM(ef(ϕ)X)=12X(f)dϕ2+divM(fh(dϕ(X),dϕ(ei))ei)-h(dϕ(X),τf(ϕ)-fX,ϕ*h+ef(ϕ)g(eiMX,ei))=12X(f)dϕ2+divM(fh(dϕ(X),dϕ(ei))ei)-h(dϕ(X),τf(ϕ))+Sf(ϕ),X.

Now, for compact domain D in M with its smooth hypersurface ∂D, taking local orthonormal frame field {ei} on M along ∂D, such that {e1, …, em−1} ∈ Γ(TD), and em = n be the unit normal vector of ∂D. Since SuppX is compact, integrating the formula (3.2) on D, by means of Green’s theorem and using (1.9), we have the desired formula.

Theorem 3.2

Let M be an m–dimensional complete, simply connected Riemannian manifold with non-positive sectional curvature KM, m > 2. Assuming that KM satisfies

  • a2 < KM < −b2, where a > 0, b>0 and(m-1)b2a; or

  • -A1+r2KM0, where0<A<m(m-2)4,

assume that ϕ is a f-harmonic map with potential H from BR(x0) to any Riemannian manifold N with ϕ |BR(x0)= P, where PN satisfiesH(P)=maxyNH(y), and X(f) ≥ 0 such thatX=rr. Then ϕ must be constant in BR(x0).

Proof

First of all, from the definition of Sf (ϕ), we obtain

Sf(ϕ),X=(ef(ϕ)g(eα,eβ)-fh(dϕ(eα),dϕ(eβ)))g(eαMX,eβ)=ef(ϕ)g(eαMX,eα)-fh(dϕ(eα),dϕ(eβ))g(eαMX,eβ).

Let eα={es,r} be the orthonormal frame field of BR(x0) and X=rr, then

rMX=r,esMX=resMr=rHessM(r)(es,et)et,divMX=g(eαMX,eα)=1+rHessM(r)(es,es).

Substituting (3.4), (3.5) and (3.6) into (3.3), we get

Sf(ϕ),X=ef(ϕ)(1+rHessM(r)(es,es))-fh(dϕ(es),dϕ(et))g(esMX,et)-fh(dϕ(r),dϕ(r))g(rMX,r)-fh(dϕ(r),dϕ(et))g(rMX,et)-fh(dϕ(es),dϕ(r))g(esMX,r)=ef(ϕ)(1+rHessM(r)(es,es))-fh(dϕ(es),dϕ(et))rHessM(r)(es,et)-fh(dϕ(r),dϕ(r)).

Under the assumption (1) in Theorem 3.2, from Hessian comparison theorem (see [8]) we have

bcoth(br)(g-drdr)HessM(r)acoth(ar)(g-drdr).

Therefore, (3.7) becomes

Sf(ϕ),Xef(ϕ)(1+(m-1)(br)coth(br))-f(ar)coth(ar)h(dϕ(es),dϕ(es))-fh(dϕ(r),dϕ(r))=f(m-12(br)coth(br)-12)h(dϕ(r),dϕ(r))+f(12+m-12(br)coth(br)-(ar)coth(ar))h(dϕ(es),dϕ(es))m-22fh(dϕ(r),dϕ(r))+f(12+rcoth(br)(m-12b-a))h(dϕ(es),dϕ(es)).

Hence, when (m-1)b2a, it follows from (3.9)

Sf(ϕ),Xδef(ϕ),

where δ > 0.

Under the assumption (2), also by Hessian comparison theorem (see [8]) we have

1r(g-drdr)HessM(r)βr(g-drdr),

where β=12+12(1+4A)12. By (3.7), it follows that

Sf(ϕ),Xmef(ϕ)-fβh(dϕ(es),dϕ(es))-fh(dϕ(r),dϕ(r))=m-22fh(dϕ(r),dϕ(r))+m-2β2fh(dϕ(es),dϕ(es))δef(u).

Then, under the two assumptions of Theorem 3.2 we obtain

Sf(ϕ),Xδef(ϕ),

where δ > 0. Now choosing the geodesic polar coordinates (θ, r) in BR(x0) and a local orthonormal frame field {e1, …, em−1, r} on M. After applying D = BR(x0), X=rr and n=r to (3.1), we get

RBR(x0)ef(ϕ)=RBR(x0)fdϕ(r)2+BR(x0)divMSf(ϕ)(rr)+BR(x0)Sf(ϕ),X.

Noting that ϕ is f-harmonic map with potential H, and using (1.10), we have

divMSf(ϕ)(rr)=BR(x0)r(Hϕ)r+12BR(x0)rfrdϕ2,

so, (3.11) becomes

RBR(x0)ef(ϕ)=RBR(x0)fdϕ(r)2+BR(x0)r(Hϕ)r+12BR(x0)rfrdϕ2+BR(x0)Sf(ϕ),X.

Since ϕ is constant at ∂BR(x0), by (3.10) and (3.12), we have

BR(x0)r(Hϕ)r+12BR(x0)rfrdϕ2+δBR(x0)ef(ϕ)0.

Denote J(θ, r)dθdr the volume element of BR(x0) in polar coordinates around x0. Since r(rJ(θ,r))>0 (see [3]), we obtain

0Rr(Hϕ)rJ(θ,r)dr=RJ(θ,R)H(P)-0RHϕ(θ,r)(rJ(θ,r))rdrRJ(θ,R)H(P)-H(P)0R(rJ(θ,r))rdr0.

Therefore

BR(x0)r(Hϕ)r=BR(x0)(0Rr(Hϕ)rJ(θ,r)dr)dθ0.

By (3.13) and (3.14) and X(f) ≥ 0, we immediately conclude that ef (ϕ) ≡ 0 in BR(x0), namely, ϕ is constant in BR(x0), which completes the proof of Theorem 3.2.

Remark 3.3

Consider the following static Landau-Lifshitz equation

Δϕ+ϕdϕ2-<H0,ϕ>3ϕ+H0=0,

where |ϕ(x)|2 = 1, x ∈ Ω ⊂ ℝm, H0 ≠ 0 is a constant vector in ℝ3. Then the solution ϕ of (3.15) can be seen as a harmonic map with potential: with the potential H(y) =< H0, y >3, (see [3]). Moreover, Hong [9] asserted that the static Landau-Lifshitz equation (3.15) with constant boundary-value problem ϕΩ=H0H0, has only constant solution, if Ω = B3, where B3 denote the unit ball in ℝ3. On the other hand, if we choose M = ℝm(m > 2), , H(y) =< H0, y >3, , then Theorem 3.2 for f ≡ 1 leads to a conclusion for the static Landau-Lifshitz equation, in particular, when m = 3, it is just the result of Hong. Theorem 3.2 also generalizes the result of [10] for the usual harmonic maps and Theorem 3 in [3] for the harmonic maps with potential.

For f-harmonic maps, we have

Theorem 3.4

Let M be as in Theorem 3.2. If ϕ is a f-harmonic map from M whose f-energy is finite or divergent slowly. Then ϕ must be a constant map when X(f) ≥ 0.

Proof

By setting D = BR(x0), X=rr and n=r in (3.1), we obtain

BR(x0)(divMSf(ϕ))(X)+BR(x0)Sf(ϕ),X=RBR(x0)ef(ϕ)-RBR(x0)fdϕ(r)2RBR(x0)ef(ϕ).

According to (1.9), (3.10) and (3.16), for a f-harmonic map ϕ, we get

RBR(x0)ef(ϕ)δBR(x0)ef(ϕ)+12BR(x0)X(f)dϕ2δBR(x0)ef(ϕ).

Now suppose that ϕ is a nonconstant map, i.e. the f-energy density ef (ϕ) does not vanish everywhere, so there exists R0 > 0 such that for R > R0,

BR(x0)ef(ϕ)C0,

where C0 be a positive constant. So when R > R0, we have from (3.17) and (3.18)

BR(x0)ef(ϕ)δC0R,

therefore, (3.19) will imply

Ef(ϕ)>BR(x0)ef(ϕ)=0R(BR(x0)ef(ϕ))drR0R(BR(x0)ef(ϕ))drR0RδC0rdr=δC0lnRR0.

Let R → ∞, this contradicts the assumption of the finite f-energy, then ϕ is constant. If the f-energy of ϕ divergent slowly, therefore (3.19) leads to

limRBR(x0)ef(ϕ)ψ(r(x))=0drψ(r)BR(x0)ef(ϕ)δC00drrψ(r)δC0R0drrψ(r)=.

Which is in contradiction with f-energy of ϕ being slowly divergent. So ϕ must be a constant map.

Remark 3.5

When f ≡ 1, it is clear that Theorem 3.4 recovers the results due to Sealey [13] and Xin [16] as special cases. If the manifold M in the Theorems 3.2 and 3.4 satisfies −a2 < KM < 0 and RicM < −b2 < 0 with b > 2a, then this two theorems remain true. Note that this kind of manifolds includes the bounded symmetric domains and complex hyperbolic spaces see ([15]).

The authors would like to thank the reviewers for their useful remarks and suggestions. The authors are supported by National Agency Scientific Research of Algeria and Laboratory of Geometry, Analysis, Controle and Applications, Algeria.

  1. Baird, P, and Wood, JC (2003). Harmonic morphisms between Riemannain manifolds. Oxford: Clarendon Press, Oxford University press
    CrossRef
  2. Caddeo, R, Montaldo, S, and Oniciuc, C (2001). Biharmonic submanifolds of . Internat J Math. 12, 867-876.
    CrossRef
  3. Chen, Q (2000). Stability and constant boundary-value problems of harmonic maps with potential. J Austral Math Soc Ser A. 68, 145-154.
    CrossRef
  4. Cherif, AM, and Djaa, M (2015). Geometry of energy and bienergy variations between Riemannian Manifolds. Kyungpook Math J. 55, 715-730.
    CrossRef
  5. Cherif, AM, Djaa, M, and Zegga, K (2015). Stable f-harmonic maps on sphere. Commun Korean Math Soc. 30, 471-479.
    CrossRef
  6. Eells, J, and Sampson, JH (1964). Harmonic mappings of Riemannian manifolds. Amer J Math. 86, 109-160.
    CrossRef
  7. Feng, S, and Han, Y (2014). Liouville type theorems of f-harmonic maps with potential. Results Math. 66, 43-64.
    CrossRef
  8. Greene, RE, and Wu, H (1979). Function theory on manifolds which possess a pole. Lecture Notes in Mathematics 699. Berlin: Springer-Verlag
    CrossRef
  9. Hong, MC, and Lemaire, L (1995). Multiple solutions of the static Landau-Lifshitz equation from B2 into S2. Math Z. 220, 295-306.
    CrossRef
  10. Karcher, J, and Wood, JC (1984). Nonexistence results and growth properties for harmonic maps and forms. J Reine Angew Math. 353, 165-180.
    CrossRef
  11. O’Neil, B (1983). Semi-Riemannian Geometry. New York: Academic Press
  12. Ouakkas, S, Nasri, R, and Djaa, M (2010). On the f-harmonic and f-biharmonic maps. JP J Geom Topol. 10, 11-27.
  13. Sealey, HCJ (1982). Some conditions ensuring the vanishing of harmonic differential forms with applications to harmonic maps and Yang–Mills theory. Math Proc Cambridge Philos Soc. 91, 441-452.
    CrossRef
  14. Xin, Y (1980). Some results on stable harmonic maps. Duke Math J. 47, 609-613.
    CrossRef
  15. Xin, Y (1995). Harmonic maps of bounded symmetric domains. Math Ann. 303, 417-433.
    CrossRef
  16. Xin, Y (1996). Geometry of harmonic maps. Boston, Boston, MA: Birkhäuser
    CrossRef