Exact Rayleigh Reduction for Direct Detection of Turning Bifurcations

Yavdat Il’yasov Address: Institute of Mathematics with Computing Centre, Ufa Federal Research Centre of the Russian Academy of Sciences, Ufa, Russia Email address: ilyasov02@gmail.com

Abstract.

We introduce an exact Rayleigh reduction for the direct construction and detection of turning bifurcations in nonlinear parameter-dependent equations. A generalized Rayleigh functional ordinarily provides only a scalar necessary constraint on solutions. We show that, along a suitable one-parameter transformation orbit $u=\mathcal{G}_{s}\phi$ , it can instead become the exact parameter law $\lambda=\mathcal{R}(\mathcal{G}_{s}\phi)$ of a genuine solution branch. Turning points are then found from nondegenerate critical points of this one-dimensional Rayleigh profile. The branch tangent automatically yields a kernel direction of the state linearization, and under the usual Fredholm, kernel-simplicity, and parameter-transversality assumptions the detected point is a simple fold.

For Kirchhoff equations, amplitude scaling on bounded domains and spatial dilation on $\mathbb{R}^{N}$ yield explicit exact branches, global parameter thresholds, and countable hierarchies of turning values. For the generalized Kirchhoff law $M_{b}(A)=a+bA^{\theta}$ , the branch geometry is governed by the dimension–homogeneity index $\theta(N-2)-2$ : an interior turning bifurcation occurs exactly when $\theta(N-2)>2$ . This gives a universal bifurcation trichotomy that is independent of the particular Berestycki–Lions nonlinearity.

Key words and phrases:

exact Rayleigh reduction, generalized Rayleigh functional, turning bifurcation, Kirchhoff equation, scaling orbit, normalized profile, dimension–homogeneity trichotomy

2020 Mathematics Subject Classification

35B32, 35J20, 35J60, 47J30, 58E05

1. Introduction

A central problem in nonlinear analysis is to determine the parameter values at which the structure of solutions changes. Such values mark the appearance, merging, or disappearance of solution branches and arise naturally in nonlinear elliptic equations, variational problems, and models with nonlocal interactions. Classical bifurcation and singularity theories provide powerful tools for describing the local structure near such points; see, for example, [1, 12, 30].

Throughout the paper, by a turning bifurcation we mean a point at which two nearby solution states coalesce as the parameter reaches a critical value and the corresponding solution branch turns with respect to the parameter. A classical simple fold is obtained when the usual additional Fredholm and transversality conditions hold.

There is, however, a complementary constructive question: can such a critical parameter be detected directly from the nonlinear equation before the singular point itself is known? This is particularly relevant when the solution set is not explicitly available and one seeks not only the critical value, but also the mechanism responsible for the change in solution geometry.

The purpose of this paper is to develop a generalized Rayleigh approach to this problem. Consider

F(\lambda,u)=0,\qquad F:\mathbb{R}\times X\to Y,

and suppose that every nontrivial solution satisfies a parameter relation

(1)

\lambda=\mathcal{R}(u),

where $\mathcal{R}$ is a generalized Rayleigh functional. In general, (1) is only a necessary scalar constraint:

\lambda=\mathcal{R}(u)\qquad\not\Longrightarrow\qquad F(\lambda,u)=0.

The main concept introduced here is an exact Rayleigh reduction. We seek a one-parameter transformation orbit

u(s)=\mathcal{G}_{s}\phi

and distinguished profiles $\phi$ for which the Rayleigh relation becomes the actual parameter law of an exact solution branch:

(2)

F\bigl(\mathcal{R}(\mathcal{G}_{s}\phi),\mathcal{G}_{s}\phi\bigr)=0.

For every such profile,

(3)

\lambda_{\phi}(s)=\mathcal{R}(\mathcal{G}_{s}\phi)

is the parameter along the exact branch

s\longmapsto\bigl(\lambda_{\phi}(s),\mathcal{G}_{s}\phi\bigr).

Thus an infinite-dimensional bifurcation problem is reduced, along an exact orbit, to the geometry of the one-dimensional Rayleigh profile $\lambda_{\phi}$ . Its nondegenerate critical points detect turning bifurcations of the exact branch.

In applications, the exactness condition (2) typically reduces to a normalized profile equation whose solutions select the admissible exact branches. A turning bifurcation candidate is then located directly from

(4)

\lambda_{\phi}^{\prime}(s)=0.

At a nondegenerate critical point $s_{*}$ , differentiation of the exact branch identity gives

F_{u}(\lambda_{*},u_{*})u^{\prime}(s_{*})=0,\qquad u_{*}=\mathcal{G}_{s_{*}}\phi,\qquad\lambda_{*}=\lambda_{\phi}(s_{*}).

Hence the tangent to the explicitly constructed branch provides a kernel direction of the state linearization. Under additional kernel-simplicity, Fredholm, and parameter-transversality assumptions, the detected turning bifurcation is a simple fold. Whenever the operator realization is of class $C^{2}$ , this agrees with the classical Lyapunov–Schmidt simple-fold criterion.

This gives generalized Rayleigh quotients a different role. Rather than merely identifying distinguished parameter levels, they may reconstruct exact portions of the nonlinear solution set itself. Multiplicity, turning thresholds, and branch hierarchies can then be read directly from the geometry of explicit Rayleigh profiles. Exact Rayleigh reduction and local bifurcation theory therefore play complementary roles: the former constructs the relevant branch and locates its critical parameter, while the latter provides a local classification of the singularity.

The approach originates from nonlinear Rayleigh quotient methods developed for Nehari and Pohozaev manifolds [14, 16, 17, 19]. There, critical levels of generalized Rayleigh functionals were used to characterize distinguished parameter values and extremal thresholds. Related minimax constructions for singular parameter values were developed through extended Rayleigh quotient methods [15, 18, 20]. The present paper takes a further step: under an appropriate transformation, the Rayleigh functional becomes the parameter law of an exact solution branch whose geometry reveals the bifurcation structure.

We develop this mechanism for Kirchhoff equations. These equations originate in Kirchhoff’s model for the transverse vibrations of a stretched elastic string, in which the tension depends on the global deformation of the string and therefore introduces a genuinely nonlocal feedback [22, 27]. Mathematically, the Kirchhoff coefficient couples a local differential operator to a global energy of the state. This structure is particularly well suited to exact Rayleigh reduction, since the global energy interacts naturally with scaling transformations. The method then isolates the competition between this nonlocal feedback and the homogeneity of the underlying scaling. In this sense, Kirchhoff equations serve here as a model class for the reduction mechanism rather than as its endpoint.

Two different reductions are considered: amplitude scaling on bounded domains and spatial dilation on $\mathbb{R}^{N}$ .

On bounded domains, the method remains exact in the presence of a positive spatial weight $k(x)$ . For the weighted Kirchhoff problem, the Nehari Rayleigh functional combined with amplitude scaling generates exact branches from solutions of a weighted normalized profile equation. The resulting Rayleigh geometry yields explicit turning bifurcation values, a global upper existence threshold, and a countable hierarchy of exact branches. Remarkably, the spatial weight changes the profile-dependent bifurcation levels but leaves the critical amplitude of every exact branch unchanged.

On $\mathbb{R}^{N}$ , spatial dilation leads naturally to a Pohozaev Rayleigh functional and reduces the Kirchhoff equation to the nonlinear scalar-field equation

-\Delta\phi=g(\phi).

Every normalized profile generates an exact spatial branch, while the Nehari representation, when applicable, gives the same parameter law along that branch. The resulting reduction separates the scalar-field profile from the geometry induced by the Kirchhoff scaling.

For some particular classes of Kirchhoff equations, parameter thresholds associated with changes in the number of solutions have previously been obtained; see [25, 26, 31]. Scalar rescaling relations underlying positive solutions, together with the corresponding two–one–zero multiplicity pattern, also appear in the Kirchhoff literature; see [2, 3, 13]. These works provide important existence, threshold, and multiplicity results, and in some cases describe the behavior of solutions near critical parameter values. The present approach addresses the bifurcation structure more directly: critical parameter values are located together with the associated turning bifurcation points, the resulting solution branches are constructed globally through an exact Rayleigh reduction, and the same construction yields a countable hierarchy of exact bifurcation branches. Moreover, the mechanism extends beyond the standard Kirchhoff law.

Indeed, for the generalized Kirchhoff law

M_{b}(A)=a+bA^{\theta},

the exact spatial reduction yields

b_{\phi}(s)=\frac{1-as^{2}}{A(\phi)^{\theta}}s^{\theta(N-2)-2}.

Thus the branch geometry is governed by the dimension–homogeneity index

\theta(N-2)-2.

In particular,

\boxed{\theta(N-2)>2}

is the sharp condition for the occurrence of an interior turning bifurcation. The supercritical, critical, and subcritical cases $\theta(N-2)>2$ , $\theta(N-2)=2$ , and $\theta(N-2)<2$ lead to three distinct branch geometries. The familiar dimensional distinction for the standard Kirchhoff equation is therefore a special case of a more general dimension–homogeneity balance.

The same reduction produces a countable hierarchy of exact spatial branches. Their turning values are ordered inversely by the energies of the normalized scalar-field profiles and accumulate at zero, while a ground-state profile generates the maximal turning value and the global upper existence threshold. Thus the exact Rayleigh reduction describes simultaneously the turning mechanism, the global threshold, and the multiplicity structure.

The paper is organized as follows. Section 2 develops the abstract exact Rayleigh reduction and its relation to classical local fold theory. The subsequent sections apply the method to weighted bounded-domain and whole-space Kirchhoff equations and study the corresponding exact branches, turning bifurcation thresholds, and branch hierarchies.

2. Exact Rayleigh reduction and direct bifurcation detection

We develop here the abstract mechanism underlying the Rayleigh approach to bifurcation. The starting point is a generalized Rayleigh representation of the parameter, but the essential new step is to combine it with a one-parameter transformation so that the resulting scalar Rayleigh profile becomes the actual parameter law of an exact solution branch.

Let

F(\lambda,u)=0,\qquad F:\mathbb{R}\times X\to Y,

where $X$ and $Y$ are Banach spaces. Suppose that every nontrivial solution satisfies a scalar constraint

\mathcal{C}(\lambda,u)=0,

which can be resolved with respect to the parameter:

(5)

\mathcal{C}(\lambda,u)=0\qquad\Longleftrightarrow\qquad\lambda=\mathcal{R}(u).

We call $\mathcal{R}$ the generalized Rayleigh functional associated with the constraint.

Such representations arise naturally from Nehari, Pohozaev, energy, and other scalar identities. They form the basis of nonlinear generalized Rayleigh quotient methods for detecting distinguished parameter levels [17, 19]. By themselves, however, they describe only constraint geometry:

\lambda=\mathcal{R}(u)\qquad\not\Longrightarrow\qquad F(\lambda,u)=0.

The purpose of the reduction below is to identify situations in which this implication becomes exact along a suitable transformation orbit.

2.1. Orbital Rayleigh profiles and exact reduction

Let

\mathcal{G}_{s}:X\to X,\qquad s>0,\qquad\mathcal{G}_{1}=\mathrm{Id},

be a one-parameter family of transformations, and let

\mathcal{O}\subset X\setminus\{0\}

be an admissible class invariant under this action:

(6)

\mathcal{G}_{s}(\mathcal{O})\subset\mathcal{O},\qquad s>0.

For $u\in\mathcal{O}$ , define the orbital Rayleigh profile

(7)

r_{u}(s):=\mathcal{R}(\mathcal{G}_{s}u),\qquad s>0.

The scalar equation

(8)

r_{u}(s)=\lambda

describes the intersections of the transformation orbit with the Rayleigh level $\mathcal{R}=\lambda$ . Its critical points satisfy

(9)

r_{u}^{\prime}(s)=0.

At this stage these are only critical points of the constraint geometry, not necessarily bifurcation points of the original equation.

The decisive situation occurs when the Rayleigh representation is exact along an orbit. We say that a profile $\phi\in\mathcal{O}$ generates an exact Rayleigh reduction if

(10)

F\bigl(\mathcal{R}(\mathcal{G}_{s}\phi),\mathcal{G}_{s}\phi\bigr)=0,\qquad s>0.

Then

(11)

\lambda_{\phi}(s):=\mathcal{R}(\mathcal{G}_{s}\phi)

is the actual parameter law of the solution branch

s\longmapsto\bigl(\lambda_{\phi}(s),\mathcal{G}_{s}\phi\bigr).

Thus the geometry of the nonlinear solution set along this branch is reduced exactly to the geometry of a one-dimensional Rayleigh profile.

This is the basic Rayleigh reduction principle: once an exact profile $\phi$ has been identified, turning bifurcations can be detected directly from the scalar condition

(12)

\boxed{\lambda_{\phi}^{\prime}(s_{*})=0.}

2.2. Rayleigh bifurcation criterion

The following result makes the preceding geometric mechanism precise.

Theorem 2.1 (Rayleigh bifurcation criterion).

Let $\phi\in\mathcal{O}$ generate an exact Rayleigh branch

F\bigl(\lambda_{\phi}(s),\mathcal{G}_{s}\phi\bigr)=0,\qquad\lambda_{\phi}(s)=\mathcal{R}(\mathcal{G}_{s}\phi).

Assume that $\lambda_{\phi}$ is of class $C^{2}$ near some $s_{*}>0$ and

\lambda_{\phi}^{\prime}(s_{*})=0,\qquad\lambda_{\phi}^{\prime\prime}(s_{*})\neq 0.

Assume also that the orbit is of class $C^{1}$ and regular at $s_{*}$ :

\left.\frac{d}{ds}\mathcal{G}_{s}\phi\right|_{s=s_{*}}\neq 0.

Then

(\lambda_{\phi}^{*},u_{*}),\qquad\lambda_{\phi}^{*}:=\lambda_{\phi}(s_{*}),\qquad u_{*}:=\mathcal{G}_{s_{*}}\phi,

is a nondegenerate turning point of the exact Rayleigh branch; in the terminology adopted above, it is a turning bifurcation of that branch. More precisely:

•

if $\lambda_{\phi}^{\prime\prime}(s_{*})<0$ , then there exists $\varepsilon>0$ such that for every

$\lambda\in(\lambda_{\phi}^{*}-\varepsilon,\lambda_{\phi}^{*})$

there are two distinct exact solutions

$u_{-}(\lambda),u_{+}(\lambda),$

with

$F(\lambda,u_{\pm}(\lambda))=0,\qquad u_{\pm}(\lambda)\to u_{*}\quad\text{as }\lambda\uparrow\lambda_{\phi}^{*};$
•

if $\lambda_{\phi}^{\prime\prime}(s_{*})>0$ , then there exists $\varepsilon>0$ such that for every

$\lambda\in(\lambda_{\phi}^{*},\lambda_{\phi}^{*}+\varepsilon)$

there are two distinct exact solutions

$u_{-}(\lambda),u_{+}(\lambda),$

with

$F(\lambda,u_{\pm}(\lambda))=0,\qquad u_{\pm}(\lambda)\to u_{*}\quad\text{as }\lambda\downarrow\lambda_{\phi}^{*}.$

Proof.

We consider the case

\lambda_{\phi}^{\prime\prime}(s_{*})<0;

the case $\lambda_{\phi}^{\prime\prime}(s_{*})>0$ is completely analogous, with the inequalities reversed.

Since

\lambda_{\phi}^{\prime}(s_{*})=0,\qquad\lambda_{\phi}^{\prime\prime}(s_{*})<0,

the point $s_{*}$ is a strict nondegenerate local maximum of the scalar profile. Hence, for $\lambda<\lambda_{\phi}^{*}$ sufficiently close to $\lambda_{\phi}^{*}$ , there exist

s_{-}(\lambda)<s_{*}<s_{+}(\lambda)

such that

\lambda_{\phi}(s_{\pm}(\lambda))=\lambda.

By exactness,

u_{\pm}(\lambda):=\mathcal{G}_{s_{\pm}(\lambda)}\phi

are solutions of $F(\lambda,u)=0$ .

The regularity condition

\left.\frac{d}{ds}\mathcal{G}_{s}\phi\right|_{s=s_{*}}\neq 0

also implies local injectivity of the orbit near $s_{*}$ . Indeed, by the Hahn–Banach theorem one can choose $\ell\in X^{*}$ such that $\ell(u^{\prime}(s_{*}))\neq 0$ , where $u(s)=\mathcal{G}_{s}\phi$ ; hence $\ell(u(s))$ is strictly monotone in a sufficiently small neighborhood of $s_{*}$ . Therefore $s_{-}(\lambda)\neq s_{+}(\lambda)$ gives

u_{-}(\lambda)\neq u_{+}(\lambda).

Finally,

s_{\pm}(\lambda)\to s_{*}

as $\lambda\uparrow\lambda_{\phi}^{*}$ , and continuity of the orbit yields

u_{\pm}(\lambda)\to\mathcal{G}_{s_{*}}\phi=u_{*}.

For $\lambda_{\phi}^{\prime\prime}(s_{*})>0$ , the point $s_{*}$ is instead a strict nondegenerate local minimum, and the same argument gives two exact solutions for $\lambda>\lambda_{\phi}^{*}$ sufficiently close to $\lambda_{\phi}^{*}$ . ∎

The importance of this criterion is that the turning point is obtained from the explicit Rayleigh geometry of the exact branch itself. Neither a singular point nor a kernel of the linearized operator needs to be known in advance.

2.3. Class-dependent Rayleigh thresholds

The exact reduction may also be combined with a global Rayleigh threshold. Let

\mathcal{S}\subseteq\mathcal{O}

be an admissible class, for example a positive, nonnegative, radial, or otherwise invariant class.

For $u\in\mathcal{S}$ , define the upper orbital level

(13)

m_{\mathcal{G}}(u):=\sup_{s>0}\mathcal{R}(\mathcal{G}_{s}u),

and the corresponding class-dependent Rayleigh threshold

(14)

\boxed{\lambda_{\mathcal{G},\mathcal{S}}^{**}:=\sup_{u\in\mathcal{S}}\sup_{s>0}\mathcal{R}(\mathcal{G}_{s}u).}

If $\mathcal{S}$ is invariant under the transformation,

\mathcal{G}_{s}(\mathcal{S})\subseteq\mathcal{S},\qquad s>0,

then

(15)

\boxed{\lambda_{\mathcal{G},\mathcal{S}}^{**}=\sup_{u\in\mathcal{S}}\mathcal{R}(u).}

Indeed, $\mathcal{G}_{1}=\mathrm{Id}$ gives

\mathcal{R}(u)\leq\sup_{s>0}\mathcal{R}(\mathcal{G}_{s}u),

whereas invariance implies

\sup_{s>0}\mathcal{R}(\mathcal{G}_{s}u)\leq\sup_{w\in\mathcal{S}}\mathcal{R}(w).

Consequently,

(16)

\boxed{F(\lambda,u)=0,\quad u\in\mathcal{S}\qquad\Longrightarrow\qquad\lambda\leq\lambda_{\mathcal{G},\mathcal{S}}^{**}.}

Thus $\lambda_{\mathcal{G},\mathcal{S}}^{**}$ is the maximal Rayleigh level compatible with solutions in $\mathcal{S}$ .

If an exact branch generated by $\phi\in\mathcal{S}$ satisfies

(17)

\lambda_{\phi}(s_{*})=\max_{s>0}\lambda_{\phi}(s)=\lambda_{\mathcal{G},\mathcal{S}}^{**},

then the turning bifurcation point detected by Theorem 2.1 occurs at the maximal Rayleigh threshold of the class. In this case the same Rayleigh construction identifies both the local turning mechanism and the global upper parameter boundary.

Remark 2.2 (Lower Rayleigh thresholds).

The preceding construction is formulated for upper Rayleigh thresholds and therefore corresponds naturally to exact branches whose Rayleigh profiles attain a maximum. If instead

\lambda_{\phi}^{\prime\prime}(s_{*})>0,

Theorem 2.1 detects a turning bifurcation at a local minimum. An analogous lower-threshold construction is obtained by replacing $\sup$ and $\max$ with $\inf$ and $\min$ and reversing the corresponding inequalities. Since the applications below concern upper existence thresholds, we use only the upper formulation.

Remark 2.3 (Relation with classical bifurcation methods).

The exact Rayleigh reduction is complementary to classical local bifurcation theory. In implicit-function and Lyapunov–Schmidt approaches one usually starts from a known singular solution

F(\lambda_{*},u_{*})=0,\qquad\ker F_{u}(\lambda_{*},u_{*})\neq\{0\},

and then studies the local solution set near $(\lambda_{*},u_{*})$ ; see, for example, [9, 21]. Global and topological methods study continua and branching of solutions; see [23, 28], while numerical continuation detects singular points along computed solution branches [10, 11].

The exact Rayleigh reduction reverses this order when an appropriate transformation is available. It first constructs an exact solution branch and detects its critical parameter from

\lambda_{\phi}^{\prime}(s_{*})=0.

The resulting point may then be analyzed by classical local, global, or numerical bifurcation methods.

2.4. Connection with the classical fold criterion

The Rayleigh bifurcation criterion is geometric and does not require a Fredholm decomposition or a Lyapunov–Schmidt reduction. Nevertheless, the geometry of an exact Rayleigh branch naturally contains the classical fold conditions when the required operator assumptions are available.

Assume that $F$ is of class $C^{2}$ and let

u(s)=\mathcal{G}_{s}\phi,\qquad\lambda_{\phi}(s)=\mathcal{R}(\mathcal{G}_{s}\phi)

be an exact branch:

F(\lambda_{\phi}(s),u(s))=0.

Differentiating with respect to $s$ gives

(18)

F_{\lambda}(\lambda_{\phi}(s),u(s))\lambda_{\phi}^{\prime}(s)+F_{u}(\lambda_{\phi}(s),u(s))u^{\prime}(s)=0.

At a critical point $s_{*}$ ,

\lambda_{\phi}^{\prime}(s_{*})=0,

and therefore

(19)

\boxed{F_{u}(\lambda_{\phi}^{*},u_{*})u^{\prime}(s_{*})=0,}

where

u_{*}=\mathcal{G}_{s_{*}}\phi,\qquad\lambda_{\phi}^{*}=\lambda_{\phi}(s_{*}).

Thus the tangent direction to the exact Rayleigh orbit automatically belongs to the kernel of the state linearization. The kernel direction is therefore a consequence of the Rayleigh reduction rather than an assumption used to locate the critical point.

Differentiating the exact branch identity once more and using $\lambda_{\phi}^{\prime}(s_{*})=0$ yields

F_{u}(\lambda_{\phi}^{*},u_{*})u^{\prime\prime}(s_{*})+D_{uu}F(\lambda_{\phi}^{*},u_{*})[u^{\prime}(s_{*}),u^{\prime}(s_{*})]+\lambda_{\phi}^{\prime\prime}(s_{*})F_{\lambda}(\lambda_{\phi}^{*},u_{*})=0.

Let

0\neq\psi\in\ker F_{u}(\lambda_{\phi}^{*},u_{*})^{*}.

Pairing with $\psi$ eliminates the first term and gives

(20)

\boxed{\left\langle\psi,D_{uu}F(\lambda_{\phi}^{*},u_{*})[u^{\prime}(s_{*}),u^{\prime}(s_{*})]\right\rangle=-\lambda_{\phi}^{\prime\prime}(s_{*})\left\langle\psi,F_{\lambda}(\lambda_{\phi}^{*},u_{*})\right\rangle.}

Suppose, in addition, that

\ker F_{u}(\lambda_{\phi}^{*},u_{*})=\operatorname{span}\{u^{\prime}(s_{*})\},

that $F_{u}(\lambda_{\phi}^{*},u_{*})$ is Fredholm of index zero, and that the parameter transversality condition

\left\langle\psi,F_{\lambda}(\lambda_{\phi}^{*},u_{*})\right\rangle\neq 0

holds. If

\lambda_{\phi}^{\prime\prime}(s_{*})\neq 0,

then (20) shows that

\left\langle\psi,D_{uu}F(\lambda_{\phi}^{*},u_{*})[u^{\prime}(s_{*}),u^{\prime}(s_{*})]\right\rangle\neq 0.

Hence the quadratic Lyapunov–Schmidt coefficient is nonzero, and $(\lambda_{\phi}^{*},u_{*})$ is a classical simple fold.

Thus exact Rayleigh reduction and classical fold theory provide two different levels of information. The Rayleigh reduction constructs the branch and detects its critical parameter; the classical fold conditions, when available, identify the local singularity type of the detected point.

3. Amplitude Rayleigh geometry on bounded domains

We first illustrate the exact Rayleigh reduction by the weighted Kirchhoff problem

(21)

\begin{cases}-\left(a+\lambda\displaystyle\int_{\Omega}|\nabla u|^{2}\,dx\right)\Delta u=k(x)|u|^{q-2}u,&x\in\Omega,\\[5.69054pt] u=0,&x\in\partial\Omega,\end{cases}

where $\Omega\subset\mathbb{R}^{N}$ is a bounded smooth domain,

a>0,\qquad k\in C(\overline{\Omega}),\qquad k(x)>0\quad\text{on }\overline{\Omega},

and

2<q<\min\{4,2^{*}\},\qquad 2^{*}=\begin{cases}\dfrac{2N}{N-2},&N\geq 3,\\ +\infty,&N\leq 2.\end{cases}

Set

A(u):=\int_{\Omega}|\nabla u|^{2}\,dx,\qquad B(u):=\int_{\Omega}k(x)|u|^{q}\,dx,

and write

F(\lambda,u):=-\bigl(a+\lambda A(u)\bigr)\Delta u-k(x)|u|^{q-2}u.

Then (21) is equivalent to

F(\lambda,u)=0,\qquad u\in H_{0}^{1}(\Omega).

Testing (21) by $u\neq 0$ gives

\bigl(a+\lambda A(u)\bigr)A(u)=B(u),

and hence every nontrivial solution satisfies the Nehari Rayleigh representation

(22)

\boxed{\lambda=\mathcal{R}_{\mathcal{N}}(u):=\frac{B(u)-aA(u)}{A(u)^{2}}.}

We now show that, along distinguished amplitude directions, this scalar parameter representation becomes the exact parameter law of a genuine solution branch.

3.1. Exact amplitude branches and turning bifurcations

Consider the amplitude action

\mathcal{G}_{s}w=sw,\qquad s>0,

and define the orbital Rayleigh profile

\lambda_{w}(s):=\mathcal{R}_{\mathcal{N}}(sw).

Since

A(sw)=s^{2}A(w),\qquad B(sw)=s^{q}B(w),

we obtain

(23)

\lambda_{w}(s)=\frac{B(w)}{A(w)^{2}}s^{q-4}-\frac{a}{A(w)}s^{-2}.

Set

(24)

\tau_{w}:=\frac{B(w)}{A(w)}.

Since $k>0$ , we have $\tau_{w}>0$ for every $w\neq 0$ . Moreover,

a+\lambda_{w}(s)A(sw)=\tau_{w}s^{q-2}.

Substituting

u=sw,\qquad\lambda=\lambda_{w}(s)

into (21), we find that

F(\lambda_{w}(s),sw)=0\qquad\text{for all }s>0

if and only if

(25)

\boxed{\begin{cases}-\tau_{w}\Delta w=k(x)|w|^{q-2}w,&x\in\Omega,\\[2.84526pt] w=0,&x\in\partial\Omega.\end{cases}}

Thus (25) is precisely the exactness condition selecting those amplitude directions along which the Nehari Rayleigh representation reconstructs an exact solution branch.

Writing

w=\tau_{w}^{1/(q-2)}\phi,

the exactness condition reduces to the normalized profile equation

(26)

\boxed{\begin{cases}-\Delta\phi=k(x)|\phi|^{q-2}\phi,&x\in\Omega,\\[2.84526pt] \phi=0,&x\in\partial\Omega.\end{cases}}

Let $\phi\neq 0$ solve (26). Testing by $\phi$ gives

(27)

A(\phi)=B(\phi)=:K_{\phi}.

Hence the amplitude orbit

\mathcal{G}_{s}\phi=s\phi,\qquad s>0,

generates the exact solution branch

(28)

\Gamma_{\phi}=\left\{(\lambda_{\phi}(s),s\phi):s>0\right\},

where

(29)

\boxed{\lambda_{\phi}(s)=\mathcal{R}_{\mathcal{N}}(s\phi)=\frac{s^{q-2}-a}{s^{2}K_{\phi}}.}

In particular,

F\bigl(\lambda_{\phi}(s),s\phi\bigr)=0,\qquad s>0.

The turning geometry is completely explicit. The equation

\lambda_{\phi}^{\prime}(s)=0

has the unique solution

(30)

\boxed{s_{*}^{q-2}=\frac{2a}{4-q}.}

Thus the critical amplitude $s_{*}$ is independent not only of the normalized profile $\phi$ , but also of the weight $k(x)$ . The spatial heterogeneity affects the turning value only through $K_{\phi}$ . Indeed,

(31)

\boxed{\lambda_{\phi}^{*}:=\lambda_{\phi}(s_{*})=\frac{q-2}{2K_{\phi}}s_{*}^{q-4}.}

Moreover,

(32)

\lambda_{\phi}^{\prime\prime}(s_{*})=-\frac{(q-2)(4-q)}{K_{\phi}}s_{*}^{q-6}<0,

so $s_{*}$ is a nondegenerate global maximum of the exact Rayleigh profile.

The remaining branch geometry follows directly from (29):

\lambda_{\phi}(s)\to-\infty\quad\text{as }s\downarrow 0,\qquad\lambda_{\phi}(s)\to 0^{+}\quad\text{as }s\to\infty.

The branch crosses $\lambda=0$ at the profile-independent amplitude

(33)

s_{\rm z}:=a^{1/(q-2)},

and

s_{\rm z}<s_{*}.

Consequently, $\Gamma_{\phi}$ contains exactly one solution for $\lambda\leq 0$ , exactly two solutions for $0<\lambda<\lambda_{\phi}^{*}$ , one turning solution for $\lambda=\lambda_{\phi}^{*}$ , and no solution for $\lambda>\lambda_{\phi}^{*}$ .

By Theorem 2.1,

\boxed{(\lambda_{\phi}^{*},s_{*}\phi)}

is a nondegenerate turning bifurcation point of the exact amplitude branch.

Theorem 3.1 (Exact amplitude bifurcation branch).

Let $\phi\neq 0$ solve the weighted normalized profile problem (26). Then

\Gamma_{\phi}=\left\{(\lambda_{\phi}(s),s\phi):s>0\right\}

is an exact solution branch of (21). It has a unique nondegenerate turning bifurcation point

\boxed{(\lambda_{\phi}^{*},s_{*}\phi),\qquad s_{*}=\left(\frac{2a}{4-q}\right)^{\frac{1}{q-2}},}

where

\lambda_{\phi}^{*}=\frac{q-2}{2K_{\phi}}s_{*}^{q-4},\qquad K_{\phi}=\int_{\Omega}|\nabla\phi|^{2}\,dx=\int_{\Omega}k(x)|\phi|^{q}\,dx.

The branch contains exactly one solution for $\lambda\leq 0$ , exactly two solutions for $0<\lambda<\lambda_{\phi}^{*}$ , one turning solution for $\lambda=\lambda_{\phi}^{*}$ , and no solution for $\lambda>\lambda_{\phi}^{*}$ .

In particular, the critical amplitude $s_{*}$ is universal: it is independent of both the normalized profile and the spatial weight $k(x)$ .

3.2. The maximal Rayleigh branch

We next identify which normalized profile generates the largest turning value. Let

\mathcal{S}=H_{0}^{1}(\Omega)\setminus\{0\}.

The upper Rayleigh threshold of the amplitude class is

(34)

\lambda_{\mathcal{G},\mathcal{S}}^{**}:=\sup_{w\in H_{0}^{1}(\Omega)\setminus\{0\}}\sup_{s>0}\mathcal{R}_{\mathcal{N}}(sw).

For fixed $w\neq 0$ , the orbital profile (23) has the unique critical point

(35)

s_{w}^{q-2}=\frac{2aA(w)}{(4-q)B(w)}.

Since

\lambda_{w}(s)\to-\infty\quad\text{as }s\downarrow 0,\qquad\lambda_{w}(s)\to 0^{+}\quad\text{as }s\to\infty,

this point is its unique global maximum. Substitution into (23) gives

(36)

\boxed{\max_{s>0}\lambda_{w}(s)=C_{a,q}\left(\frac{B(w)^{2/q}}{A(w)}\right)^{\frac{q}{q-2}},}

where

(37)

C_{a,q}:=\frac{q-2}{2}\left(\frac{2a}{4-q}\right)^{\frac{q-4}{q-2}}.

Introduce the optimal weighted Sobolev quotient

(38)

S_{q,k}:=\inf_{w\in H_{0}^{1}(\Omega)\setminus\{0\}}\frac{A(w)}{B(w)^{2/q}}.

Then (36) yields

(39)

\boxed{\lambda_{\mathcal{G},\mathcal{S}}^{**}=C_{a,q}S_{q,k}^{-\frac{q}{q-2}}.}

Since $q<2^{*}$ and $\Omega$ is bounded, the compact embedding

H_{0}^{1}(\Omega)\hookrightarrow L^{q}(\Omega)

and the boundedness of $k$ imply that the infimum in (38) is attained. Let $w_{0}$ be a minimizer normalized by

B(w_{0})=1.

Then

A(w_{0})=S_{q,k},

and its Euler–Lagrange equation is

(40)

-\Delta w_{0}=S_{q,k}\,k(x)|w_{0}|^{q-2}w_{0}.

Define

(41)

\phi_{0}:=S_{q,k}^{1/(q-2)}w_{0}.

Then $\phi_{0}$ solves the weighted normalized profile equation (26). Moreover,

(42)

K_{\phi_{0}}=A(\phi_{0})=B(\phi_{0})=S_{q,k}^{\frac{q}{q-2}}.

Consequently,

\lambda_{\phi_{0}}^{*}=\frac{q-2}{2}s_{*}^{q-4}S_{q,k}^{-\frac{q}{q-2}}=C_{a,q}S_{q,k}^{-\frac{q}{q-2}},

and comparison with (39) gives

(43)

\boxed{\lambda_{\phi_{0}}^{*}=\lambda_{\mathcal{G},\mathcal{S}}^{**}.}

Moreover, $\phi_{0}$ is a least-energy nontrivial solution of the weighted normalized profile equation. Indeed, if $\phi\neq 0$ solves (26), then

A(\phi)=B(\phi),

and hence, by the definition of $S_{q,k}$ ,

S_{q,k}\leq\frac{A(\phi)}{B(\phi)^{2/q}}=A(\phi)^{\frac{q-2}{q}}.

Therefore

A(\phi)\geq S_{q,k}^{\frac{q}{q-2}}=A(\phi_{0}).

Since

J_{\Omega,k}(\phi)=\frac{1}{2}A(\phi)-\frac{1}{q}B(\phi)=\frac{q-2}{2q}A(\phi)

on every normalized profile, $\phi_{0}$ has the least nontrivial critical energy.

Thus the weighted ground-state profile $\phi_{0}$ generates the maximal exact amplitude branch, and

(44)

\boxed{\left(\lambda_{\mathcal{G},\mathcal{S}}^{**},s_{*}\phi_{0}\right)}

is its nondegenerate turning bifurcation point.

Moreover, this threshold is global. Indeed, every nontrivial solution of (21) satisfies

\lambda=\mathcal{R}_{\mathcal{N}}(u)\leq\lambda_{\mathcal{G},\mathcal{S}}^{**}.

Hence no nontrivial solution exists for

\lambda>\lambda_{\mathcal{G},\mathcal{S}}^{**}.

Corollary 3.2 (Maximal amplitude bifurcation branch).

The weighted ground-state profile $\phi_{0}$ generates the maximal exact amplitude branch, and its turning value satisfies

\boxed{\lambda_{\phi_{0}}^{*}=\lambda_{\mathcal{G},\mathcal{S}}^{**}=C_{a,q}S_{q,k}^{-\frac{q}{q-2}}.}

In particular,

\boxed{\left(\lambda_{\mathcal{G},\mathcal{S}}^{**},s_{*}\phi_{0}\right)}

is a nondegenerate turning bifurcation point at the global upper existence threshold for nontrivial solutions.

3.3. Variational hierarchy of amplitude bifurcations

The exact amplitude reduction is not restricted to the weighted ground-state profile $\phi_{0}$ . Every nontrivial solution of (26) generates its own exact Rayleigh branch. The variational multiplicity of the weighted normalized problem therefore produces a countable hierarchy of amplitude bifurcations for the Kirchhoff equation.

Consider the even functional

J(u)=\frac{1}{2}A(u)-\frac{1}{q}B(u),\qquad u\in H_{0}^{1}(\Omega).

Its nontrivial critical points are precisely the weak solutions of (26). Since $2<q<2^{*}$ , the compact embedding

H_{0}^{1}(\Omega)\hookrightarrow L^{q}(\Omega)

and the boundedness of $k$ imply the Palais–Smale compactness condition for $J$ . Moreover, since $k\in C(\overline{\Omega})$ and $k>0$ on $\overline{\Omega}$ , there exists $k_{0}>0$ such that

k(x)\geq k_{0}\qquad\text{on }\overline{\Omega},

which yields the standard symmetric superlinear geometry. Hence the symmetric critical point theorem gives an unbounded sequence of critical values; see, for example, [29, Theorem 9.38].

As shown above, $\phi_{0}$ is a least-energy nontrivial critical point. We may therefore choose $\phi_{1}=\phi_{0}$ and select the remaining profiles from an unbounded variational sequence. Accordingly, there exists a sequence of pairwise distinct antipodal solution pairs

\{\pm\phi_{j}\}_{j\geq 1}

such that

\phi_{1}=\phi_{0},\qquad c_{j}:=J(\phi_{j}),\qquad 0<c_{1}\leq c_{2}\leq\cdots,\qquad c_{j}\to+\infty.

Testing (26) by $\phi_{j}$ gives

A(\phi_{j})=B(\phi_{j})=:K_{j}.

Therefore

(45)

c_{j}=\frac{q-2}{2q}K_{j},\qquad K_{j}=\frac{2q}{q-2}c_{j}\longrightarrow+\infty.

Each antipodal profile pair $\{\pm\phi_{j}\}$ generates the pair of exact amplitude branches

\Gamma_{j}^{\pm}=\left\{(\lambda_{j}(s),\pm s\phi_{j}):s>0\right\},

where

(46)

\lambda_{j}(s)=\frac{s^{q-2}-a}{s^{2}K_{j}}.

Since both the normalized equation and the Kirchhoff equation are odd in $u$ , the two branches have the same Rayleigh profile $\lambda_{j}(s)$ . In what follows, solutions are counted in antipodal pairs.

All these branch pairs have the same critical amplitude

s_{*}^{q-2}=\frac{2a}{4-q},

independent of both the profile and the weight $k(x)$ . Their turning values depend on the profiles only through $K_{j}$ :

(47)

\boxed{\lambda_{j}^{*}=\frac{q-2}{2K_{j}}s_{*}^{q-4}=\frac{(q-2)^{2}}{4q}s_{*}^{q-4}\frac{1}{c_{j}}.}

Since $c_{j}$ is nondecreasing and tends to infinity,

(48)

\boxed{\lambda_{1}^{*}=\lambda_{\mathcal{G},\mathcal{S}}^{**}\geq\lambda_{2}^{*}\geq\cdots\longrightarrow 0.}

Thus the global upper Rayleigh threshold is the first element of a countable hierarchy of turning bifurcation values accumulating at $\lambda=0$ .

Theorem 3.3 (Countable hierarchy of amplitude bifurcations).

Problem (21) possesses infinitely many distinct antipodal pairs of exact amplitude branches

\Gamma_{j}^{\pm},\qquad j\geq 1.

The branches $\Gamma_{j}^{\pm}$ have the antipodal nondegenerate turning bifurcation points

(\lambda_{j}^{*},\pm s_{*}\phi_{j}),

and the corresponding bifurcation values satisfy

\boxed{\lambda_{\mathcal{G},\mathcal{S}}^{**}=\lambda_{1}^{*}\geq\lambda_{2}^{*}\geq\cdots\longrightarrow 0.}

In particular, $\lambda_{1}^{*}$ is the global upper existence threshold for nontrivial solutions.

More precisely:

(i)

for every $\lambda\leq 0$ , problem (21) possesses infinitely many distinct antipodal pairs of nontrivial solutions;
(ii)

for every $m\in\mathbb{N}$ and

$0<\lambda<\lambda_{m}^{*},$

problem (21) possesses at least $2m$ distinct antipodal pairs of nontrivial solutions;
(iii)

for

$\lambda>\lambda_{1}^{*}=\lambda_{\mathcal{G},\mathcal{S}}^{**},$

problem (21) has no nontrivial solutions.

Proof.

We first show that distinct antipodal profile pairs generate distinct antipodal branch pairs. Suppose that, for some $i\neq j$ and $s,r>0$ ,

s\phi_{i}=\pm r\phi_{j}.

Then

\phi_{i}=\pm\alpha\phi_{j},\qquad\alpha:=\frac{r}{s}>0.

Since both profiles solve (26), substitution of $\phi_{i}=\pm\alpha\phi_{j}$ gives

\alpha=\alpha^{q-1}.

Because $q>2$ , we obtain $\alpha=1$ , and hence

\{\pm\phi_{i}\}=\{\pm\phi_{j}\},

a contradiction. Thus the branch pairs $\Gamma_{j}^{\pm}$ are mutually distinct.

Formula (47) and $c_{j}\to+\infty$ imply

\lambda_{j}^{*}\to 0,

while the ordering of $c_{j}$ gives

\lambda_{1}^{*}\geq\lambda_{2}^{*}\geq\cdots.

Since $\phi_{1}=\phi_{0}$ , equality

\lambda_{1}^{*}=\lambda_{\mathcal{G},\mathcal{S}}^{**}

follows from (43).

For every $j$ ,

\lambda_{j}^{\prime\prime}(s_{*})<0,\qquad\left.\frac{d}{ds}\mathcal{G}_{s}\phi_{j}\right|_{s=s_{*}}=\phi_{j}\neq 0.

Hence Theorem 2.1 shows that

(\lambda_{j}^{*},s_{*}\phi_{j})

is a nondegenerate turning bifurcation point of $\Gamma_{j}^{+}$ . By odd symmetry,

(\lambda_{j}^{*},-s_{*}\phi_{j})

is the corresponding turning bifurcation point of $\Gamma_{j}^{-}$ .

Let $\lambda\leq 0$ . By Theorem 3.1, for each $j$ there is exactly one amplitude $s>0$ at the level $\lambda$ . Hence the branch pair $\Gamma_{j}^{\pm}$ yields exactly one antipodal pair

\{\pm s\phi_{j}\}

of nontrivial solutions. Since the branch pairs are mutually distinct, there are infinitely many such solution pairs. This proves (i).

Now let $m\in\mathbb{N}$ and

0<\lambda<\lambda_{m}^{*}.

For every $j\leq m$ ,

\lambda<\lambda_{m}^{*}\leq\lambda_{j}^{*}.

Hence the scalar profile $\lambda_{j}(s)$ has two distinct amplitudes

s_{j}^{-}(\lambda)<s_{*}<s_{j}^{+}(\lambda)

at the level $\lambda$ . These amplitudes generate the two distinct antipodal solution pairs

\{\pm s_{j}^{-}(\lambda)\phi_{j}\},\qquad\{\pm s_{j}^{+}(\lambda)\phi_{j}\}.

Thus each of the first $m$ profile pairs contributes two distinct antipodal pairs of solutions, and (21) has at least $2m$ such pairs. This proves (ii).

Finally, suppose

\lambda>\lambda_{1}^{*}=\lambda_{\mathcal{G},\mathcal{S}}^{**}.

By the definition of the maximal Rayleigh threshold,

\mathcal{R}_{\mathcal{N}}(u)\leq\lambda_{\mathcal{G},\mathcal{S}}^{**}<\lambda\qquad\text{for every }u\neq 0.

Every nontrivial solution of (21), however, must satisfy

\lambda=\mathcal{R}_{\mathcal{N}}(u)=\frac{B(u)-aA(u)}{A(u)^{2}},

which is impossible. This proves (iii). ∎

Thus the unbounded variational hierarchy of the weighted normalized profile problem is transformed by the exact amplitude Rayleigh reduction into a countable hierarchy of explicit turning bifurcation values,

\lambda_{j}^{*}\longrightarrow 0.

The weighted ground-state branch determines the global upper threshold, while higher variational profiles generate successively smaller bifurcation levels. The spatial heterogeneity $k(x)$ changes the profile energies and hence the individual thresholds $\lambda_{j}^{*}$ , but leaves the universal critical amplitude $s_{*}$ unchanged.

Refer to caption — Figure 1. Exact amplitude Rayleigh profiles for representative normalized profiles. All branches have the same zero-crossing amplitude $s_{\rm z}$ and the same critical amplitude $s_{*}$ . Their turning bifurcation values $\lambda_{j}^{*}$ are determined by the profile levels $K_{j}$ of the weighted normalized problem and accumulate at $\lambda=0$ .

4. Exact spatial Rayleigh reduction for generalized Kirchhoff equations

The variational theory of nonlinear scalar-field equations on $\mathbb{R}^{N}$ goes back, in particular, to the pioneering work of Strauss [32]. Under the nearly optimal hypotheses developed by Berestycki and Lions [6, 7], the scalar-field equation

-\Delta\phi=g(\phi)\qquad\text{in }\mathbb{R}^{N}

admits a positive radial ground state and an infinite sequence of distinct radial bound states.

We consider the corresponding Kirchhoff generalization

(49)

-\left(a+b\left(\int_{\mathbb{R}^{N}}|\nabla u|^{2}\,dx\right)^{\theta}\right)\Delta u=g(u)\qquad\text{in }\mathbb{R}^{N},

where

N\geq 3,\qquad a>0,\quad b\geq 0,\quad\theta>0.

The standard Kirchhoff law corresponds to $\theta=1$ .

We assume that $g\in C(\mathbb{R},\mathbb{R})$ is odd and satisfies the Berestycki–Lions hypotheses

(50)		$\displaystyle-\infty$	$\displaystyle<\liminf_{s\to 0^{+}}\frac{g(s)}{s}\leq\limsup_{s\to 0^{+}}\frac{g(s)}{s}=-m<0,$
(51)		$\displaystyle-\infty$	$\displaystyle\leq\limsup_{s\to+\infty}\frac{g(s)}{s^{(N+2)/(N-2)}}\leq 0,$

and that there exists $\zeta>0$ such that

(52)

G(\zeta)>0,\qquad G(t):=\int_{0}^{t}g(\tau)\,d\tau.

The power model is recovered by taking

g(t)=|t|^{p-2}t-\mu t,\qquad\mu>0,\qquad 2<p<2^{*}:=\frac{2N}{N-2}.

Thus the negative linear behavior of $g$ at the origin incorporates the mass term of the usual scalar-field equation.

Set

A(u):=\int_{\mathbb{R}^{N}}|\nabla u|^{2}\,dx,\qquad\mathcal{B}(u):=\int_{\mathbb{R}^{N}}G(u)\,dx,

whenever the latter integral is finite, and define

\mathcal{D}_{G}:=\left\{u\in H^{1}(\mathbb{R}^{N})\setminus\{0\}:\mathcal{B}(u)\in\mathbb{R}\right\}.

The set $\mathcal{D}_{G}$ is invariant under spatial dilations. Throughout this section, by a nontrivial solution we mean a finite-energy solution $u\in\mathcal{D}_{G}$ ; whenever the Pohozaev identity is used, the standard regularity required for its validity is understood.

Then (49) takes the form

-\bigl(a+bA(u)^{\theta}\bigr)\Delta u=g(u).

We write it as

F_{\theta}(b,u)=0,

where

(53)

F_{\theta}(b,u):=-\bigl(a+bA(u)^{\theta}\bigr)\Delta u-g(u).

The associated energy functional on $\mathcal{D}_{G}$ is

(54)

E_{b,\theta}(u)=\frac{a}{2}A(u)+\frac{b}{2(\theta+1)}A(u)^{\theta+1}-\mathcal{B}(u).

The crucial feature of (49) is that spatial dilation acts on the nonlocal coefficient through a fixed homogeneity law, whereas no homogeneity assumption on $g$ is required. This makes it possible to separate the geometry of the exact Kirchhoff branches from the particular form of the local nonlinearity.

4.1. Pohozaev parameter representation

We first derive the scalar parameter relation naturally associated with spatial dilation. For a fixed solution $u$ , the quantity

\kappa_{u}:=a+bA(u)^{\theta}

is a positive constant, and (49) becomes

-\kappa_{u}\Delta u=g(u).

Hence every sufficiently regular nontrivial solution for which the Pohozaev identity is valid satisfies

(55)

\frac{N-2}{2}\bigl(aA(u)+bA(u)^{\theta+1}\bigr)-N\mathcal{B}(u)=0.

Solving for $b$ gives the Pohozaev Rayleigh functional on $\mathcal{D}_{G}$ ,

(56)

\boxed{\mathcal{R}_{\mathcal{P},\theta}(u):=\frac{2N\mathcal{B}(u)-(N-2)aA(u)}{(N-2)A(u)^{\theta+1}}.}

Thus every sufficiently regular nontrivial solution satisfies

(57)

\boxed{b=\mathcal{R}_{\mathcal{P},\theta}(u).}

As in the abstract setting, this identity alone is only a necessary scalar constraint:

b=\mathcal{R}_{\mathcal{P},\theta}(u)\qquad\not\Longrightarrow\qquad F_{\theta}(b,u)=0.

The exact spatial reduction below identifies distinguished dilation orbits on which the Pohozaev Rayleigh functional becomes the actual parameter law of an exact solution branch. Remarkably, the resulting branch geometry depends on $N$ and $\theta$ , but not on the particular form of the Berestycki–Lions nonlinearity $g$ .

4.2. Exact spatial reduction and normalized profiles

We restrict the Pohozaev Rayleigh functional to a spatial dilation orbit

(\mathcal{G}_{s}w)(x)=w(sx),\qquad s>0,

where

w\in\mathcal{D}_{G}.

Under spatial dilation,

A(\mathcal{G}_{s}w)=s^{2-N}A(w),\qquad\mathcal{B}(\mathcal{G}_{s}w)=s^{-N}\mathcal{B}(w).

Hence $\mathcal{G}_{s}w\in\mathcal{D}_{G}$ for every $s>0$ , and

\mathcal{R}_{\mathcal{P},\theta}(\mathcal{G}_{s}w)=\frac{2Ns^{-N}\mathcal{B}(w)-(N-2)as^{2-N}A(w)}{(N-2)\bigl(s^{2-N}A(w)\bigr)^{\theta+1}}.

Introduce

(58)

\sigma_{w}:=\frac{2N\mathcal{B}(w)}{(N-2)A(w)}.

Then the orbital Pohozaev Rayleigh profile becomes

(59)

\boxed{b_{w}(s):=\mathcal{R}_{\mathcal{P},\theta}(\mathcal{G}_{s}w)=\frac{s^{\theta(N-2)-2}}{A(w)^{\theta}}\left(\sigma_{w}-as^{2}\right).}

At this stage (59) enforces only the Pohozaev constraint. To determine when it generates an exact solution branch, set

u=\mathcal{G}_{s}w,\qquad b=b_{w}(s).

Since

A(\mathcal{G}_{s}w)^{\theta}=s^{-\theta(N-2)}A(w)^{\theta},

we obtain

b_{w}(s)A(\mathcal{G}_{s}w)^{\theta}=s^{-2}\left(\sigma_{w}-as^{2}\right),

and therefore

(60)

\boxed{a+b_{w}(s)A(\mathcal{G}_{s}w)^{\theta}=\sigma_{w}s^{-2}.}

On the other hand,

\Delta(\mathcal{G}_{s}w)(x)=s^{2}\Delta w(sx).

Thus the factors $s^{-2}$ and $s^{2}$ cancel exactly, and substitution into (49) shows that

F_{\theta}\bigl(b_{w}(s),\mathcal{G}_{s}w\bigr)=0\qquad\text{for all }s>0

if and only if

(61)

\boxed{-\sigma_{w}\Delta w=g(w)\qquad\text{in }\mathbb{R}^{N}.}

Thus (61) is the exactness condition selecting the spatial directions along which the Pohozaev Rayleigh profile reconstructs a genuine solution branch. Crucially, the orbit parameter $s$ , the Kirchhoff parameters $a$ and $b$ , and the Kirchhoff exponent $\theta$ have all disappeared from the profile equation.

For an exact direction contributing to the original problem $b\geq 0$ , one necessarily has $\sigma_{w}>0$ , since

\sigma_{w}s^{-2}=a+bA(\mathcal{G}_{s}w)^{\theta}>0.

Define

\phi(x)=w(\sqrt{\sigma_{w}}\,x).

Then (61) reduces to the normalized scalar-field equation

(62)

\boxed{-\Delta\phi=g(\phi)\qquad\text{in }\mathbb{R}^{N}.}

Conversely, let $\phi\in\mathcal{D}_{G}$ be a sufficiently regular solution of (62). Its Pohozaev identity is

(63)

\frac{N-2}{2}A(\phi)-N\mathcal{B}(\phi)=0.

Hence

\sigma_{\phi}=\frac{2N\mathcal{B}(\phi)}{(N-2)A(\phi)}=1.

Therefore every such normalized profile generates the extended exact spatial curve

(64)

\boxed{\widetilde{\Gamma}_{\phi}=\left\{\bigl(b_{\phi}(s),\phi(s\,\cdot)\bigr):s>0\right\},}

where

(65)

\boxed{b_{\phi}(s)=\frac{1-as^{2}}{A_{\phi}^{\theta}}s^{\theta(N-2)-2},\qquad A_{\phi}:=A(\phi).}

Thus the exact parameter law has the same form for every admissible Berestycki–Lions nonlinearity. For the original parameter range $b\geq 0$ , the relevant part of this curve is restricted below to $0<s\leq a^{-1/2}$ . The nonlinearity $g$ selects the normalized profiles and their energies $A_{\phi}$ , whereas the dependence of the parameter on the dilation scale is governed entirely by $N$ , $\theta$ , and $a$ .

Remark 4.1 (Universality of the spatial reduction).

The normalized equation (62) is independent of $a$ , $b$ , and $\theta$ . Consequently, for a fixed nonlinearity $g$ , the same family of normalized scalar-field profiles generates exact branches for the whole class of Kirchhoff laws

M_{b}(A)=a+bA^{\theta},\qquad\theta>0.

The nonlinearity enters each branch only through $A_{\phi}$ , while its dependence on the scaling variable is always proportional to

(1-as^{2})s^{\theta(N-2)-2}.

In particular, the dimension–homogeneity bifurcation geometry derived below is independent of the particular choice of $g$ within the admissible class.

Remark 4.2 (Nehari representation).

If testing (49) by $u$ is justified, every nontrivial solution also satisfies

\bigl(a+bA(u)^{\theta}\bigr)A(u)=\int_{\mathbb{R}^{N}}g(u)u\,dx.

Hence one may introduce

\mathcal{R}_{\mathcal{N},\theta}(u):=\frac{\displaystyle\int_{\mathbb{R}^{N}}g(u)u\,dx-aA(u)}{A(u)^{\theta+1}}.

Along every exact spatial branch generated by (62),

\mathcal{R}_{\mathcal{N},\theta}(\mathcal{G}_{s}\phi)=\mathcal{R}_{\mathcal{P},\theta}(\mathcal{G}_{s}\phi)=b_{\phi}(s).

Thus the Nehari representation yields the same parameter law on the exact branch, although it is not needed for the spatial reduction.

4.3. Dimension–homogeneity bifurcation trichotomy

Let $\phi\in\mathcal{D}_{G}$ be a sufficiently regular solution of (62), and set

A_{\phi}:=A(\phi).

By (65), the corresponding exact spatial branch is governed by

(66)

\boxed{b_{\phi}(s)=\frac{1-as^{2}}{A_{\phi}^{\theta}}s^{\theta(N-2)-2},\qquad s>0.}

Remark 4.3 (Generalized Azzollini rescaling relation).

For $u=\phi(s\,\cdot)$ ,

A(u)^{\theta}=s^{-\theta(N-2)}A_{\phi}^{\theta}.

Hence the exact rescaling relation is equivalent to

(67)

\boxed{as^{2}+bA_{\phi}^{\theta}s^{\,2-\theta(N-2)}=1.}

Solving for $b$ recovers (66). For $\theta=1$ , this reduces to

as^{2}+bA_{\phi}s^{4-N}=1,

the usual Azzollini-type rescaling relation [2, 3] for the standard Kirchhoff law.

For the original problem $b\geq 0$ , the relevant part of the exact curve is

\Gamma_{\phi}=\left\{\bigl(b_{\phi}(s),\phi(s\,\cdot)\bigr):0<s\leq s_{\rm z}\right\},\qquad s_{\rm z}:=a^{-1/2},

since

b_{\phi}(s)\geq 0\quad\Longleftrightarrow\quad 0<s\leq s_{\rm z}.

The geometry of every exact spatial branch is governed by the dimension–homogeneity index

(68)

\boxed{k:=\theta(N-2)-2.}

Indeed,

b_{\phi}(s)=\frac{1-as^{2}}{A_{\phi}^{\theta}}s^{k},

and

(69)

b_{\phi}^{\prime}(s)=\frac{s^{k-1}}{A_{\phi}^{\theta}}\left[k-a(k+2)s^{2}\right].

The factor $A_{\phi}^{-\theta}$ changes only the vertical scale of the profile. Its qualitative geometry is therefore determined solely by $N$ and $\theta$ . This dimension–homogeneity balance leads to three distinct regimes, illustrated in Figure 2.

Supercritical scaling regime: $\theta(N-2)>2$ . Here $k>0$ and

b_{\phi}(s)\to 0\quad\text{as }s\downarrow 0,\qquad b_{\phi}(s_{\rm z})=0.

The equation $b_{\phi}^{\prime}(s)=0$ has the unique solution

(70)

\boxed{s_{*}=\left(\frac{\theta(N-2)-2}{a\theta(N-2)}\right)^{1/2},}

with

0<s_{*}<s_{\rm z}.

Moreover,

(71)

b_{\phi}^{\prime\prime}(s_{*})=-\frac{2a\theta(N-2)}{A_{\phi}^{\theta}}s_{*}^{\theta(N-2)-2}<0.

Hence $s_{*}$ is the unique nondegenerate global maximum of the exact Rayleigh profile, with

(72)

\boxed{b_{\phi}^{*}=\frac{2}{\theta(N-2)A_{\phi}^{\theta}}\left(\frac{\theta(N-2)-2}{a\theta(N-2)}\right)^{\frac{\theta(N-2)-2}{2}}.}

The dilation orbit

s\longmapsto\phi(s\,\cdot)

is regular for $\phi\neq 0$ : otherwise $x\cdot\nabla\phi=0$ , which is incompatible with a nonzero $H^{1}(\mathbb{R}^{N})$ profile. Therefore, by Theorem 2.1,

\bigl(b_{\phi}^{*},\phi(s_{*}\cdot)\bigr)

is a nondegenerate turning bifurcation point of the exact spatial branch.

Critical scaling regime: $\theta(N-2)=2$ . Here

b_{\phi}(s)=\frac{1-as^{2}}{A_{\phi}^{\theta}}

is strictly decreasing on $(0,s_{\rm z}]$ , with

\lim_{s\downarrow 0}b_{\phi}(s)=A_{\phi}^{-\theta},\qquad b_{\phi}(s_{\rm z})=0.

Thus $A_{\phi}^{-\theta}$ is a nonattained upper threshold, and there is no interior turning point.

Subcritical scaling regime: $\theta(N-2)<2$ . Here $k<0$ and, since

k+2=\theta(N-2)>0,

we have

k-a(k+2)s^{2}<0,\qquad s>0.

Hence $b_{\phi}$ is strictly decreasing, with

b_{\phi}(s)\to+\infty\quad\text{as }s\downarrow 0,\qquad b_{\phi}(s_{\rm z})=0.

Again there is no interior turning point.

We have therefore obtained the following universal spatial bifurcation trichotomy.

Theorem 4.4 (Dimension–homogeneity trichotomy).

Let $\phi\in\mathcal{D}_{G}$ be a sufficiently regular solution of (62). Then:

(i)

if $\theta(N-2)>2$ , the exact spatial branch $\Gamma_{\phi}$ has a unique nondegenerate turning bifurcation point

$\bigl(b_{\phi}^{*},\phi(s_{*}\cdot)\bigr).$

The branch contains exactly two points for $0<b<b_{\phi}^{*}$ , one point for $b=0$ , one turning solution for $b=b_{\phi}^{*}$ , and none for $b>b_{\phi}^{*}$ ;
(ii)

if $\theta(N-2)=2$ , the branch contains exactly one point for

$0\leq b<A_{\phi}^{-\theta},$

and none for $b\geq A_{\phi}^{-\theta}$ ;
(iii)

if $\theta(N-2)<2$ , the branch contains exactly one point for every $b\geq 0$ .

In particular,

\boxed{\theta(N-2)>2}

is the sharp dimension–homogeneity condition for the occurrence of an interior turning bifurcation on an exact spatial branch. This criterion is independent of the particular Berestycki–Lions nonlinearity $g$ .

4.4. Ground states and the global threshold

Let

\mathcal{P}:=\left\{\phi\in\mathcal{D}_{G}:-\Delta\phi=g(\phi)\ \text{in }\mathbb{R}^{N}\right\},

and

\mathcal{P}_{\rm rad}:=\mathcal{P}\cap H^{1}_{\rm rad}(\mathbb{R}^{N}).

Define the scalar-field action

J(\phi):=\frac{1}{2}A(\phi)-\mathcal{B}(\phi).

Under the Berestycki–Lions assumptions, the normalized problem admits a positive radial ground state; see [6]. Let $U\in\mathcal{P}_{\rm rad}$ be any such ground state and set

c_{g}:=J(U)=\inf_{\phi\in\mathcal{P}}J(\phi),\qquad A_{g}:=A(U).

For every $\phi\in\mathcal{P}$ , the Pohozaev identity

\frac{N-2}{2}A(\phi)-N\mathcal{B}(\phi)=0

gives

(73)

\boxed{J(\phi)=\frac{1}{N}A(\phi).}

Consequently,

(74)

\boxed{A_{g}=\min_{\phi\in\mathcal{P}}A(\phi)=Nc_{g}.}

In particular, all ground states have the same gradient energy $A_{g}$ .

By the exact spatial reduction, every $\phi\in\mathcal{P}$ generates the branch

\Gamma_{\phi}=\left\{\bigl(b_{\phi}(s),\phi(s\,\cdot)\bigr):0<s\leq s_{\rm z}\right\},

where

b_{\phi}(s)=\frac{1-as^{2}}{A(\phi)^{\theta}}s^{\theta(N-2)-2}.

Thus every ground state generates the same scalar parameter profile

(75)

\boxed{b_{g}(s)=\frac{1-as^{2}}{A_{g}^{\theta}}s^{\theta(N-2)-2}.}

The corresponding solution branches themselves need not coincide when the normalized ground state is nonunique.

Conversely, the exact spatial class exhausts all nontrivial finite-energy solutions of the Kirchhoff equation. Indeed, let $u\neq 0$ solve (49) and set

\kappa:=a+bA(u)^{\theta}.

Then

\phi(x):=u(\sqrt{\kappa}\,x)

satisfies

-\Delta\phi=g(\phi).

Since $u\in\mathcal{D}_{G}$ and $\mathcal{D}_{G}$ is invariant under spatial dilations, we also have $\phi\in\mathcal{D}_{G}$ , and hence $\phi\in\mathcal{P}$ . Moreover,

u(x)=\phi(sx),\qquad s=\kappa^{-1/2}.

Since $b\geq 0$ ,

\kappa\geq a,\qquad 0<s\leq s_{\rm z}=a^{-1/2}.

Thus every nontrivial finite-energy solution belongs to one of the exact branches $\Gamma_{\phi}$ .

Assume first that

\theta(N-2)>2.

By (72),

b_{\phi}^{*}=\frac{C_{a,N,\theta}}{A(\phi)^{\theta}},

where

(76)

C_{a,N,\theta}:=\frac{2}{\theta(N-2)}\left(\frac{\theta(N-2)-2}{a\theta(N-2)}\right)^{\frac{\theta(N-2)-2}{2}}.

Since the ground states minimize $A(\phi)$ on $\mathcal{P}$ , they maximize the turning value. Hence

(77)

\boxed{b_{g}^{*}:=\sup_{\phi\in\mathcal{P}}b_{\phi}^{*}=\frac{C_{a,N,\theta}}{A_{g}^{\theta}}.}

Equivalently,

(78)

\boxed{b_{g}^{*}=\frac{C_{a,N,\theta}}{(Nc_{g})^{\theta}}.}

Thus every ground-state profile generates the maximal turning parameter. Moreover, if $u\neq 0$ is any solution, then

u=\phi(s\,\cdot)

for some $\phi\in\mathcal{P}$ , and therefore

b=b_{\phi}(s)\leq b_{\phi}^{*}\leq b_{g}^{*}.

Consequently, $b_{g}^{*}$ is the global upper existence threshold:

(79)

\boxed{b>b_{g}^{*}\quad\Longrightarrow\quad\text{\eqref{eq:general-K} has no nontrivial solution}.}

The same ground-state selection determines the global threshold in the critical regime. If

\theta(N-2)=2,

then the branch generated by $\phi\in\mathcal{P}$ exists precisely for

0\leq b<A(\phi)^{-\theta}.

Since $A(\phi)\geq A_{g}$ , the value

(80)

\boxed{b_{g}^{\,0}:=A_{g}^{-\theta}=(Nc_{g})^{-\theta}}

is the sharp nonattained global upper existence threshold. Every ground-state branch realizes

0\leq b<b_{g}^{\,0},

whereas no nontrivial solution exists for

b\geq b_{g}^{\,0}.

\theta(N-2)<2,

there is no finite upper parameter threshold: every ground-state branch contains exactly one solution for each $b\geq 0$ .

We summarize these observations as follows.

Corollary 4.5 (Ground-state maximality).

Let $U$ be any ground state of (62), and let $A_{g}=A(U)=Nc_{g}$ .

(i)

If $\theta(N-2)>2$ , every ground-state branch has the same nondegenerate turning value

$b_{g}^{*}=\frac{C_{a,N,\theta}}{A_{g}^{\theta}}=\sup_{\phi\in\mathcal{P}}b_{\phi}^{*},$

and $b_{g}^{*}$ is the global upper existence threshold.
(ii)

If $\theta(N-2)=2$ , the value

$b_{g}^{\,0}=A_{g}^{-\theta}$

is the sharp nonattained global upper existence threshold.
(iii)

If $\theta(N-2)<2$ , the Kirchhoff problem has no finite upper threshold, and every ground-state branch contains exactly one solution for every $b\geq 0$ .

For the standard Kirchhoff law $\theta=1$ , the governing exponent becomes

k=N-4.

Hence the classical distinction between $N=3$ , $N=4$ , and $N\geq 5$ is precisely the specialization of the dimension–homogeneity balance.

Corollary 4.6 (Standard Kirchhoff law).

For $\theta=1$ :

(i)

if $N\geq 5$ , every exact spatial branch has a unique nondegenerate turning bifurcation point, with

s_{*}=\left(\frac{N-4}{a(N-2)}\right)^{1/2},\qquad b_{\phi}^{*}=\frac{2}{A_{\phi}(N-2)}\left(\frac{N-4}{a(N-2)}\right)^{\frac{N-4}{2}};

(ii)

if $N=4$ , every branch is strictly decreasing from the nonattained value $A_{\phi}^{-1}$ to $0$ ;
(iii)

if $N=3$ , every branch is strictly decreasing from $+\infty$ to $0$ .

Remark 4.7 (Positive radial solutions and the power case).

For a general Berestycki–Lions nonlinearity, uniqueness of a positive radial normalized profile is not assumed. Hence the positive radial solution set of the Kirchhoff problem is, in general, the union of the exact branches generated by all positive profiles in $\mathcal{P}_{\rm rad}$ .

For the power nonlinearity

g(t)=|t|^{p-2}t-\mu t,\qquad 2<p<2^{*},

Kwong’s theorem [24] gives uniqueness of the positive radial normalized profile $U_{\mu}$ . In this case the positive radial solution set reduces to a single exact spatial branch.

4.5. Variational hierarchy of turning thresholds

Continue to assume

\theta(N-2)>2.

Under the Berestycki–Lions assumptions, the normalized scalar-field equation possesses infinitely many distinct radial solution pairs; see [7]. Let $\phi_{1}=U$ be a ground state and choose the remaining profiles from a radial variational sequence so that

\{\pm\phi_{j}\}_{j\geq 1}\subset\mathcal{P}_{\rm rad},\qquad J(\phi_{j})\longrightarrow+\infty.

Since

J(\phi)=\frac{1}{N}A(\phi)\qquad\text{on }\mathcal{P},

after passing to a subsequence and relabeling we may assume

A_{j}:=A(\phi_{j}),\qquad A_{1}=A_{g}\leq A_{2}\leq\cdots,\qquad A_{j}\longrightarrow+\infty.

Each profile generates the exact branch

\Gamma_{j}=\left\{\bigl(b_{j}(s),\phi_{j}(s\,\cdot)\bigr):0<s\leq s_{\rm z}\right\},

where

(81)

b_{j}(s)=\frac{1-as^{2}}{A_{j}^{\theta}}s^{\theta(N-2)-2}.

All branches have the same critical spatial scale $s_{*}$ , while their turning values are

(82)

\boxed{b_{j}^{*}=\frac{2}{\theta(N-2)A_{j}^{\theta}}\left(\frac{\theta(N-2)-2}{a\theta(N-2)}\right)^{\frac{\theta(N-2)-2}{2}}.}

Hence

(83)

\boxed{b_{1}^{*}=b_{g}^{*}\geq b_{2}^{*}\geq\cdots\longrightarrow 0.}

The branches are mutually distinct. Indeed, suppose that

\phi_{i}(s\,\cdot)=\pm\phi_{j}(r\,\cdot)=:u.

Since $g$ is odd and both $\phi_{i}$ and $\phi_{j}$ solve the normalized equation, $u$ satisfies

-s^{-2}\Delta u=g(u),\qquad-r^{-2}\Delta u=g(u).

Subtracting gives

(s^{-2}-r^{-2})\Delta u=0.

If $s\neq r$ , then $\Delta u=0$ . Since $u\in H^{1}(\mathbb{R}^{N})$ , this implies $u=0$ , a contradiction. Hence $s=r$ , and consequently

\phi_{i}=\pm\phi_{j}.

Let

\mathcal{S}_{\rm ex}:=\left\{\mathcal{G}_{s}\phi:\phi\in\mathcal{P},\ 0<s\leq s_{\rm z}\right\}.

Define the upper Rayleigh threshold of the exact spatial class by

(84)

b_{\mathcal{G},\mathcal{S}_{\rm ex}}^{**}:=\sup_{\phi\in\mathcal{P}}\sup_{0<s\leq s_{\rm z}}\mathcal{R}_{\mathcal{P},\theta}(\mathcal{G}_{s}\phi).

Since

\mathcal{R}_{\mathcal{P},\theta}(\mathcal{G}_{s}\phi)=b_{\phi}(s),

and the ground states minimize $A(\phi)$ on $\mathcal{P}$ , we obtain

(85)

\boxed{b_{\mathcal{G},\mathcal{S}_{\rm ex}}^{**}=b_{1}^{*}=b_{g}^{*}.}

By the exhaustivity established in Subsection 4.4, this is also the global upper existence threshold for (49).

Thus the variational hierarchy of the normalized scalar-field equation is transformed by the exact Rayleigh reduction into an inverse hierarchy of Kirchhoff turning thresholds.

Theorem 4.8 (Spatial bifurcation hierarchy).

Assume $\theta(N-2)>2$ . Then (49) possesses infinitely many distinct exact radial branches $\Gamma_{j}$ , each with a unique nondegenerate turning bifurcation point

\bigl(b_{j}^{*},\phi_{j}(s_{*}\cdot)\bigr).

The corresponding turning values satisfy

\boxed{b_{\mathcal{G},\mathcal{S}_{\rm ex}}^{**}=b_{1}^{*}=b_{g}^{*}\geq b_{2}^{*}\geq\cdots\longrightarrow 0.}

More precisely:

(i)

for $b=0$ , problem (49) possesses infinitely many distinct antipodal pairs of nontrivial radial solutions;
(ii)

for every $m\in\mathbb{N}$ and

$0<b<b_{m}^{*},$

it possesses at least $2m$ distinct antipodal pairs of nontrivial radial solutions;
(iii)

for

$b>b_{g}^{*}=b_{\mathcal{G},\mathcal{S}_{\rm ex}}^{**},$

problem (49) has no nontrivial solutions.

Proof.

The distinctness of the branches was proved above, and (82)–(83) give

b_{1}^{*}=b_{g}^{*}\geq b_{2}^{*}\geq\cdots\longrightarrow 0.

For every $j$ , the profile

b_{j}(s)=\frac{1-as^{2}}{A_{j}^{\theta}}s^{\theta(N-2)-2}

has the same unique nondegenerate maximum at $s=s_{*}$ . The dilation orbit

s\longmapsto\phi_{j}(s\,\cdot)

is regular for $\phi_{j}\neq 0$ ; otherwise $x\cdot\nabla\phi_{j}=0$ , which is incompatible with a nonzero $H^{1}(\mathbb{R}^{N})$ profile. Hence Theorem 2.1 yields the nondegenerate turning bifurcation point

\bigl(b_{j}^{*},\phi_{j}(s_{*}\cdot)\bigr).

For $b=0$ , each profile pair gives the solution pair

\{\pm\phi_{j}(s_{\rm z}\,\cdot)\},

and the distinctness of the branches proves (i).

Now let $m\in\mathbb{N}$ and

0<b<b_{m}^{*}.

For every $j\leq m$ ,

b<b_{m}^{*}\leq b_{j}^{*}.

Hence

b_{j}(s)=b

has two distinct roots

s_{j}^{-}(b)<s_{*}<s_{j}^{+}(b).

Thus each of the first $m$ profile pairs produces two distinct antipodal solution pairs

\{\pm\phi_{j}(s_{j}^{-}(b)\,\cdot)\},\qquad\{\pm\phi_{j}(s_{j}^{+}(b)\,\cdot)\}.

Therefore (49) has at least $2m$ distinct antipodal pairs, proving (ii).

Finally, (85) and the exhaustivity of the exact spatial class imply

b\leq b_{g}^{*}

for every nontrivial solution. This proves (iii). ∎

4.6. Radial simple-fold classification

The exact Rayleigh reduction detects turning bifurcation points without requiring any spectral analysis of the linearization. Under additional regularity and nondegeneracy assumptions on a normalized radial profile, the corresponding turning point can also be classified as a radial simple fold.

Corollary 4.9 (Radial simple fold).

Assume

\theta(N-2)>2,

and let $\phi\in\mathcal{P}_{\rm rad}$ . Set

u_{*}:=\phi(s_{*}\cdot).

Suppose, in addition, that $g\in C^{1}(\mathbb{R})$ , that the Nemytskii operator induced by $g$ yields a $C^{1}$ realization

F_{\theta}:\mathbb{R}\times H^{1}_{\rm rad}(\mathbb{R}^{N})\longrightarrow H^{-1}_{\rm rad}(\mathbb{R}^{N})

in a neighborhood of $(b_{\phi}^{*},u_{*})$ , and that

s\longmapsto\phi(s\,\cdot)

is a $C^{1}$ regular curve in $H^{1}_{\rm rad}(\mathbb{R}^{N})$ near $s_{*}$ . Assume also that the scalar-field linearization

\mathcal{L}_{\phi}:=-\Delta-g^{\prime}(\phi):H^{1}_{\rm rad}(\mathbb{R}^{N})\longrightarrow H^{-1}_{\rm rad}(\mathbb{R}^{N})

is Fredholm of index zero and radially nondegenerate:

\ker\mathcal{L}_{\phi}=\{0\}.

Then $(b_{\phi}^{*},u_{*})$ is a radial simple fold of (49). More precisely, in a neighborhood of $(b_{\phi}^{*},u_{*})$ , the radial solution set is a one-dimensional $C^{1}$ manifold locally parametrized by the exact spatial branch

s\longmapsto\bigl(b_{\phi}(s),\phi(s\,\cdot)\bigr),

and its projection onto the parameter axis has a nondegenerate quadratic maximum at $s=s_{*}$ . Moreover,

\boxed{\ker\left(F_{\theta,u}(b_{\phi}^{*},u_{*})\big|_{H^{1}_{\rm rad}(\mathbb{R}^{N})}\right)=\operatorname{span}\left\{\left.\frac{d}{ds}\phi(s\,\cdot)\right|_{s=s_{*}}\right\}.}

Proof.

Set

u_{s}:=\phi(s\,\cdot),\qquad\dot{u}_{*}:=\left.\frac{d}{ds}u_{s}\right|_{s=s_{*}}.

By the regularity assumption on the dilation orbit,

\dot{u}_{*}\neq 0.

Differentiating

F_{\theta}\bigl(b_{\phi}(s),u_{s}\bigr)=0

at $s=s_{*}$ and using $b_{\phi}^{\prime}(s_{*})=0$ gives

(86)

F_{\theta,u}(b_{\phi}^{*},u_{*})\dot{u}_{*}=0.

For radial $v$ ,

	$\displaystyle F_{\theta,u}(b,u)v={}$	$\displaystyle-\bigl(a+bA(u)^{\theta}\bigr)\Delta v-g^{\prime}(u)v$
		$\displaystyle-2b\theta A(u)^{\theta-1}\left(\int_{\mathbb{R}^{N}}\nabla u\cdot\nabla v\,dx\right)\Delta u.$

Let

L_{*}:=F_{\theta,u}(b_{\phi}^{*},u_{*}):H^{1}_{\rm rad}(\mathbb{R}^{N})\longrightarrow H^{-1}_{\rm rad}(\mathbb{R}^{N}),

and write

L_{*}=B_{*}+K_{*},

where

B_{*}v=-\bigl(a+b_{\phi}^{*}A(u_{*})^{\theta}\bigr)\Delta v-g^{\prime}(u_{*})v

and $K_{*}$ is the rank-one nonlocal term.

By the exact rescaling relation,

a+b_{\phi}^{*}A(u_{*})^{\theta}=s_{*}^{-2}.

Under the dilation $y=s_{*}x$ , the operator $B_{*}$ is conjugate to

\mathcal{L}_{\phi}=-\Delta-g^{\prime}(\phi).

The Fredholm index-zero property together with

\ker\mathcal{L}_{\phi}=\{0\}

implies that $\mathcal{L}_{\phi}$ , and hence $B_{*}$ , is invertible. Since $K_{*}$ has rank one, $L_{*}$ is Fredholm of index zero and

\dim\ker L_{*}\leq 1.

Together with (86) and $\dot{u}_{*}\neq 0$ , this yields

\ker L_{*}=\operatorname{span}\{\dot{u}_{*}\}.

It remains to verify parameter transversality. Since

F_{\theta,b}(b,u)=-A(u)^{\theta}\Delta u,

we have

	$\displaystyle\left\langle F_{\theta,b}(b_{\phi}^{},u_{}),\dot{u}_{*}\right\rangle$	$\displaystyle=A(u_{})^{\theta}\int_{\mathbb{R}^{N}}\nabla u_{}\cdot\nabla\dot{u}_{*}\,dx$
		$\displaystyle=\frac{A(u_{})^{\theta}}{2}\left.\frac{d}{ds}A(u_{s})\right\|_{s=s_{}}.$

Because

A(u_{s})=s^{2-N}A_{\phi},

it follows that

\left.\frac{d}{ds}A(u_{s})\right|_{s=s_{*}}=(2-N)A_{\phi}s_{*}^{1-N}\neq 0.

Thus

(87)

\left\langle F_{\theta,b}(b_{\phi}^{*},u_{*}),\dot{u}_{*}\right\rangle\neq 0.

The bilinear form associated with $L_{*}$ is symmetric. Since $L_{*}$ is Fredholm of index zero and

\ker L_{*}=\operatorname{span}\{\dot{u}_{*}\},

we have

\operatorname{Ran}L_{*}=\left\{f\in H^{-1}_{\rm rad}(\mathbb{R}^{N}):\langle f,\dot{u}_{*}\rangle=0\right\}.

Hence (87) implies

F_{\theta,b}(b_{\phi}^{*},u_{*})\notin\operatorname{Ran}L_{*}.

Consequently, the full derivative

DF_{\theta}(b_{\phi}^{*},u_{*})[\beta,v]=\beta F_{\theta,b}(b_{\phi}^{*},u_{*})+L_{*}v

is surjective, with

\ker DF_{\theta}(b_{\phi}^{*},u_{*})=\operatorname{span}\{(0,\dot{u}_{*})\}.

By the regular level-set theorem, the radial solution set near $(b_{\phi}^{*},u_{*})$ is therefore a one-dimensional $C^{1}$ manifold.

The exact curve

\gamma(s)=\bigl(b_{\phi}(s),u_{s}\bigr)

lies in this manifold and satisfies

\gamma^{\prime}(s_{*})=(0,\dot{u}_{*})\neq 0.

It therefore gives a local parametrization of the radial solution manifold. Finally,

b_{\phi}^{\prime}(s_{*})=0,\qquad b_{\phi}^{\prime\prime}(s_{*})<0,

so the projection onto the parameter axis has a nondegenerate quadratic maximum at $s=s_{*}$ . Hence $(b_{\phi}^{*},u_{*})$ is a radial simple fold. ∎

Remark 4.10 (Role of the additional assumptions).

The $C^{1}$ regularity and radial nondegeneracy assumptions above are not part of the basic Berestycki–Lions hypotheses and are needed only for the local simple-fold classification. They play no role in the exact Rayleigh detection of turning bifurcations, in the dimension–homogeneity trichotomy, or in the variational hierarchy. Whenever the operator realization is of class $C^{2}$ , the preceding geometric fold agrees with the classical Lyapunov–Schmidt simple-fold criterion discussed in Subsection 2.4.

Remark 4.11 (Translation modes).

In the full space $H^{1}(\mathbb{R}^{N})$ , translation invariance produces the additional kernel directions

\partial_{x_{1}}u_{*},\ldots,\partial_{x_{N}}u_{*}.

Hence the one-dimensional-kernel formulation is naturally made in $H^{1}_{\rm rad}(\mathbb{R}^{N})$ , or after factoring out translations.

5. Discussion

5.1. Beyond amplitude and spatial scaling

The examples above show that amplitude scaling and spatial dilation are particular realizations of a more general mechanism. Its essential ingredients are a one-parameter transformation orbit

u(s)=\mathcal{G}_{s}\phi

and a generalized Rayleigh functional for which

(88)

\lambda_{\phi}(s)=\mathcal{R}(\mathcal{G}_{s}\phi)

becomes the actual parameter law of an exact solution branch:

(89)

F\bigl(\mathcal{R}(\mathcal{G}_{s}\phi),\mathcal{G}_{s}\phi\bigr)=0.

The decisive point is exactness. A critical point of an arbitrary fiber map or generalized Rayleigh functional may reflect only a degeneracy of an auxiliary constraint. Once (89) holds, however, the scalar Rayleigh geometry becomes part of the geometry of the genuine solution set. Nondegenerate critical points of the exact Rayleigh profile then detect turning bifurcations directly through

\frac{d}{ds}\mathcal{R}(\mathcal{G}_{s}\phi)=0.

This changes the role of the Rayleigh functional. Instead of merely assigning a distinguished parameter level to a state, it may reconstruct the parameter evolution along a nonlinear solution branch. Consequently, multiplicity, turning thresholds, and branch hierarchies can become visible through the geometry of a one-dimensional profile.

The weighted bounded-domain problem illustrates an important robustness of this mechanism. The introduction of a positive spatial coefficient $k(x)$ changes the normalized profile equation and hence the profile energies and turning levels, but leaves the exact amplitude geometry unchanged. In particular, the critical amplitude

s_{*}=\left(\frac{2a}{4-q}\right)^{1/(q-2)}

is independent of both the profile and the spatial weight. Thus exact Rayleigh reduction need not rely on complete spatial homogeneity of the underlying equation.

This viewpoint suggests a broader strategy for nonlinear bifurcation problems: search simultaneously for a scalar parameter representation and a transformation family for which the corresponding exactness condition reduces to a tractable normalized profile equation. The transformation need not be a pure amplitude or spatial scaling. Natural candidates include combined transformations

u_{s}(x)=s^{\alpha}u(sx),

anisotropic dilations, and transformations adapted to conservation laws, homogeneity, or other geometric structures of the equation. For example, $\alpha=N/2$ preserves the $L^{2}$ norm and is natural in normalized nonlinear Schrödinger problems and related models; see [4, 5, 33].

The method also allows several scalar constraints to coexist. Their generalized Rayleigh functionals may differ on the ambient state space but recover the same external parameter along a common exact orbit. The coincidence of the Nehari and Pohozaev Rayleigh functionals on the exact spatial Kirchhoff branches provides a concrete example.

It is useful to distinguish this mechanism from an ordinary symmetry transformation. A symmetry maps solutions of a fixed equation to solutions of the same equation, while an equivalence transformation prescribes the change of parameters together with the transformation. In an exact Rayleigh reduction, by contrast, the parameter evolution is reconstructed from the transformed state itself:

\lambda_{\phi}(s)=\mathcal{R}(\mathcal{G}_{s}\phi).

Its geometry is therefore not imposed in advance. In particular, nonmonotonicity of the resulting parameter law may reveal turning bifurcations and parameter thresholds that are hidden in the original operator equation.

The broader problem suggested by this work is therefore to identify nonlinear equations, scalar constraints, and transformation families for which a generalized Rayleigh representation admits such an exact realization.

5.2. Linearization, spectral transitions, and stability

Exact Rayleigh reduction also creates a direct link between branch geometry and the spectrum of the linearized equation. At a critical point of an exact branch,

F(\lambda_{\phi}(s),u(s))=0,\qquad\lambda_{\phi}^{\prime}(s_{*})=0,

differentiation gives

F_{u}(\lambda_{*},u_{*})u^{\prime}(s_{*})=0.

Thus the tangent to the explicitly constructed orbit provides a kernel direction of the state linearization. The loss of invertibility is not used to locate the critical parameter; rather, it emerges as a consequence of the Rayleigh turning geometry.

This suggests a natural relation between critical points of the exact Rayleigh profile and spectral transitions of the linearized operator. In the Kirchhoff case this question is especially interesting because the linearization contains a nonlocal rank-one perturbation. The radial simple-fold result above provides one instance in which the Rayleigh tangent generates the relevant kernel mode and the detected turning bifurcation is classified as a radial simple fold.

A complete stability theory requires further information. Along a spatial branch $u_{s}(x)=U(sx)$ , variational constraints and conserved quantities generally change with $s$ , so stability cannot simply be transferred from the normalized local equation. The relation between turning bifurcations, Morse-index changes, spectral crossings, and orbital stability therefore remains a natural direction for further study; see [8, 34].

Acknowledgements

During the preparation of this manuscript, ChatGPT (OpenAI) was used for language editing and structural suggestions. All mathematical arguments, calculations, references, and conclusions were independently checked and verified by the author, who takes full responsibility for the final manuscript.

Declarations

Funding

No funding was received for conducting this study.

Competing interests

The author has no relevant financial or non-financial interests to disclose.

Data availability

No datasets were generated or analysed during the current study.

Author contributions

Yavdat Il’yasov: Conceptualization, methodology, formal analysis, writing–original draft, writing–review and editing.

References

[1] V. I. Arnol’d, Catastrophe Theory, Springer Science & Business Media, 2003.
[2] A. Azzollini, The elliptic Kirchhoff equation in $\mathbb{R}^{N}$ perturbed by a local nonlinearity, Differential Integral Equations 25 (2012), no. 5–6, 543–554, doi:10.57262/die/1356012678.
[3] A. Azzollini, A note on the elliptic Kirchhoff equation in $\mathbb{R}^{N}$ perturbed by a local nonlinearity, Commun. Contemp. Math. 17 (2015), no. 4, 1450039, doi:10.1142/S0219199714500394.
[4] T. Bartsch and N. Soave, A natural constraint approach to normalized solutions of nonlinear Schrödinger equations and systems, J. Funct. Anal. 272 (2017), no. 12, 4998–5037, doi:10.1016/j.jfa.2017.01.025.
[5] J. Bellazzini and G. Siciliano, Scaling properties of functionals and existence of constrained minimizers, J. Funct. Anal. 261 (2011), no. 9, 2486–2507, doi:10.1016/j.jfa.2011.06.014.
[6] H. Berestycki and P.-L. Lions, Nonlinear scalar field equations. I. Existence of a ground state, Arch. Ration. Mech. Anal. 82 (1983), no. 4, 313–345.
[7] H. Berestycki and P.-L. Lions, Nonlinear scalar field equations. II. Existence of infinitely many solutions, Arch. Ration. Mech. Anal. 82 (1983), no. 4, 347–375.
[8] T. Cazenave and P.-L. Lions, Orbital stability of standing waves for some nonlinear Schrödinger equations, Commun. Math. Phys. 85 (1982), 549–561.
[9] M. G. Crandall and P. H. Rabinowitz, Bifurcation from simple eigenvalues, J. Funct. Anal. 8 (1971), 321–340.
[10] E. J. Doedel, H. B. Keller, and J.-P. Kernevez, Numerical analysis and control of bifurcation problems (I): Bifurcation in finite dimensions, Internat. J. Bifur. Chaos 1 (1991), 493–520.
[11] E. J. Doedel, H. B. Keller, and J.-P. Kernevez, Numerical analysis and control of bifurcation problems (II): Bifurcation in infinite dimensions, Internat. J. Bifur. Chaos 1 (1991), 745–772.
[12] M. Golubitsky and D. G. Schaeffer, Singularities and Groups in Bifurcation Theory, Vol. I, Applied Mathematical Sciences, Vol. 51, Springer-Verlag, New York, 1985.
[13] Y. Huang, Z. Liu, and Y. Wu, On finding solutions of a Kirchhoff type problem, Proc. Amer. Math. Soc. 144 (2016), no. 7, 3019–3033, doi:10.1090/proc/12946.
[14] Ya. Sh. Il’yasov, Non-local investigation of bifurcations of solutions of non-linear elliptic equations, Izv. Math. 66 (2002), no. 6, 1103–1130, doi:10.1070/IM2002v066n06ABEH000408.
[15] Y. Il’yasov, Bifurcation calculus by the extended functional method, Funct. Anal. Appl. 41 (2007), no. 1, 18–30.
[16] Y. Il’yasov and Y. Egorov, Hopf boundary maximum principle violation for semilinear elliptic equations, Nonlinear Anal. 72 (2010), no. 7–8, 3346–3355.
[17] Y. Il’yasov, On extreme values of Nehari manifold method via nonlinear Rayleigh’s quotient, Topol. Methods Nonlinear Anal. 49 (2017), no. 2, 683–714.
[18] Y. Il’yasov, Finding saddle-node bifurcations via a nonlinear generalized Collatz–Wielandt formula, Internat. J. Bifur. Chaos Appl. Sci. Engrg. 31 (2021), no. 1, 2150008, doi:10.1142/S0218127421500085.
[19] Y. Il’yasov, Rayleigh quotients of the level set manifolds related to the nonlinear PDE, Minimax Theory Appl. 7 (2022), no. 2, 277–302.
[20] Y. S. Il’yasov, On the minimax bifurcation formula, arXiv:2605.17331, 2026.
[21] H. Kielhöfer, Bifurcation Theory: An Introduction with Applications to Partial Differential Equations, 2nd ed., Applied Mathematical Sciences, Vol. 156, Springer, New York, 2012.
[22] G. Kirchhoff, Vorlesungen über mathematische Physik. Mechanik, 3rd ed., B. G. Teubner, Leipzig, 1883.
[23] M. A. Krasnosel’skii and P. P. Zabreiko, Geometrical Methods of Nonlinear Analysis, Grundlehren der mathematischen Wissenschaften, Vol. 263, Springer-Verlag, Berlin, 1984.
[24] M. K. Kwong, Uniqueness of positive solutions of $\Delta u-u+u^{p}=0$ in $\mathbb{R}^{N}$ , Arch. Ration. Mech. Anal. 105 (1989), no. 3, 243–266, doi:10.1007/BF00251502.
[25] Z. Liang, F. Li, and J. Shi, Positive solutions of Kirchhoff-type non-local elliptic equation: a bifurcation approach, Proc. Roy. Soc. Edinburgh Sect. A 147 (2017), no. 4, 875–894, doi:10.1017/S0308210516000378.
[26] J. Liu, J.-F. Liao, H.-L. Pan, and C.-L. Tang, A bifurcation-type result for Kirchhoff equations, C. R. Math. Acad. Sci. Paris 360 (2022), 247–254, doi:10.5802/crmath.294.
[27] T. F. Ma, Remarks on an elliptic equation of Kirchhoff type, Nonlinear Anal. 63 (2005), e1967–e1977.
[28] P. H. Rabinowitz, Some global results for nonlinear eigenvalue problems, J. Funct. Anal. 7 (1971), 487–513.
[29] P. H. Rabinowitz, Minimax Methods in Critical Point Theory with Applications to Differential Equations, CBMS Regional Conference Series in Mathematics, Vol. 65, American Mathematical Society, Providence, RI, 1986.
[30] R. Seydel, Practical Bifurcation and Stability Analysis, Interdisciplinary Applied Mathematics, Vol. 5, Springer, 2010.
[31] K. Silva, The bifurcation diagram of an elliptic Kirchhoff-type equation with respect to the stiffness of the material, Z. Angew. Math. Phys. 70 (2019), Art. 93, doi:10.1007/s00033-019-1137-8.
[32] W. A. Strauss, Existence of solitary waves in higher dimensions, Comm. Math. Phys. 55 (1977), 149–162.
[33] H. Ye, The existence of normalized solutions for $L^{2}$ -critical constrained problems related to Kirchhoff equations, Z. Angew. Math. Phys. 66 (2015), no. 4, 1483–1497, doi:10.1007/s00033-014-0474-x.
[34] J. Zhang, Z. Liu, and M. Squassina, Modulational stability of ground states to nonlinear Kirchhoff equations, J. Math. Anal. Appl. 477 (2019), no. 1, 844–859, doi:10.1016/j.jmaa.2019.04.067.