Nonlinear Differ. Equ. Appl.

(2024) 31:41
c 2024 The Author(s)
1021-9722/24/030001-28
published online March 20, 2024 Nonlinear Differential Equations
https://doi.org/10.1007/s00030-024-00925-8 and Applications NoDEA

Boundary value problems associated with

Hamiltonian systems coupled with
positively-(p,q)-homogeneous systems
Alessandro Fonda and Wahid Ullah

Abstract. We study the multiplicity of solutions for a two-point boundary

value problem of Neumann type associated with a Hamiltonian system
which couples a system with periodic Hamiltonian in the space variable
with a second one with positively-(p,q)-homogeneous Hamiltonian. The
periodic problem is also treated.
Mathematics Subject Classification. 34C25.
Keywords. Hamiltonian systems, Periodic solutions, Multiplicity, Poincaré–
Birkhoff theorem, Positively-(p,q)-homogeneous systems.

1. Introduction and statement of the main result

In the recent paper [7], a multiplicity result for a Neumann-type boundary
value problem associated with a Hamiltonian system has been proved. It is the
aim of this paper to extend this result to coupled systems, the first of which is
of the type considered in [7], while the second one involves a positively-(p, q)-
homogeneous and positive Hamiltonian function.
Denoting by J the standard symplectic matrix, our Hamiltonian system

J ż = ∇z H(t, z),

when writing z = ((x, y), (u, v)) ∈ R2M × R2L , is driven by a Hamiltonian
function of the type

H(t, z) = H(t, x, y) + H (u, v) + P (t, x, y, u, v).

41 Page 2 of 28 A. Fonda and W. Ullah NoDEA

To be more precise, we are dealing with the Hamiltonian system

⎧
⎪
⎪ ẋ = ∇y H(t, x, y) + ∇y P (t, x, y, u, v) ,
⎪
⎨ẏ = −∇ H(t, x, y) − ∇ P (t, x, y, u, v),
x x
(1.1)
⎪
⎪ u̇ = ∇v H (u, v) + ∇v P (t, x, y, u, v),
⎪
⎩
v̇ = −∇u H (u, v) − ∇u P (t, x, y, u, v),
with Neumann-type boundary conditions

y(a) = 0 = y(b),
(1.2)
v(a) = 0 = v(b).
We write
x = (x1 , . . . , xM ) ∈ RM , y = (y1 , . . . , yM ) ∈ RM ,
u = (u1 , . . . , uL ) ∈ RL , v = (v1 , . . . , vL ) ∈ RL .
The functions H : [a, b]×R2M → R, H : R2L → R and P : [a, b]×R2M +2L → R
are continuous, and continuously differentiable with respect to (x, y), (u, v) and
(x, y, u, v), respectively.
In the sequel, the notation for the standard symplectic matrix
 
0 −I
J= ,
I 0
where I is the identity matrix, will also be used in any other dimensions.
Here are our hypotheses.
A1. For every i ∈ {1, . . . , M } there exists κi > 0 such that the functions
H(t, x, y) and P (t, x, y, u, v) are κi -periodic in the variable xi .
The periodicity assumption A1 naturally leads us to consider the torus
TM = (R/κ1 Z) × · · · × (R/κM Z).
Indeed, in view of this assumption, the x component of the solutions could
sometimes be interpreted as belonging to TM .
A2. The function P (t, x, y, u, v) has a bounded gradient with respect to
(x, y, u, v).
Assumption A2 guarantees that the coupling term P (t, x, y, u, v) can be
seen as some kind of not so large perturbation term.
A3. All the solutions of system (1.1) satisfying y(a) = v(a) = 0 are defined on
[a, b].
In view of the results in [5,7], assumption A3 is surely satisfied if there
exists a constant K1 such that
|∇x H(t, x, y)| ≤ K1 (1 + |y|), for every (t, x, y) ∈ [a, b] × TM × RM .
A4. The function H : R2L → R is of the type
L
H (u, v) = Hj (uj , vj ),
j=1
NoDEA Boundary value problems associated Page 3 of 28 41

for some functions Hj : R2 → R which are positively-(pj , qj )-homogeneous and

positive, meaning that for some pj > 1 and qj > 1 with 1/pj + 1/qj = 1
we have
Hj (γ qj r, γ pj s) = γ pj +qj Hj (r, s) > 0, for every (r, s) ∈ R2 \ {0} and γ > 0.
In this setting, the origin (0, 0) is an isochronous center for the planar
autonomous system
J ζ̇ = ∇Hj (ζ). (1.3)
For every j ∈ {1, . . . , L}, besides the origin all solutions of system (1.3) are
periodic and have the same minimal period, which will be denoted by τj .
Moreover, for all solutions ζ(t) = (ζ1 (t), ζ2 (t)) of (1.3) starting with ζ1 (0) < 0
and ζ2 (0) = 0, there is a first time τj+ > 0 for which ζ2 (τj+ ) = 0, while
ζ2 (t) > 0 for all t ∈ ]0, τj+ [, and this time τj+ is independent of ζ1 (0) < 0.
Similarly, if ζ1 (0) > 0 and and ζ2 (0) = 0, there is a first time τj− > 0 for which
ζ2 (τj− ) = 0, while ζ2 (t) < 0 for all t ∈ ]0, τj− [, and this time τj− is independent
of ζ1 (0) > 0. Clearly enough, τj = τj+ + τj− .
Here is our main result.

Theorem 1.1. Assume that A1 – A4 hold true. Let τj+ = τj− and
b−a
∈
/ N, for every j ∈ {1, . . . , L}.
τj+
Then there are at least M + 1 geometrically distinct solutions of the boundary
value problem (1.1)–(1.2).

Notice that, when a solution has been found, infinitely many others ap-
pear by just adding an integer multiple of κi to the xi -th component. We say
that two solutions are geometrically distinct if they cannot be obtained from
each other in this way.
Let us remark here that a sufficient condition for having satisfied the
assumption τj+ = τj− is that the function Hj is even in v. This is a frequent
case in the applications, where, e.g., Hj is quadratic in v.
Theorem 1.1 generalizes the result in [7], where the case P ≡ 0 was
treated, dealing only with the system in (x, y). In order to prove it, we first
consider the case when, writing w = (u, v), the second Hamiltonian functions is
of the type H (w) = 12 Aw, w, where A is a particular diagonal matrix. Then,
by a symplectic change of variables, we are able to transform the positively-
(p, q)-homogeneous Hamiltonian in the quadratic one.
We also study the periodic problem for such kind of Hamiltonian sys-
tems, and obtain a similar multiplicity result when a suitable twist condition
is assumed. This part of the paper is related to the Poincaré–Birkhoff Theorem
[15], and we exploit some results obtained in [4], where any symmetric matrix
A can be considered, provided that a nonresonance condition is also assumed.
We thus generalize to this setting some results obtained in [3,8,9].
At the end of the paper we will analyze the possibility of dealing with
any symmetric matrix A, provided that a nonresonance condition is assumed,
41 Page 4 of 28 A. Fonda and W. Ullah NoDEA

also for the Neumann-type problem. However, we succeed doing this only in
the case L = 1, while the case L ≥ 2 remains an open problem.
Let us describe more in detail how the paper is organized.
In Sect. 2 we study the Neumann-type boundary value problem in the
particular case when
L
1
H (u, v) = λj (u2j + vj2 ),
2 j=1

for some positive constants λ1 , . . . , λL . The proof is variational, and it is mod-

eled on the method developed in [7]. However, some delicate estimates are
needed in order to prove the invertibility of the involved selfadjoint operator.
In Sect. 3 we provide the proof of Theorem 1.1. The idea is to construct
a symplectic change of variables, so to reduce the problem to the one already
treated in Sect. 2.
In Sect. 4 we study the periodic problem. Here we need to introduce a
twist condition, which recalls the classical assumption in the Poincaré–Birkhoff
Theorem. We obtain a similar multiplicity result as in Theorem 1.1 by applying
a corollary of the main result in [4].
Some possible applications are given in Sect. 5. For example, we propose
a system of the type

ẋ = f (y) + E(t), ẏ = −A sin x − ∂x P (t, x, u),
u̇ = |v|q−2 v, v̇ = −μ(u+ )p−1 + ν(u− )p−1 + ∂u P (t, x, u),
where u+ = max{u, 0} and u− = max{−u, 0}. The first two equations can be
seen as a generalization of the pendulum equation (obtained when f (y) = y),
while the last two equations correspond to the scalar equation
d
|u̇|p−2 u̇ + μ(u+ )p−1 − ν(u− )p−1 = ∂u P (t, x, u).
dt
Notice that the particular case p = 2 leads to a classical asymmetric oscillator.
Both Neumann-type and periodic problems are analyzed.
Finally, in Sect. 6 we end with some further remarks and proposing an
open problem.
In all the rest of the paper we will denote by ·, · and | · | the Euclidean
scalar product and norm on Rk , for any k ∈ N.

2. Coupling with a linear system

In this section we consider a Hamiltonian system of the type
⎧
⎪
⎨ẋ = ∇y H(t, x, y) + ∇y P (t, x, y, w) ,
ẏ = −∇x H(t, x, y) − ∇x P (t, x, y, w), (2.1)
⎪
⎩
J ẇ = Aw + ∇w P (t, x, y, w).
Here, the functions H : [a, b] × R2M → R and P : [a, b] × R2M +2L → R are
continuous, and continuously differentiable with respect to (x, y) and (x, y, w),
NoDEA Boundary value problems associated Page 5 of 28 41

respectively. The 2L × 2L matrix A is of the type

 
BL 0
A= , (2.2)
0 BL
where
⎡ ⎤
λ1 0 ... 0
⎢0 λ2 ... 0 ⎥
⎢ ⎥
BL = ⎢ . .. .. .. ⎥,
⎣ .. . . . ⎦
0 ... 0 λL
for some positive real numbers λ1 , . . . , λL . Writing
x = (x1 , . . . , xM ) ∈ RM , y = (y1 , . . . , yM ) ∈ RM ,
and w = (u, v) ∈ R2L , with
u = (u1 , . . . , uL ) ∈ RL , v = (v1 , . . . , vL ) ∈ RL ,
we consider the Neumann-type boundary conditions

y(a) = 0 = y(b),
(2.3)
v(a) = 0 = v(b).
Here is the main result of this section.
Theorem 2.1. Assume that A1 – A3 hold true, and
b−a
λj ∈
/ N, for every j ∈ {1, . . . , L}.
π
Then, the boundary value problem (2.1)–(2.3) has at least M + 1 geometrically
distinct solutions.
Proof. Without loss of generality, we may assume that [a, b] = [0, π]. By A3
and a standard compactness argument, there exists a constant K2 > 0 such
that, for any solution (x, y, w) of (2.1) satisfying y(0) = v(0) = 0, one has that
|y(t)| ≤ K2 , for every t ∈ [0, π]. (2.4)
Notice indeed that, by the nonresonance assumption we have, one can find a C 1
a priori bound on w, which hence belongs to a compact subset of C([0, π], R2L ).
Let σ : R → R be a C ∞ -function such that

1, if |s| ≤ K2 ,
σ(s) =
0, if |s| ≥ K2 + 1,
set
 x, y) = σ(|y|)H(t, x, y),
H(t, (2.5)
and consider the modified system
⎧
⎪ 
⎨ẋ = ∇y H(t, x, y) + ∇y P (t, x, y, w) ,
 x, y) − ∇x P (t, x, y, w),
ẏ = −∇x H(t, (2.6)
⎪
⎩
J ẇ = Aw + ∇w P (t, x, y, w).
41 Page 6 of 28 A. Fonda and W. Ullah NoDEA

The new Hamiltonian function is thus

 x, y, w) = H(t,
H(t,  x, y) + 1 Aw, w + P (t, x, y, w). (2.7)
2
We will prove that the boundary value problem (2.6)–(2.3) has at least M + 1
geometrically distinct solutions. By the above argument, these solutions will
satisfy (2.4), hence they will be the solutions of (2.1)–(2.3) we are looking for.
The proof is variational, and it is based on a theorem by Szulkin recalled
below. We will now introduce the function spaces and the needed functionals.
2.1. The function spaces
For any α ∈ ]0, 1[ , we define Xα as the set of those real valued functions
x̃ ∈ L2 (0, π) such that
∞
x̃(t) ∼ x̃m cos(mt),
m=1

where (x̃m )m≥1 is a sequence in R satisfying

∞
m2α x̃2m < ∞.
m=1
The space Xα is endowed with the inner product and norm

∞  ∞

x̃, φ̃Xα = 2α
m x̃m φ̃m , ||x̃||Xα =  m2α x̃2m .
m=1 m=1

For any β ∈ ]0, 1[ , we define Yβ as the set of those real valued functions y ∈
L2 (0, π) such that
∞
y(t) ∼ ym sin(mt),
m=1

where (ym )m≥1 is a sequence in R satisfying

∞
m2β ym
2
< ∞.
m=1
The space Yβ is endowed with the inner product and norm

∞  ∞

y, ρYβ = 2β
m ym ρm , ||y||Yβ =  m2β ym
2 .

m=1 m=1

From now on, we will consider functions x, y, u, v which can be written as


1 π
x(t) = x̄ + x̃(t) , x̄ = x(t) dt ,
π 0

1 π
u(t) = ū + ũ(t) , ū = u(t) dt ,
π 0
where x̃ and y belong to the spaces XαM and YβM respectively, while functions
ũ and v belong to the spaces XαL and YβL respectively.
NoDEA Boundary value problems associated Page 7 of 28 41

Choose two positive numbers α, β such that

1
α< 2 <β and α + β = 1.
Consider the space E = XαM × YβM × (RL × XαL ) × YβL , and the torus TM =
(R/κ1 Z) × · · · × (R/κM Z). The space E is endowed with the scalar product
(x̃ , y , ū , ũ , v), (X  , V )E =x̃, X
 ,Y ,U ,U  X M + y, Y Y M +
α β

 X L + v, V Y L ,
+ ū , U  + ũ, U α β

and the corresponding norm


||(x̃, y, ū, ũ, v)||E = ||x̃||2X M + ||y||2Y M + |ū|2 + ||ũ||2X L + ||v||2Y L .
α β α β

Since Xα , Yβ and R are separable Hilbert spaces [7, Proposition 2.3 and 2.6],
the same is true for E.
By A1, the Hamiltonian function H  in (2.7) is κi -periodic in xi for i ∈
{1, . . . , M }, hence writing x(t) = x̄ + x̃(t), with

1 π
x̄ = x(t) dt,
π 0
we can assume that x̄ ∈ TM and look for solutions (z, x̄) ∈ E × TM , where
z = (x̃, y, ū, ũ, v).
These solutions will be found as critical points of a suitable functional, by
applying the following theorem of Szulkin [18] (see also [10,13]).

Theorem 2.2. ([18]) If ϕ : E × TM → R is a continuously differentiable

functional of the type
1
ϕ(z, x̄) = L z, zE + ψ(z, x̄),
2
where L : E → E is a bounded selfadjoint invertible operator and dψ(E ×TM )
is relatively compact, then ϕ has at least M + 1 critical points.

2.2. The functional and the bilinear form

We define a functional ψ : E × TM → R as
ψ(z, x̄) = ψ (x̃ , y , ū , ũ, v) , x̄
 π
=  t , x̄ + x̃(t) , y(t) , ū + ũ(t) , v(t) dt .
H
0

In the following, we will treat TM as being lifted to RM , so E × TM will often

be identified with E × RM . It has been shown in [7, Proposition 2.10] and
[6, Proposition 19, Proposition 22] that ψ is continuously differentiable, and
the gradient function ∇ψ has a relatively compact image. In what follows we
introduce the operator L .
We first consider the space
 1 ([0, π])]M × [C01 ([0, π])]M × FL ,
D = [C
41 Page 8 of 28 A. Fonda and W. Ullah NoDEA

where
 1 ([0, π])]L × [C01 ([0, π])]L ,
FL = RL × [C
and define a symmetric bilinear form B : D × D → R as follows. For every
z = (x̃, y, ū, ũ, v) and Z = (X,  Y, U , U
 , V ) in D,
 π 
B(z, Z) =  − x̃ , Y  − J ẇ, W  + Aw, W  dt,
y  , X
0
 , V ) are in FL .
where w = (ū + ũ, v), W = (U + U
Proposition 2.3. The set D is dense in E, and the bilinear form B : D×D → R
is continuous with respect to the topology of E × E .
Proof. We know by [7, Proposition 2.5 and 2.8] that D is a dense subspace of
E. In order to prove the second part of the statement, let us write
 Y ) + B2 (w, W ),
B(z, Z) = B1 (x̃, y), (X,
where
 π  
 Y) =
B1 (x̃, y), (X,  − x̃ , Y  dt,
y  , X (2.8)
0
and
 π  
B2 (w, W ) = − J ẇ, W  + Aw, W  dt. (2.9)
0
It has been proved in [6, Section 3.4] that B1 is continuous with respect to the
topology of XαM × YβM . We need to prove that B2 is continuous with respect to
the topology of RL × XαL × YβL . For w = (w1 , . . . , wL ) and W = (W1 , . . . , WL )
in FL we have
 π L  π
J ẇ, W dt = J ẇj , Wj dt, (2.10)
0 j=1 0

and, writing wj = (ūj + ũj , vj ), Wj = (U j + U j , Vj ),

 π  π  π  π
J ẇj , Wj dt = u̇j Vj dt − v̇j U j dt − j dt.
v̇j U (2.11)
0 0 0 0
We decompose the involved functions as
∞ ∞
j
vj = vm sin(mt), Vj = Vmj sin(mt),
m=1 m=1
∞ ∞
ũj = ũjm cos(mt), j =
U  j cos(mt).
Um
m=1 m=1

By the boundary condition v(0) = 0 = v(π), we see that

 π
v̇j U j dt = 0.
0
Recalling that α + β = 1, we have
NoDEA Boundary value problems associated Page 9 of 28 41

   
 π  π  ∞


 u̇j Vj dt =  −m ũjm Vmj 
 2 
0 m=1
∞
π  α j β j
≤ m ũm m Vm 
2 m=1
π
≤ ||ũj ||Xα ||Vj ||Yβ ,
2
and
   
 π  π  ∞
j j 

 j dt = 
v̇j U m vm Um 
  2 
0 m=1
∞  
π  α j β j 
≤
m Um m vm 
2 m=1
π 
≤ ||U j ||Xα ||vj ||Yβ .
2
Going back to (2.11), for each j ∈ {1, . . . , L}, we thus have
 
 π  π
 J ẇj , Wj dt ≤ ||wj ||R×Xα ×Yβ ||Wj ||R×Xα ×Yβ .
 2
0
Hence, by (2.10),
 π 
  π
 J ẇ, W dt ≤ ||w||RL ×XαL ×YβL ||W ||RL ×XαL ×YβL .
 2
0
We have thus proved the continuity of the first part of the bilinear form defined
in (2.9).
For the second part, we can write
 π L  π
Aw, W dt = λj wj , Wj  dt, (2.12)
0 j=1 0

where
 
π π  
wj , Wj  dt = (ūj + ũj , vj ), (U j + Uj , Vj ) dt
0
0 π  π
= (ūj + ũj )(U j + U j )dt + vj Vj dt .
0 0
Now for every j = 1, . . . , L, we have
 π   π   π 
     
  
(ūj + ũj )(U j + Uj )dt ≤    
ūj U j  +  
ũj Uj 

0
 0 0
π ∞ 
 mj 
≤ π|ūj ||U j | +  ũjm U 
2 
m=1
∞
π  α j α j
≤ π|ūj ||U j | + m ũm m U m 
2 m=1
π j ||X ,
≤ π|ūj ||U j | + ||ũj ||Xα ||U α
2
41 Page 10 of 28 A. Fonda and W. Ullah NoDEA

while
   ∞


 π  π 
 vj Vj dt =  j
vm V mj 
 2 
0 m=1
∞
π  β j β j
≤  m vm m V m 
2 m=1
π
≤ ||vj ||Yβ ||Vj ||Yβ .
2
Thus we have
 π 
 
 w , W dt  ≤ π||wj ||R×X ×Y ||Wj ||R×X ×Y ,
 j j  α β α β
0
and, going back to (2.12),
 
 π   L  π 
   
 
Aw, W dt =  λj wj , Wj dt

0  j=1 0 
 π 
L
 
≤ λj  wj , Wj dt
j=1 0

L
≤ πλj ||wj ||R×Xα ×Yβ ||Wj ||R×Xα ×Yβ
j=1
≤ πλ||w||RL ×XαL ×YβL ||W ||RL ×XαL ×YβL ,
where λ = max{λ1 , . . . , λL }. This shows that also the second part of the
bilinear form B2 : D × D → R in (2.9) is continuous, and the proof is complete.

The bilinear form B : D × D → R can thus be extended in a unique way
to a continuous symmetric bilinear form B : E ×E → R, for which we maintain
the same notation. A bounded selfadjoint operator L : E → E can thus be
defined by
L z, ZE = B(z, Z),
for z and Z in E. Referring to (2.8) and (2.9), we can write
L (x̃, y, ū, ũ, v) = (L1 (x̃, y), L2 (w)),
where
 Y )X M ×Y M = B1 (x̃, y), (X,
L1 (x̃, y), (X,  Y) ,
α β

and
L2 (w), W RL ×XαL ×YβL = B2 (w, W ),
 Y, U , U
for every z = (x̃, y, ū, ũ, v) and Z = (X,  , V ) in E with w = (ū, ũ, v),

and W = (U , U , V ). It has been proved in [7, Proposition 2.14] that
π
||L1 (x̃, y)||XαM ×YβM = ||(x̃, y)||XαM ×YβM . (2.13)
2
NoDEA Boundary value problems associated Page 11 of 28 41

We now need the following.

Lemma 2.4. There exist positive constants α, β, δ with α < 12 < β, and α+β =
1 such that
||L2 (w)|| L L L ≥ δ ||w|| L L L ,
R ×Xα ×Yβ R ×Xα ×Yβ (2.14)
for every w ∈ RL × XαL × YβL .
 ξ) ∈ R × Xα × Yβ be such that
Proof. We first assume L = 1. Let (ζ, ζ,

L2 (w) = (ζ, ζ, ξ), so that
 ξ), W R×X ×Y ,
B2 (w, W ) = (ζ, ζ, (2.15)
α β

for every W = (U, V ) ∈ R × Xα × Yβ . Recalling that w = (ū, ũ, v), we decom-

pose
∞ ∞
ũ = um cos(mt), v= vm sin(mt),
m=1 m=1
∞ ∞
ζ = ζm cos(mt), ξ= ξm sin(mt).
m=1 m=1

By taking first V = 0 and then U = 0 in (2.15), and using (2.9), we obtain the
following identities
⎧
⎪
⎨ζ = λ1 π ū,
ζm m2α = π2 [λ1 um + mvm ], (2.16)
⎪
⎩ 2β π
ξm m = 2 [mum + λ1 vm ].
Thus we have
π π
ζm mα = [λ1 m−α um + mβ vm ], ξm mβ = [mα um + λ1 m−β vm ],
2 2
and, by using the Young inequality,
2 π 2  2 −2α 2
ζm m2α + ξm2
m2β = λ m um + m2β vm
2
+ m2α u2m
4 1 
+ λ21 m−2β vm
2
+ 2λ1 [mα−β + mβ−α ]um vm
π 2  2 −2α 2
≥ λ m um + m2β vm2
+ m2α u2m
4 1 
+ λ21 m−2β vm
2
− λ1 [mα−β + mβ−α ] u2m + vm2

π2  
= m−4α (λ1 − m) λ1 − m4α−1 m2α u2m
4
π 2 −4β  
+ m (λ1 − m) λ1 − m4β−1 m2β vm 2
. (2.17)
4
By hypothesis, we know that there exists a nonnegative integer n1 such that
n1 < λ1 < n1 + 1.
We now discuss separately the cases for n1 = 0 and n1 ≥ 1.
41 Page 12 of 28 A. Fonda and W. Ullah NoDEA

Case 1. If n1 = 0, then 0 < λ1 < 1, and so λ1 < m for all m ≥ 1. Now for
m = 1, (2.17) implies that
π2
ζ12 + ξ12 ≥ (λ1 − 1)2 (u21 + v12 ). (2.18)
4
For m ≥ 2, we have
(λ1 − m)(λ1 − m4α−1 ) > (1 − m)(1 − m4α−1 ) = (m − 1)(m4α−1 − 1).
By writing m−4α = m−1 m−4α+1 , and choosing α such that
 
1 log(4/3) 1
+1 <α< ,
4 log 2 2
we have
  
−4α 1 1
m (λ1 − m) λ1 − m 4α−1
> 1− 1 − 4α−1
m m
  
1 1 1 λ2
≥ 1− 1 − 4α−1 ≥ ≥ 1 ,
2 2 8 8
1
since λ1 < 1. Similarly, since β > 2 > α, we get
λ21
m−4β (λ1 − m) λ1 − m4β−1 ≥ ,
8
and thus (2.17) implies that
π 2 λ21 2α 2 !
2
ζm m2α + ξm
2
m2β ≥ m um + m2β vm 2
. (2.19)
4 8
Combining (2.18), (2.19), and the first identity in (2.16), we have
 2 + ||ζ||
||L2 (ū, ũ, v)||2R×Xα ×Yβ = |ζ|2 + ||ζ||  2
Xα Yβ
∞
= π 2 λ21 |ū|2 + (ζ12 + ξ12 ) + 2
ζm m2α + ξm
2
m2β
m=2

π 2 λ21  2  1 2 2 
∞
≥ |ū| + 1 − [u1 + v12 ] + u2m m2α + vm
2
m2β
4 8 λ1 m=2
 ∞ 
≥ δ2 |ū|2 + u2m m2α + vm
2
m2β = δ2 ||(ū, ũ, v)||2R×Xα ×Yβ ,
m=1
where
"  1 #
π 
δ = λ1 min 1, 1 −  .
8 λ1
This implies that (2.14) holds in this case, for L = 1.
Case 2. If n1 ≥ 1, then for m ∈ {1, . . . , n1 } we have λ1 − m ≥ λ1 − n1 > 0,
and so
λ1 − m4α−1 ≥ λ1 − m ≥ λ1 − n1 > 0.
This implies that
m−4α (λ1 − m) λ1 − m4α−1 ≥ n−4α
1 (λ1 − n1 )2 ≥ n−4β
1 (λ1 − n1 )2 .
NoDEA Boundary value problems associated Page 13 of 28 41

By choosing β such that

 1 
1 1 log 2 (λ1+ n1 )
<β< +1 , (2.20)
2 4 log n1
we obtain that λ1 − m4β−1 ≥ λ1 − n4β−1
1 > 12 (λ1 − n1 ) > 0 , and so
1
m−4β (λ1 − m) λ1 − m4β−1 ≥ n−4β 1 (λ1 − n1 )2 .
2
Thus, for m ∈ {1, . . . , n1 }, (2.17) and (2.20) imply that
π 2 −4β !
2
ζm m2α + ξm
2
m2β ≥ n1 (λ1 − n1 )2 m2α u2m + m2β vm
2
. (2.21)
8
For m = n1 + 1, we have λ1 − m = λ1 − (n1 + 1) < 0, and so
λ1 − m4β−1 = λ1 − (n1 + 1)4β−1 < λ1 − (n1 + 1) < 0.
This implies that
m−4β (λ1 − m) λ1 − m4β−1 ≥ (n1 + 1)−4β (λ1 − (n1 + 1))2 .
By choosing α such that
 
1 log 12 (λ1 + n1 + 1) 1
+1 ≤α< , (2.22)
4 log(n1 + 1) 2
we obtain
1
λ1 − m4α−1 = λ1 − (n1 + 1)4α−1 ≤ λ1 − (n1 + 1) < 0,
2
and so
1 2
m−4α (λ1 − m) λ1 − m4α−1 ≥ (n1 + 1)−4α λ1 − (n1 + 1)
2
1 2
≥ (n1 + 1)−4β λ1 − (n1 + 1) .
2
Thus, for m = n1 + 1, (2.17) and (2.22) imply that
π2 !
2
ζm m2α + ξm
2
m2β ≥ (n1 + 1)−4β (λ1 − (n1 + 1))2 m2α u2m + m2β vm
2
.
8
(2.23)
Lastly, for m ≥ n1 + 2, by choosing α such that
 2(n1 +1)(n1 +2) 
1 log 2n1 +3 1
+1 ≤α< , (2.24)
4 log(n1 + 2) 2
we have
(λ1 − m)(λ1 − m4α−1 ) > (n1 + 1 − m)(n1 + 1 − m4α−1 )
= (m − (n1 + 1))(m4α−1 − (n1 + 1)) ,
and, writing m−4α = m−1 m−4α+1 ,
41 Page 14 of 28 A. Fonda and W. Ullah NoDEA

  
−4α n1 + 1 n1 + 1
m (λ1 − m) λ1 − m 4α−1
> 1− 1−
m m4α−1
  
n1 + 1 n1 + 1
≥ 1− 1−
n1 + 2 (n1 + 2)4α−1
 2
1 n1 + 1 1 1
≥ 1− = .
2 n1 + 2 2 (n1 + 2)2
1
Similarly, since β > 2 > α, we obtain
1 1
m−4β (λ1 − m) λ1 − m4β−1 ≥ .
2 (n1 + 2)2
Hence for m ≥ n1 + 2, (2.17) and (2.24) imply that
π2 1 !
2
ζm m2α + ξm
2
m2β ≥ 2
m2α u2m + m2β vm
2
. (2.25)
8 (n1 + 2)
Combining (2.21), (2.23), (2.25), and the first identity in (2.16) we have
 2 + ||ζ||
||L2 (ū, ũ, v)||2R×Xα ×Yβ = |ζ|2 + ||ζ||  2
Xα Yβ
n1
!
= π 2 λ21 |ū|2 + 2
ζm m2α + ξm
2
m2β +
m=1
∞
! !
+ ζn21 +1 (n1 + 1)2α + ξn2 1 +1 (n1 + 1)2β + 2
ζm m2α + ξm
2
m2β
m=n1 +2
$
n1 2 1 !
2 2 n
π π
≥ |ū|2 + (n1 + 1)−4β 1 − m2α u2m + m2β vm 2
8 8 λ1 m=1
 
n1 + 1 2  %
+ 1− (n1 + 1)2α u2n1 +1 + (n1 + 1)2β vn2 1 +1 +
λ1
∞
π2 1 !
+ m2α u2m + m2β vm
2
8 (n1 + 2)2 m=n1 +2
& ∞
'
≥δ 2 |ū| +
2
u2m m2α + 2
vm m2β = δ2 ||(ū, ũ, v)||2R×Xα ×Yβ ,
m=1
where
(  )
 π 1 −2β  n1 + 1  
−2β  n1 
δ = √ min , (n1 + 1) 1 − , (n1 + 1) 1 −  .
2 2 n1 + 2 λ1 λ1
This implies that (2.14) holds also in this case, for L = 1.
Finally, by using (2.10) and (2.12), we can easily see that (2.14) holds for
any L ≥ 1. 
By combining (2.13) and (2.14) in Lemma 2.4, we can say that the selfad-
joint operator L : E → E is invertible, and the inverse operator L −1 : E → E
is continuous.
By Theorem 2.2, we conclude that the functional ϕ has at least M + 1
critical points. Arguing as in [6, Proposition 24], it can be seen that these
NoDEA Boundary value problems associated Page 15 of 28 41

critical points correspond to the solutions of the boundary value problem (2.6)–
(2.3) that we are looking for. The proof of Theorem 2.1 is thus completed. 

3. Proof of Theorem 1.1

Without loss of generality, we may assume that [a, b] = [0, π]. We start assum-
ing L = 1, and we first work on the planar system (1.3) so to transform it, by
a symplectic change of variables, into a linear one. We will follow the approach
developed in [1,8,11].

3.1. A symplectic change of variables

By using A4, we have that H (0, 0) = 0 and the generalized Euler Identity
holds true, i.e.,
*  +
u v
∇H (u, v), , = H (u, v). (3.1)
p q
Choose the positive constant
( )
1
Υ = min H (w) : 1 ≤ |w| ≤ 2 , (3.2)
|w|2
and let η : R → R be a C ∞ -function such that η  (s) ≤ 0 for all s ∈ R and

1, if s ≤ 1,
η(s) =
0, if s ≥ 2.
For w = (u, v), set
,(w) = η(|w|)Υ|w|2 + (1 − η(|w|))H (w),
H (3.3)
and consider the new system
,(w).
J ẇ = ∇H (3.4)
Notice that H,(0) = 0. For every w = 0, we have
  
∇H,(w) = Υη  (|w|)|w| + 2Υη(|w|) − η (|w|) H (w) w + (1 − η(|w|))∇H (w).
|w|
Then, using (3.1) and (3.2), if w = (u, v) is such that 1 ≤ |w| ≤ 2, we have
*  +  2 
, u v  u v2 1
∇H (w), , = η (|w|)|w| + Υ− H (w)
p q p q |w|2
 2 
u v2
+2η(|w|) Υ + + (1 − η(|w|))H (w) > 0.
p q
This implies that ∇H ,(w) = 0, for 1 ≤ |w| ≤ 2. For 0 < |w| ≤ 1, the
Hamiltonian function H, is quadratic, so that ∇H
,(w) = 0. Lastly, for |w| ≥ 2,
,
we have ∇H (w) = ∇H (w), and it is clear from (3.1) that ∇H (w) = 0. Hence
∇H ,(w) = 0 for every w = 0, and this shows that every non-zero solution of
system (3.4) does not pass through the origin, and by Poincaré–Bendixson
41 Page 16 of 28 A. Fonda and W. Ullah NoDEA

theory, all the solutions of system (3.4) are periodic. Thus the origin is still a
global center for the system (3.4).
Now for any w0 ∈ R2 \ {0}, we denote by T(w0 ) the minimal period of
the solution of (3.4) passing through w0 . We notice here that this solution is
unique, even if we are not assuming ∇H to be locally Lipschitz continuous, cf.
[16]. The function T : R2 \{0} → R thus defined is continuously differentiable
(see [1]).
Define
δ  = [0, +∞[×{0},
and a function ξ : ]0, +∞[ → ]0, +∞[ as follows: for every r > 0, the level line
{w ∈ R2 : H ,(w) = r} intersects δ  at the point (ξ(r), 0). Such a point is
unique, because for every (ξ, 0) ∈ δ  with ξ = 0 we have
*  +
, ξ
∇H (ξ, 0), ,0 > 0,
p
which implies that
 
,(ξ, 0), (ξ, 0) > 0.
∇H
Thus, if w(t0 ) = (u(t0 ), v(t0 )) = (u(t0 ), 0) is such that u(t0 ) > 0, then v  (t0 ) <
0, and so it is impossible for the level line {w ∈ R2 : H ,(w) = r} to intersect

δ at two different points.
Now define K  : R2 → R as
 ,
 1 H (w) 
K(w) = T (ξ(r), 0) dr.
τ 0
This function is continuously differentiable, and

 T(w) ,
∇K(w) = ∇H (w).
τ
Hence, the origin is an isochronous center for the system

J ẇ = ∇K(w), (3.5)
since all solutions except the equilibrium 0 are periodic with minimal period
τ . Moreover,
 π
K(w) = |w|2 , if |w| ≤ 1.
τ
Now, for every w0 ∈ R2 \ {0}, let ζ(t; w0 ) be the solution of system (3.5)
satisfying ζ(0; w0 ) = w0 , and define θ(w0 ) ∈ [0, 2π[ as the minimum time for
which
 τ 
ζ − θ(w0 ); w0 ∈ δ  .
2π
As shown in [1], the restricted function θ : R2 \ δ  → ]0, 2π[ is continuously
differentiable, and its gradient ∇θ can be continuously extended to R2 \ {0}.
We will still denote this extension by ∇θ : R2 \ {0} → R2 .
NoDEA Boundary value problems associated Page 17 of 28 41

Hence, by [1, Proposition 2.2.], there exists a symplectic diffeomorphism

Λ : R2 → R2 defined by
⎧-
⎨ τ 
K(w) cos θ(w), − sin θ(w) , if w = 0,
Λ(w) = π
⎩
0, if w = 0,
such that, by the change of variable z = Λ(w), system (3.5) is changed to the
linear one
2π
J ż = z.
τ
3.2. The proof in the case L = 1
First, we can modify the function H as in (2.5) so to obtain the modified
system
⎧
⎪  x, y) + ∇y P (t, x, y, u, v) ,
ẋ = ∇y H(t,
⎪
⎪
⎨ẏ = −∇ H(t,
 x, y) − ∇x P (t, x, y, u, v),
x
(3.6)
⎪
⎪ u̇ = ∇v H (u, v) + ∇v P (t, x, y, u, v),
⎪
⎩
v̇ = −∇u H (u, v) − ∇u P (t, x, y, u, v).
Using the argument in [5, Section 3], it can be seen that all the solutions
of this system are globally defined. Moreover, those satisfying the boundary
conditions

y(0) = 0 = y(π),
(3.7)
v(0) = 0 = v(π)
are solutions of the original system (1.1).
Recalling the change of variables Λ(w) = z in Sect. 3.1, we define a map
P(t, x, y, z) = P (t, x, y, Λ−1 (z)).
Lemma 3.1. The function P has a bounded gradient with respect to (x, y, z).
Proof. Clearly, by A2 both
∂x P(t, x, y, z) = ∂x P (t, x, y, Λ−1 (z)), ∂y P(t, x, y, z) = ∂y P (t, x, y, Λ−1 (z))
are bounded and denoting by M∗ the transpose of a matrix M,
!∗
∇z P(t, x, y, z) = (Λ−1 (z)) ∇w P (t, x, y, Λ−1 (z))
!−1
= (Λ (Λ−1 (z)))∗ ∇w P (t, x, y, Λ−1 (z))) .
Again by A2, ∇w P (t, x, y, w) is bounded, so it is sufficient to show that

(Λ (w))−1 is bounded. For |w| large enough, we have that K(w) = H (w).
By denoting c(w) = cos θ(w) and s(w) = sin θ(w), we have
$ %
 a11 (w) a12 (w)
Λ (w) = ,
a21 (w) a22 (w)
where
-  
τ ∂u H (w) .
a11 (w) = . c(w) − H (w)∂u θ(w)s(w) ,
π 2 H (w)
41 Page 18 of 28 A. Fonda and W. Ullah NoDEA

-  
τ ∂v H (w) .
a12 (w) = . c(w) − H (w)∂v θ(w)s(w) ,
π 2 H (w)
-  
τ ∂u H (w) .
a21 (w) = − . s(w) − H (w)∂u θ(w)c(w) ,
π 2 H (w)
-  
τ ∂v H (w) .
a22 (w) = − . s(w) − H (w)∂v θ(w)c(w) .
π 2 H (w)
Recalling that Λ is symplectic, so det Λ (w) = 1, the inverse matrix is
$ %
a22 (w) −a12 (w)
(Λ (w))−1 = .
−a21 (w) a11 (w)
From the definition of θ, for w = 0 and γ > 0 we see that θ(γ q u, γ p v) =
θ(u, v). Indeed, if w(t) = (u(t), v(t)) is a solution of system (3.5), then wγ =
(γ q u, γ p v) is also a solution of system (3.5) with vertical component of the
τ
velocity equal to γ p v̇(t). Hence, if w(t) needs a time 2π θ(u0 , v0 ) to go from δ 
to (u0 , v0 ) (it has a vertical speed v̇(t)), then the time for wγ (t) to go from δ 
to (γ q u0 , γ p v0 ) must be the same, since its vertical speed is just γ p times the
vertical speed of w(t). Thus we have
∂u θ(γ q u, γ p v)γ q = ∂u θ(u, v), ∂v θ(γ q u, γ p v)γ p = ∂v θ(u, v),
for every γ > 0. For w = (u, v) with |w| ≥ 2, since H is positively-(p, q)-
homogeneous, the following identities have been proved in [5]:
∂H q ∂H ∂H
(γ u, γ p v) = γ q(p−1) (u, v) = γ p (u, v),
∂u ∂u ∂u
∂H q ∂H ∂H
(γ u, γ p v) = γ p(q−1) (u, v) = γ q (u, v).
∂v ∂v ∂v
Thus we have
-  
τ |∂v H (w)| .
|a22 (w)| ≤ . + H (w) |∂v θ(w)|
π 2 H (w)
  
 u v 
- |w|q ∂v H , 
τ |w| |w| 
q p
= /  +
π u v
2|w| p+q H ,
|w|2q |w|2p
- /    
τ |w|p+q u v u v
+ H , ∂v θ ,
π |w|p(p+q) |w|2q |w|2p |w|q(p+q) |w|p(p+q)
  
 u v 
- ∂v H , 
τ |w| |w| 
q p
≤ /  +
π u v
2 H ,
|w|2q |w|2p
- /    
τ u v u v
+ H , ∂ v θ , .
π |w|2q |w|2p |w|q(p+q) |w|p(p+q)
NoDEA Boundary value problems associated Page 19 of 28 41

Define three types of sets as follow:

(  )
u v
S= , : w = (u, v), |w| ≥ 1 ,
|w|q |w|p
(  )
u v
S = , : w = (u, v), |w| ≥ 1 ,
|w|2q |w|2p
and
(  )
 u v
S = , : w = (u, v), |w| ≥ 1 .
|w|q(p+q) |w|p(p+q)
It is easy to see that the sets S, S  , and S  are subsets of the closed unit ball
B(0, 1) of R2 . This implies that |a22 (w)| is bounded, since the functions H and
θ are C 1 . Similarly we can show that all of the other elements of the matrix
(Λ (w))−1 are bounded, which thus proves that the map P has a bounded
gradient with respect to z. 

Now we consider the modified system

⎧
 
⎪ẋ = ∇y H(t, x, y) + ∇y P (t, x, y, ξ, ζ),
⎪
⎪
⎪
⎪  x, y) − ∇x P(t, x, y, ξ, ζ),
⎨ẏ = −∇x H(t,
2π (3.8)
⎪
⎪ ξ˙ = ζ + ∂ζ P(t, x, y, ξ, ζ),
⎪
⎪ τ
⎪
⎩ζ̇ = − 2π ξ − ∂ P(t, x, y, ξ, ζ),
ξj
τ
where z = (ξ, ζ). By the assumption τ+ = τ− , the boundary conditions become

y(0) = 0 = y(π),
(3.9)
ζ(0) = 0 = ζ(π).

Thus, by taking λ1 = 2π τ , all the assumptions of Theorem 2.1 are satisfied, so

that the boundary value problem (3.8)–(3.9) has at least M + 1 geometrically
distinct solutions.
Recalling that Λ is a diffeomorphism, we can apply the inverse change of
variables w = Λ−1 (z), and obtain the M + 1 geometrically distinct solutions
of system (3.6) satisfying the boundary conditions (3.7) we were looking for.
This completes the proof of Theorem 1.1 in the case L = 1. 

3.3. The proof in the higher dimensional case

We now consider the case L ≥ 2, for which we will follow briefly the lines of
,j as in (3.3) and consider
the proof in the previous section. We can define H
the new system
,j (ζ).
J ζ̇ = ∇H
 j : R2 → R as
We define K
 H,j (ζ)
 j (ζ) = 1
K Tj (ξj (r), 0) dr,
τj 0
41 Page 20 of 28 A. Fonda and W. Ullah NoDEA

so that the origin is an isochronous center for the system

 j (ζ),
J ζ̇ = ∇K (3.10)
i.e., for every j ∈ {1, . . . , L}, all solutions of system (3.10) except the origin are
periodic and have the same minimal period τj . Now, for every j ∈ {1, . . . , L},
there exists a symplectic diffeomorphism Λj : R2 → R2 such that, by the
change of variables ρ = Λj (ζ), system (3.10) becomes
2π
J ρ̇ = ρ.
τj
By the use of a cut-off function, we modify the Hamiltonian H like in (2.5), so
that the new Hamiltonian H  has a bounded gradient with respect to (x, y).
Defining Λ : R → R by
2L 2L

Λ(u, v) = (Λ1 (u1 , v1 ), . . . , ΛL (uL , vL )) ,

we see that Λ is a symplectic diffeomorphism. By writing
P(t, x, y, z) = P (t, x, y, Λ−1 (z)),
as in Lemma 3.1 we can show that the function P has a bounded gradient
with respect to (x, y, z).
We apply the change of variables z = Λ(w) and write z = (ξ, ζ) with
ξ = (ξ1 , . . . , ξL ), ζ = (ζ1 , . . . , ζL ),
so to obtain the modified system
⎧
⎪  x, y) + ∇y P(t, x, y, z),
ẋ = ∇y H(t,
⎪
⎪
⎪
⎪  
⎨ẏ = −∇x H(t, x, y) − ∇x P (t, x, y, z),
⎪
2π
⎪ ξ˙j = ζj + ∂ζj P(t, x, y, z), j = 1, . . . , L, (3.11)
⎪
⎪ τ j
⎪
⎪ 2π
⎪
⎩ζ̇j = − ξj − ∂ξj P(t, x, y, z), j = 1, . . . , L.
τj
Moreover, since τj+ = τj− , the boundary conditions become the same as those
in (3.9). Hence, by taking λj = 2π τj , Theorem 2.1 implies that the modified
system (3.11) has at least M + 1 geometrically distinct solutions satisfying the
boundary conditions (3.9).
Recalling that Λ is a diffeomorphism, we can apply the inverse change of
variables w = Λ−1 (z) and obtain the solutions of problem (1.1)–(1.2) we are
looking for. 

4. The periodic problem

In this section, we consider the Hamiltonian system (1.1), where besides the
regularity assumptions already made on the functions involved, we assume
that all these functions are T -periodic in t. While maintaining assumptions
A1, A2 and A4 we will reinforce assumption A3 by a twist condition, and for
this we first recall some definitions.
NoDEA Boundary value problems associated Page 21 of 28 41

By a convex body of RM , we mean a closed convex bounded subset D

of R having nonempty interior. If in addition, D has a smooth boundary,
M

then we denote the unit outward normal at ζ ∈ ∂D by νD (ζ). Moreover, we

say that D is strongly convex if for any p̄ ∈ ∂D, the map F : D → R defined
by F(ξ) = ξ − p̄, νD (p̄) has a unique maximum point at ξ = p̄. Below is our
twist condition.
B3 . There are a strongly convex body D of RM having a smooth boundary
and a symmetric regular M × M matrix B such that, for every C 1 -function
W : [0, T ] → R2L , all the solutions (x, y) of system

ẋ = ∇y H(t, x, y) + ∇y P (t, x, y, W(t)),
(4.1)
ẏ = −∇x H(t, x, y) − ∇x P (t, x, y, W(t)),
starting with y(0) ∈ D are defined on [0, T ], and
y(0) ∈ ∂D ⇒ x(T ) − x(0), BνD (y(0)) > 0.
Here is our first result for the periodic problem.

Theorem 4.1. Assume that A1, A2, B3 and A4 hold true, and let
T
∈
/ N, for every j ∈ {1, . . . , L}.
τj
Then there are at least M + 1 geometrically distinct T -periodic solutions of
system (1.1), with y(0) ∈ D̊.
Proof. Following the lines of the proof of Theorem 1.1, we modify the problem
so to have a coupling with a perturbed linear system. Then, [4, Corollary 2.4]
applies (instead of Theorem 2.1), and the proof is readily completed. 
We can state some variants of Theorem 4.1 replacing the twist assumption
B3 by B3 or by B3 given below.
B3 . There exists a convex body D of RM , having a smooth boundary, such
that for σ ∈ {−1, 1} and for every C 1 -function W : [0, T ] → R2L , all the
solutions (x, y) of system (4.1) starting with y(0) ∈ D are defined on [0, T ],
and
y(0) ∈ ∂D ⇒ / {σλνD (y(0)) : λ ≥ 0}.
x(T ) − x(0) ∈
B3 . Let D be a rectangle in RM , i.e.
D = [c1 , d1 ] × · · · × [cM , dM ].
There exists an M -tuple σ = (σ1 , . . . , σM ) ∈ {−1, 1}M such that, for every
C 1 -function W : [0, T ] → R2L , all the solutions (x, y) of system (4.1) starting
with y(0) ∈ D are defined on [0, T ], and, for every i ∈ {1, . . . , M }, we have

yi (0) = ci ⇒ σi (xi (T ) − xi (0)) < 0,
yi (0) = di ⇒ σi (xi (T ) − xi (0)) > 0.
The proofs of such results are similar to those of [8, Theorem 4.2, Theorem
4.3], so we avoid them for briefness.
41 Page 22 of 28 A. Fonda and W. Ullah NoDEA

5. Some possible applications

As an example of application of Theorem 1.1, we consider the following system
for L = M = 1:

ẋ = f (y) + E(t), ẏ = −A sin x − ∂x P (t, x, u),
(5.1)
u̇ = |v|q−2 v, v̇ = −μ(u+ )p−1 + ν(u− )p−1 + ∂u P (t, x, u),

with the Neumann-type boundary conditions


y(a) = 0 = y(b),
(5.2)
v(a) = 0 = v(b).

Here we use the notation u+ = max{u, 0}, u− = max{−u, 0}. We assume

that the constants A, μ, ν are positive, and the functions f : R → R, E :
[a, b] → R and P : [a, b] × R2 → R are continuous. Assume further that
P (t, x, u) is 2π-periodic in x, continuously differentiable in (x, u), and that it
has a bounded gradient with respect to (x, u). Since sin x and ∂x P (t, x, u) are
bounded, assumption A3 clearly holds.
On the other hand, notice that the last two equations in system (5.1)
correspond to the scalar equation
d
|u̇|p−2 u̇ + μ(u+ )p−1 − ν(u− )p−1 = ∂u P (t, x, u).
dt
If we define H by
|v|q 1
H (u, v) = + μ(u+ )p + ν(u− )p ,
q p
then H is positively-(p, q)-homogeneous and positive, and all the solutions of
system J ẇ = ∇H (w) with w = (u, v) are periodic with the same minimal
period
τ = πp (μ−1/p + ν −1/p ), (5.3)
(see [12,17]), where
2(p − 1)1/p
πp = π.
p sin(π/p)
We thus get the following immediate consequence of Theorem 1.1.

Corollary 5.1. In the above setting, assume moreover that

(μν)1/p nπp
= , for every n ∈ N.
μ1/p + ν 1/p 2(b − a)
Then problem (5.1)–(5.2) has at least two geometrically distinct solutions.

Remark 5.2. Surprisingly enough, besides continuity, in the above corollary

no further assumption is needed on the function f .
NoDEA Boundary value problems associated Page 23 of 28 41

Concerning the periodic problem, as a first example of application of

Theorem 4.1 we consider the system

ẍ + A sin x = e(t) + ∂x P (t, x, u),
(5.4)
dt |u̇|
d p−2
u̇ + μ(u+ )p−1 − ν(u− )p−1 = ∂u P (t, x, u),
where the constants A, μ, ν are positive. Assume that P (t, x, u) is T -periodic
in t and 2π-periodic in0 x, and that it has a bounded gradient with respect to
t
(x, u). Setting E(t) = 0 e(s) ds, system (5.4) is equivalent to

ẋ = y + E(t), ẏ = −A sin x + ∂x P (t, x, u),
(5.5)
u̇ = |v|q−2 v, v̇ = −μ(u+ )p−1 + ν(u− )p−1 + ∂u P (t, x, u).
Assuming e(t) to be T -periodic with
 T
e(t) dt = 0,
0
the function E(t) is T -periodic, as well.
Let us verify that the first two equations in (5.5) satisfy the twist condi-
tion B3 , with M = 1. Notice that there exists K3 > 0 such that, for every
C 1 -function U : [0, T ] → R, all the solutions (x, y) of the system
ẋ = y + E(t), ẏ = −A sin x + ∂x P (t, x, U(t))
are defined on [0, T ] and satisfy
|ẏ(t)| ≤ K3 , for every t ∈ [0, T ].
Define d = K3 T + E∞ + 1 and c = −(K3 T + E∞ + 1). Then, if y(0) = d,
we have
 t
ẋ(t) = y(t) + E(t) = y(0) + ẏ(s) ds + E(t) ≥ d − K3 T − E∞ > 0,
0
for every t ∈ [0, T ], and so x(T ) − x(0) > 0. Similarly, if y(0) = c, then
x(T ) − x(0) < 0, which shows that the twist condition is satisfied.
As a consequence of Theorem 4.1 we then immediately have the following.
Corollary 5.3. In the above setting, assume moreover that
(μν)1/p nπp
= , for every n ∈ N.
μ1/p
+ ν 1/p T
Then system (5.4) has at least two geometrically distinct T -periodic solutions.
A variant of the previous example is provided by the system
⎧
⎪
⎨ẍ + A sin x = e(t) + ∂x P (t, x, u),
u̇ = ν(v − )q−1 − μ(v + )q−1 , (5.6)
⎪
⎩
v̇ = μ(u+ )p−1 − ν(u− )p−1 − ∂u P (t, x, u).
where, being w = (u, v), one has w+ = (u+ , v + ) and w− = (u− , v − ). Assuming
μ, ν to be positive, if we define H by
1 1
H (u, v) = μ(v + )q + ν(v − )q + μ(u+ )p + ν(u− )p ,
q p
41 Page 24 of 28 A. Fonda and W. Ullah NoDEA

then H is positively-(p, q)-homogeneous and positive, and all the solutions of

system J ẇ = ∇H (w) with w = (u, v) are periodic having the same minimal
period τ , which can be computed as follows.
We first consider the dynamics in the first quadrant, i.e., when u > 0 and
v > 0. In this case we can write J ẇ = ∇H (w) as
u̇ = μv q−1 , v̇ = −μup−1 ,
leading to the equation
d
|u̇|p−2 u̇ + μp up−1 = 0.
dt
Then, recalling (5.3), the time needed to pass from the positive v-axis to the
positive u-axis is
1 1 πp
τ1 = πp 2(μp )− p = .
4 2μ
Similarly, in the fourth quadrant, where u > 0 and v < 0, the system becomes
u̇ = −ν|v|q−2 v, v̇ = μup−1 ,
leading to the equation
d
|u̇|p−2 u̇ + μν p−1 up−1 = 0.
dt
So, the time needed to pass from the positive u-axis to the negative v-axis is
1 1 πp
τ2 = πp 2(μν p−1 )− p = 1 1 .
4 2μ p ν q
In a similar way, we obtain that the time needed to pass from the negative
v-axis to the negative u-axis is
πp
τ3 =
2ν
and the time needed to pass from the negative u-axis to the positive v-axis is
πp
τ4 = 1 1 .
2μ q ν p
Hence,
 
πp 1 1 1 1
τ = τ1 + τ 2 + τ 3 + τ 4 = + + 1 1 + 1 1 .
2 μ ν μp ν q μq ν p
We thus get the following consequence of Theorem 4.1.
Corollary 5.4. In the above setting, assume moreover that
 
πp 1 1 1 1 T
+ + 1 1 + 1 1 = , for every n ∈ N \ {0}.
2 μ ν μ ν
p q μ ν
q p n
Then system (5.6) has at least two geometrically distinct T -periodic solutions.
Both Corollaries 5.3 and 5.4 generalize a classical theorem of Mawhin
and Willem [14] on the multiplicity of periodic solutions for the pendulum
equation.
NoDEA Boundary value problems associated Page 25 of 28 41

6. Final remarks
In Theorem 2.1, dealing with the Neumann problem, we have only considered
a diagonal matrix A like in (2.2). However, for the T -periodic problem, the
first author with Gidoni in [4] were able to deal with any symmetric matrix A,
provided that the nonresonance condition σ(JA) ∩ 2π T iZ = ∅ is assumed. We
are confident that a similar result should also hold for the Neumann problem,
but we have been able to prove it only when L = 1 and the matrix has a
positive determinant. Here is our result.
Theorem 6.1. Assume L = 1 and that A1 – A3 hold true. Let A be a symmetric
2 × 2 matrix such that det A > 0. If the non-resonance condition σ(JA) ∩
b−a iZ = ∅ holds, then there are at least M + 1 geometrically distinct solutions
π

of the boundary value problem (2.1)–(2.3).

Proof. Consider the planar Hamiltonian system
J ẇ = Aw. (6.1)
We can diagonalize A by a symplectic transformation. Indeed, there exist a
matrix U with det U = 1 and a diagonal matrix D such that
A = U−1 DU.
Since det U = 1, and the dimension is 2, the change of variables  = Uw
is symplectic. Hence, system (6.1) is transformed into the new Hamiltonian
system
J ˙ = D, (6.2)
with
 
α 0
D= ,
0 β
for some α, β such that αβ > 0. Now, the symplectic change of variables
 = M, with
1 α 2
4
β 0
M=  .
4 β
0 α

transforms system (6.2) into

J
˙ = λ,
√
with λ = ± αβ, according to the signs of α and β. However, if λ < 0, a
final change of variables t → −t will lead to a positive λ. We can now apply
Theorem 2.1 to conclude the proof. 
The general case L ≥ 2 remains an open problem. However, writing
w = (w1 , . . . , wL ) with wj = (uj , vj ) and applying componentwise the same
procedure in the proof of Theorem 6.1 above, Theorem 2.1 can be generalized
to the case when the last equation in system (2.1) is of the type
J ẇj = Aj wj + ∇wj P (t, x, y, w), j = 1, . . . , L,
41 Page 26 of 28 A. Fonda and W. Ullah NoDEA

where Aj is a symmetric 2 × 2 matrix such that det Aj > 0. We omit the

details, for briefness.
As a final remark, we recall that, for the periodic problem, Chen and Qian
in [2] proved a multiplicity result, coupling resonant linear components with
twisting components by using Ahmad-Lazer-Paul type resonance condition. In
our case, a similar result can be expected for Neumann problem without any
twist assumption. The problem remains open for further investigation.

Acknowledgements
We are very grateful to Andrea Sfecci for his insightful feedback and discus-
sions.

Author contributions All authors contributed in an equal way to the manu-

script.

Funding Open access funding provided by Università degli Studi di Trieste

within the CRUI-CARE Agreement.

Declarations

Conflict of interest The authors declare no competing interests.

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

Publisher’s Note Springer Nature remains neutral with regard to jurisdic-

tional claims in published maps and institutional affiliations.

References

[1] Boscaggin, A., Garrione, M.: Planar Hamiltonian systems at resonance: the
Ahmad–Lazer–Paul condition. Nonlinear Differ. Equ. Appl. 20, 825–843 (2013)

[2] Chen, F., Qian, D.: An extension of the Poincaré–Birkhoff Theorem for Hamil-
tonian systems coupling resonant linear components with twisting components.
J. Differ. Equ. 321, 415–448 (2022)
NoDEA Boundary value problems associated Page 27 of 28 41

[3] Conley, C.C., Zehnder, E.J.: The Birkhoff–Lewis fixed point theorem and a con-
jecture of V.I. Arnold. Invent. Math. 73, 33–49 (1983)

[4] Fonda, A., Gidoni, P.: Coupling linearity and twist: an extension of the Poincaré–
Birkhoff Theorem for Hamiltonian systems. Nonlinear Differ. Equ. Appl. 27, 1–26
(2020)

[5] Fonda, A., Klun, G., Obersnel, F., Sfecci, A.: On the Dirichlet problem associ-
ated with bounded perturbations of positively-(p, q)-homogeneous Hamiltonian
systems. J. Fixed Point Theory Appl. 24, 66 (2022)

[6] Fonda, A., Mamo, N.G., Obersnel, F., Sfecci, A.: Multiplicity results for Hamil-
tonian systems with Neumann-type boundary conditions. Nonlinear Differ. Equ.
Appl. 31, Paper No. 31 (2024)

[7] Fonda, A., Ortega, R.: A two-point boundary value problem associated with
Hamiltonian systems on a cylinder. Rend. Circ. Mat. Palermo 72, 3931–3947
(2023).

[8] Fonda, A., Ullah, W.: Periodic solutions of Hamiltonian systems coupling twist
with an isochronous center. Differ. Integral Equ. 37, 323–336 (2024).

[9] Fonda, A., Ureña, A.J.: A higher dimensional Poincaré–Birkhoff theorem for
Hamiltonian flows. Ann. Inst. H. Poincaré Anal. Non Linéaire 34, 679–698 (2017)

[10] Fournier, G., Lupo, D., Ramos, M., Willem, M.: Limit relative category and
critical point theory. In: Dynamics Reported. Expositions in Dynamical Systems,
vol. 3, pp. 1–24. Springer, Berlin (1994)

[11] Jiang, M.Y.: A Landesman–Lazer type theorem for periodic solutions of the
resonant asymmetric p-Laplacian equation. Acta Math. Sin. (Engl. Ser.) 21,
1219–1228 (2005)

[12] Lindqvist, P.: Some remarkable sine and cosine functions. Ricerche Mat. 44,
269–290 (1995)

[13] Liu, J.Q.: A generalized saddle point theorem. J. Differ. Equ. 82, 372–385 (1989)

[14] Mawhin, J., Willem, M.: Multiple solutions of the periodic boundary value prob-
lem for some forced pendulum-type equations. J. Differ. Equ. 52, 264–287 (1984)

[15] Poincaré, H.: Sur un théorème de géométrie. Rend. Circ. Mat. Palermo 33, 375–
407 (1912)

[16] Rebelo, C.: A note on uniqueness of Cauchy problems associated to planar

Hamiltonian systems. Portugal. Math. 57, 415–419 (2000)

[17] Yang, X.: Existence of periodic solutions of a class of planar systems. Z. Anal.
Anwendungen 25(2), 237–248 (2006)

[18] Szulkin, A.: A relative category and applications to critical point theory for
strongly indefinite functionals. Nonlinear Anal. 15, 725–739 (1990)
41 Page 28 of 28 A. Fonda and W. Ullah NoDEA

Alessandro Fonda and Wahid Ullah

Dipartimento di Matematica e Geoscienze
Università degli Studi di Trieste
P.le Europa 1
34127 Trieste
Italy
e-mail: [email protected]

Wahid Ullah
e-mail: [email protected]

Received: 20 September 2023.

Revised: 20 September 2023.
Accepted: 31 January 2024.