Journal Details
Format
Journal
eISSN
2444-8656
First Published
01 Jan 2016
Publication timeframe
2 times per year
Languages
English
Open Access

# Limit cycles of a generalised Mathieu differential system

###### Accepted: 24 Sep 2021
Journal Details
Format
Journal
eISSN
2444-8656
First Published
01 Jan 2016
Publication timeframe
2 times per year
Languages
English
Introduction and statement of the main results

A limit cycle of a differential system is a periodic orbit having a neighbourhood where it is the unique periodic orbit of the differential system. The notion of limit cycle was introduced in 1881 by Poincaré [10].

The study of the existence and number of limit cycles that a differential system in ℝ2 can exhibit is one of the more difficult problems in the qualitative theory of the differential system in the plane. Thus in 1900, Hilbert [6] presented a list of 23 problems at the International Conference of Mathematicians in Paris; most of these problems were solved partially or completely, but the second part of the 16th problem remains unsolved till today. This problem ask about the existence of an upper bound for the maximal number of limit cycles that polynomial differential systems in ℝ2 of a given degree can exhibit.

A source of producing limit cycles is by perturbing the periodic orbits of a centre; see for instance the papers [3, 11] and the book of Christhoper and Li [5], and the hundreds of references quoted there.

The classical Mathieu's differential equation [9] is $x¨+b(1+cosθ)x=0,$ \ddot x + b(1 + \cos \theta )x = 0, where b is real parameter, and the dots denote second derivative with respect to time t. This equation was first discussed in 1868 by Mathieu while studying the problem of vibrations on an elliptical drumhead. Matthieu's equation has many applications in engineering [12, 13] and also in theoretical and experimental physics [2, 7, 14]. Information on its periodic orbits can be found in [16].

Mathieu's equation can be written as the differential system $x˙=y, y˙=−b(1+cosθ)x,$ \dot x = y,\quad \dot y = - b(1 + \cos \theta )x,

In [4], Chen and Llibre studied the limit cycles of the differential system $x˙=y, y˙=−x−ε(1+cosmθ)Q(x, y),$ \dot x = y,\quad \dot y = - x - \varepsilon \left( {1 + \mathop {\cos }\nolimits^m \theta } \right)Q(x,{\kern 1pt} y), where ɛ > 0 is a small parameter and Q(x, y) is an arbitrary polynomial of degree n.

In the present work we study the limit cycles of the following generalisation of differential system (2) as $x˙=y−ε(1+coslθ)P(x, y), y˙=−x−ε(1+cosmθ)Q(x, y),$ \dot x = y - \varepsilon (1 + \mathop {\cos }\nolimits^l \theta )P(x,{\kern 1pt} y),\quad \dot y = - x - \varepsilon \left( {1 + \mathop {\cos }\nolimits^m \theta } \right)Q(x,{\kern 1pt} y), where ɛ > 0 is a small parameter, l and m are positive integers, and P(x, y) and Q(x, y) are arbitrary polynomials of degree n. More precisely, we study the maximum number of limit cycles which bifurcate from the periodic orbits of the linear centre $x˙=y$ \dot x = y , $y˙=−x$ \dot y = - x , when it is perturbed in the form (3).

Our main result is as follows.

Theorem 1

Using the averaging theory of first-order the maximum number of limit cycles of the differential system (3) bifurcating from the periodic solutions of the linear centre $x˙=y$ \dot x = y , $y˙=−x$ \dot y = - x is at most:

n if n is even, and l and m are not both even;

n/2 if n,l and m are even;

n if n is odd and l and m are one odd and the other even;

(n − 1)/2 if n is odd, and l and m are even;

n if n, l and m are odd.

Theorem 1 is proved in section 3.

In section 2, we present a summary of the averaging theory of first-order and of the Descartes theorem that we shall need for proving Theorem 1.

The averaging theory of first-order and the Descartes theorem
Averaging theory of first-order

In these subsection we summarise the result stated in Theorems 11.5 of the book of Verhulst [15] on the averaging theory. For a general introduction to the computation of periodic orbits using the averaging theory see book [8].

Consider the periodic differential system $dxdθ=X(θ, x)=εℱ(θ, x)+ε2Φ(θ, x, ε),$ {{dx} \over {d\theta }} = {\cal X}(\theta ,{\kern 1pt} x) = \varepsilon {\cal F}(\theta ,{\kern 1pt} x) + {\varepsilon ^2}\Phi (\theta ,{\kern 1pt} x,{\kern 1pt} \varepsilon ), where ɛ is a small parameter, x ∈ ℝ, θ ∈ 𝕊1 = ℝ/(2πℤ) and ℱ : 𝕊1 × D → ℝ, Φ : 𝕊1 × D × (−ɛ0, ɛ0) → ℝ are C2 functions, being D an open interval of ℝ and ℱ and Φ are periodic with period 2π in the variable θ.

Now we consider the averaging function f : D → ℝ defined by $f(x)=12π∫02πℱ(θ, x)dθ.$ f(x) = {1 \over {2\pi }}\int_0^{2\pi } {\cal F}(\theta ,{\kern 1pt} x)d\theta . It is known that if x (θ, x0) is the solution of system (4) such that x (0, x0) = x0, then we have $x(2π, x0)−x0=εf(x0)+O(ε2).$ x\left( {2\pi ,{\kern 1pt} {x_0}} \right) - {x_0} = \varepsilon f\left( {{x_0}} \right) + O\left( {{\varepsilon ^2}} \right). So for ɛ > 0 sufficiently small the simple zeros of the averaged function f (x) provide limit cycles of differential equation (4).

The Descartes theorem

In order to study the real zeros of the function f (x) we shall use the Descartes theorem (for a proof see [1]).

Theorem 2

(Descartes Theorem). Consider the real polynomial p(ρ) = ai1ρi1 + a i2ρi2 + ⋯ + ainρin with 0 ≤ i1i2 ≤ ⋯ ≤ in and aij ≠ 0 real constants for j ∈ {1,2,...,n}. When aijaij+1 < 0, we say that aij and aij+1 have a variation of sign. If the number of variations of signs in the polynomial p(ρ) is m, then p(ρ) has at most m positive real roots. Moreover, it is always possible to choose the coefficients of p(ρ) in such a way that p(ρ) has exactly n − 1 positive real roots.

Proof of Theorem 1

Let the polynomials $P(x, y)=∑i+j=0naijxiyj$ P(x,{\kern 1pt} y) = \sum\nolimits_{i + j = 0}^n {a_{ij}}{x_i}{y_j} and $Q(x, y)=∑i+j=0nbijxiyj$ Q(x,{\kern 1pt} y) = \sum\nolimits_{i + j = 0}^n {b_{ij}}{x_i}{y_j} be.

We write the differential system (3) in polar coordinates (ρ, θ) defined by x = ρ cos θ and y = ρ sin θ, with ρ > 0, we obtain $ρ˙=−ε(cosθ(1+cosLθ)P(ρcosθ, ρsinθ)+sinθ(1+cosmθ)Q(ρcosθ, ρsinθ)),θ˙=−1+ερ(sinθ(1+cosLθ)P(ρcosθ, ρsinθ)−cosθ(1+cosmθ)Q(ρcosθ, ρsinθ)).$ \matrix{ {\dot \rho } \hfill & { = - \varepsilon \left( {\cos \theta \left( {1 + \mathop {\cos }\nolimits^L \theta } \right)P(\rho \cos \theta ,{\kern 1pt} \rho \sin \theta ) + \sin \theta \left( {1 + \mathop {\cos }\nolimits^m \theta } \right)Q(\rho \cos \theta ,{\kern 1pt} \rho \sin \theta )} \right),} \hfill \cr {\dot \theta } \hfill & { = - 1 + {\varepsilon \over \rho }\left( {\sin \theta \left( {1 + \mathop {\cos }\nolimits^L \theta } \right)P(\rho \cos \theta ,{\kern 1pt} \rho \sin \theta ) - \cos \theta \left( {1 + \mathop {\cos }\nolimits^m \theta } \right)Q(\rho \cos \theta ,{\kern 1pt} \rho \sin \theta )} \right).} \hfill \cr } Taking in the differential system (7) the variable θ as the new independent variable, system (7) reduces to the differential equation $dρdθ=ε(cosθ(1+cosLθ)P(ρcosθ, ρsinθ)+sinθ(1+cosmθ)Q(ρcosθ, ρsinθ))+O(ε2)=ε∑i+j=0ncosiθsinjθ(aij(cosθ+cosL+1θ)+bij(sinθ+sinθcosmθ))ρi+j+O(ε2)=εℱ(θ, x)+O(ε2).$ \matrix{ {{{d\rho } \over {d\theta }}} \hfill & { = \varepsilon \left( {\cos \theta \left( {1 + \mathop {\cos }\nolimits^L \theta } \right)P(\rho \cos \theta ,{\kern 1pt} \rho \sin \theta ) + \sin \theta \left( {1 + \mathop {\cos }\nolimits^m \theta } \right)Q(\rho \cos \theta ,{\kern 1pt} \rho \sin \theta )} \right) + O\left( {{\varepsilon ^2}} \right)} \hfill \cr {} \hfill & { = \varepsilon \sum\limits_{i + j = 0}^n \mathop {\cos }\nolimits^i \theta \mathop {\sin }\nolimits^j \theta \left( {{a_{ij}}\left( {\cos \theta + \mathop {\cos }\nolimits^{L + 1} \theta } \right) + {b_{ij}}\left( {\sin \theta + \sin \theta \mathop {\cos }\nolimits^m \theta } \right)} \right){\rho ^{i + j}} + O\left( {{\varepsilon ^2}} \right)} \hfill \cr {} \hfill & { = \varepsilon {\cal F}(\theta ,{\kern 1pt} x) + O\left( {{\varepsilon ^2}} \right).} \hfill \cr } Since this differential equation is written in the normal form (4), so we can apply to it the averaging theory of first-order.

In our study we shall use the following formulas for computing the averaged function: $∫02πcospθsin2qθdθ=(2q−1)!!(2q+p)(2q+p−2)…(p+2)∫02πcospθdθ=2παp,2q,∫02πcos2sθdθ=(2s−1)!!2ss!2π=2πβ2s,∫02πcos2s+1θdθ=0,∫02πcospΘsin2q+1θdθ=0.$ \matrix{\hfill {\int_0^{2\pi } \mathop {\cos }\nolimits^p \theta \mathop {\sin }\nolimits^{2q} \theta d\theta } & { = {{(2q - 1)!!} \over {(2q + p)(2q + p - 2) \ldots (p + 2)}}\int_0^{2\pi } \mathop {\cos }\nolimits^p \theta d\theta = 2\pi {\alpha _{p,2q}},} \hfill \cr \hfill {\int_0^{2\pi } \mathop {\cos }\nolimits^{2s} \theta d\theta } & { = {{(2s - 1)!!} \over {{2^s}s!}}2\pi = 2\pi {\beta _{2s}},} \hfill \cr \hfill {\int_0^{2\pi } \mathop {\cos }\nolimits^{2s + 1} \theta d\theta } & { = 0,} \hfill \cr \hfill {\int_0^{2\pi } \mathop {\cos }\nolimits^p \Theta \mathop {\sin }\nolimits^{2q + 1} \theta d\theta } & { = 0.} \hfill \cr } where in the first and second formulas p, q and s are positive integers, in the third one s is a non-negative integer, and in the fourth one p is a positive integer and q is a non-negative integer. For more details of these four integrals see [17, pages 152 and 153].

Proof of Theorem 1 when n is even, and l and m are not both even

We consider the following three cases with n even.

Case 1: m and l are odd. Then we compute the averaged function (5), and we obtain $f(ρ)=12π∫02πℱ(ρ,Θ)dθ=12π∫02π∑i+j=0n(aijcosi+1θsinjθ+aijcosi+l+1θsinjθ +bijcosiθsinj+1θ+bijcosi+mθsinj+1θ)ρi+jdθ=12π∫02π∑i+2q=2n+1((ai,2q−1cosi+1θ+ai,2q−1cosi+l+1θ)sin2q−1θ +(bi,2q−1cosiθ+bi,2q−1cosi+mθ)sin2qθ)ρi+2q−1 +∑i+2q=2n((ai,2qcosi+1(θ)+ai,2qcosi+l+1θ)sin2q(θ) +(bi,2qcosiθ+bi,2qcosi+mθ)sin2q+1θ)ρi+2qdθ=12π[∑2s+1+2q=3n+1∫02πb2s+1,2q−1cos2s+1+mθsin2qθρ2s+2qdθ +∑2s+2q=2n∫02πb2s,2q−1cos2sθsin2qθρ2s+2q−1dθ +∑2s+1+2q=3n+1∫02π(a2s+1,2qcos2s+2θsin2q(θ)ρ2s+1+2qdθ +∑2s+2q=2n∫02πa2s,2qcos2s+l+1θsin2q(θ)ρ2s+2qdθ] +∑s+q=1n/2b2s+1,2q−1α2s+1+m,2qρ2s+2q+∑s+q=1n/2b2s,2q−1α2s,2qρ2s+2q−1 +∑s+q=1n/2a2s+1,2qα2s+2,2qρ2s+1+2q+∑s+q=1n/2a2s,2qα2s+l+1,2qρ2s+2q=∑k=1n+1Akρk.$ \matrix{ {f(\rho )} \hfill & { = {1 \over {2\pi }}\int_0^{2\pi } {\cal F}(\rho ,\Theta )d\theta } \hfill \cr {} \hfill & { = {1 \over {2\pi }}\int_0^{2\pi } \sum\limits_{i + j = 0}^n \left( {{a_{ij}}\mathop {\cos }\nolimits^{i + 1} \theta \mathop {\sin }\nolimits^j \theta + {a_{ij}}\mathop {\cos }\nolimits^{i + l + 1} \theta \mathop {\sin }\nolimits^j \theta } \right.} \hfill \cr {} \hfill & {\quad \left. { + {b_{ij}}\mathop {\cos }\nolimits^i \theta \mathop {\sin }\nolimits^{j + 1} \theta + {b_{ij}}\mathop {\cos }\nolimits^{i + m} \theta \mathop {\sin }\nolimits^{j + 1} \theta } \right){\rho ^{i + j}}d\theta } \hfill \cr {} \hfill & { = {1 \over {2\pi }}\int_0^{2\pi } \sum\limits_{i + 2q = 2}^{n + 1} \left( {\left( {{a_{i,2q - 1}}\mathop {\cos }\nolimits^{i + 1} \theta + {a_{i,2q - 1}}\mathop {\cos }\nolimits^{i + l + 1} \theta } \right)\mathop {\sin }\nolimits^{2q - 1} \theta } \right.\;} \hfill \cr {} \hfill & {\quad \left. { + \left( {{b_{i,2q - 1}}\mathop {\cos }\nolimits^i \theta + {b_{i,2q - 1}}\mathop {\cos }\nolimits^{i + m} \theta } \right)\mathop {\sin }\nolimits^{2q} \theta } \right){\rho ^{i + 2q - 1}}} \hfill \cr {} \hfill & {\quad + \sum\limits_{i + 2q = 2}^n \left( {\left( {{a_{i,2q}}\mathop {\cos }\nolimits^{i + 1} (\theta ) + {a_{i,2q}}\mathop {\cos }\nolimits^{i + l + 1} \theta } \right)\mathop {\sin }\nolimits^{2q} (\theta )} \right.\;} \hfill \cr {} \hfill & {\quad \left. { + \left( {{b_{i,2q}}\mathop {\cos }\nolimits^i \theta + {b_{i,2q}}\mathop {\cos }\nolimits^{i + m} \theta } \right)\mathop {\sin }\nolimits^{2q + 1} \theta } \right){\rho ^{i + 2q}}d\theta } \hfill \cr {} \hfill & { = {1 \over {2\pi }}\left[ {\sum\limits_{2s + 1 + 2q = 3}^{n + 1} \int_0^{2\pi } {b_{2s + 1,2q - 1}}\mathop {\cos }\nolimits^{2s + 1 + m} \theta \mathop {\sin }\nolimits^{2q} \theta {\rho ^{2s + 2q}}d\theta } \right.\;} \hfill \cr {} \hfill & {\quad + \sum\limits_{2s + 2q = 2}^n \int_0^{2\pi } {b_{2s,2q - 1}}\mathop {\cos }\nolimits^{2s} \theta \mathop {\sin }\nolimits^{2q} \theta {\rho ^{2s + 2q - 1}}d\theta } \hfill \cr {} \hfill & {\quad + \sum\limits_{2s + 1 + 2q = 3}^{n + 1} \int_0^{2\pi } ({a_{2s + 1,2q}}\mathop {\cos }\nolimits^{2s + 2} \theta \mathop {\sin }\nolimits^{2q} (\theta ){\rho ^{2s + 1 + 2q}}d\theta } \hfill \cr {} \hfill & {\quad + \left. {\sum\limits_{2s + 2q = 2}^n \int_0^{2\pi } {a_{2s,2q}}\mathop {\cos }\nolimits^{2s + l + 1} \theta \mathop {\sin }\nolimits^{2q} (\theta ){\rho ^{2s + 2q}}d\theta } \right]} \hfill \cr {} \hfill & {\quad + \sum\limits_{s + q = 1}^{n/2} {b_{2s + 1,2q - 1}}{\alpha _{2s + 1 + m,2q}}{\rho ^{2s + 2q}} + \sum\limits_{s + q = 1}^{n/2} {b_{2s,2q - 1}}{\alpha _{2s,2q}}{\rho ^{2s + 2q - 1}}} \hfill \cr {} \hfill & {\quad + \sum\limits_{s + q = 1}^{n/2} {a_{2s + 1,2q}}{\alpha _{2s + 2,2q}}{\rho ^{2s + 1 + 2q}} + \sum\limits_{s + q = 1}^{n/2} {a_{2s,2q}}{\alpha _{2s + l + 1,2q}}{\rho ^{2s + 2q}}} \hfill \cr {} \hfill & { = \sum\limits_{k = 1}^{n + 1} {A_k}{\rho ^k}.} \hfill \cr }

Since f (ρ) is a polynomial generated by a linear combination of the monomials {ρ, ρ2, . . ., ρn+1}. Using Descartes theorem it follows that the polynomial f (ρ) have at most n simple positive zeros, and consequently from subsection 2.1 we get using the averaging theory of first-order that for ɛ > 0 sufficiently small the differential system (3) has at most n limit cycles bifurcating from the periodic solutions of the linear centre (3) with ɛ = 0.

Case 2: m is odd and l is even. Working as in the case 1 we obtain $f(ρ)=12π∫02πℱ(ρ,θ)dθ=∑s+q=1n/2b2s+1,2q−1α2s+1+m,2qρ2s+2q+∑s+q=1n/2b2s,2q−1α2s,2qρ2s+2q−1 +∑s+q=1n/2a2s+1,2qα2s+2,2qρ2s+1+2q+∑s+q=1n/2a2s+1,2qα2s+l+2,2qρ2s+1+2q=∑k=1n+1A˜kρk.$ \matrix{ {f(\rho )} \hfill & { = {1 \over {2\pi }}\int_0^{2\pi } {\cal F}(\rho ,\theta )d\theta } \hfill \cr {} \hfill & { = \sum\limits_{s + q = 1}^{n/2} {b_{2s + 1,2q - 1}}{\alpha _{2s + 1 + m,2q}}{\rho ^{2s + 2q}} + \sum\limits_{s + q = 1}^{n/2} {b_{2s,2q - 1}}{\alpha _{2s,2q}}{\rho ^{2s + 2q - 1}}} \hfill \cr {} \hfill & {\quad + \sum\limits_{s + q = 1}^{n/2} {a_{2s + 1,2q}}{\alpha _{2s + 2,2q}}{\rho ^{2s + 1 + 2q}} + \sum\limits_{s + q = 1}^{n/2} {a_{2s + 1,2q}}{\alpha _{2s + l + 2,2q}}{\rho ^{2s + 1 + 2q}} = \sum\limits_{k = 1}^{n + 1} {{\tilde A}_k}{\rho ^k}.} \hfill \cr }

Again, f (ρ) is a polynomial generated by the monomials {ρ, ρ2, ⋯, ρn+1}, and as in case 1 by Descartes theorem it follows that for ɛ > 0 sufficiently small, the differential system (3) has at most n limit cycles bifurcating from the periodic solutions of the linear centre (3) with ɛ = 0.

Case 3: m is even and l is odd. Then $f(ρ)=12π∫02πℱ(ρ,θ)dθ=∑s+q=1n/2b2s,2q−1α2s+m,2qρ2s+2q−1+∑s+q=1n/2b2s,2q−1α2s,2qρ2s+2q−1 +∑s+q=1n/2a2s+1,2qα2s+2,2qρ2s+1+2q+∑s+q=1n/2a2s,2qα2s+l+1,2qρ2s+2q=∑k=1n+1A^kρk.$ \matrix{ {f(\rho )} \hfill & { = {1 \over {2\pi }}\int_0^{2\pi } {\cal F}(\rho ,\theta )d\theta } \hfill \cr {} \hfill & { = \sum\limits_{s + q = 1}^{n/2} {b_{2s,2q - 1}}{\alpha _{2s + m,2q}}{\rho ^{2s + 2q - 1}} + \sum\limits_{s + q = 1}^{n/2} {b_{2s,2q - 1}}{\alpha _{2s,2q}}{\rho ^{2s + 2q - 1}}} \hfill \cr {} \hfill & {\quad + \sum\limits_{s + q = 1}^{n/2} {a_{2s + 1,2q}}{\alpha _{2s + 2,2q}}{\rho ^{2s + 1 + 2q}} + \sum\limits_{s + q = 1}^{n/2} {a_{2s,2q}}{\alpha _{2s + l + 1,2q}}{\rho ^{2s + 2q}} = \sum\limits_{k = 1}^{n + 1} {{\hat A}_k}{\rho ^k}.} \hfill \cr }

As in the previous two cases, we conclude that for ɛ > 0 sufficiently small the differential system (3) has at most n limit cycles bifurcating from the periodic solutions of the linear centre (3) with ɛ = 0.

Proof of Theorem 1 when n, l and m are even

Working as in case 1, we compute the averaged function and we get $f(ρ)=12π∫02πℱ(ρ,θ)dt=∑s+q=1n/2b2s,2q−1α2s+m,2qρ2s+2q−1+∑s+q=1n/2b2s,2q−1α2s,2qρ2s+2q−1 +∑s+q=1n/2a2s+1,2qα2s+2,2qρ2s+1+2q+∑s+q=1n/2a2s+1,2qα2s+l+2,2qρ2s+1+2q=∑k=1(n/2)+1B˜kρ2k−1.$ \matrix{ {f(\rho )} \hfill & { = {1 \over {2\pi }}\int_0^{2\pi } {\cal F}(\rho ,\theta )dt} \hfill \cr {} \hfill & { = \sum\limits_{s + q = 1}^{n/2} {b_{2s,2q - 1}}{\alpha _{2s + m,2q}}{\rho ^{2s + 2q - 1}} + \sum\limits_{s + q = 1}^{n/2} {b_{2s,2q - 1}}{\alpha _{2s,2q}}{\rho ^{2s + 2q - 1}}} \hfill \cr {} \hfill & {\quad + \sum\limits_{s + q = 1}^{n/2} {a_{2s + 1,2q}}{\alpha _{2s + 2,2q}}{\rho ^{2s + 1 + 2q}} + \sum\limits_{s + q = 1}^{n/2} {a_{2s + 1,2q}}{\alpha _{2s + l + 2,2q}}{\rho ^{2s + 1 + 2q}} = \sum\limits_{k = 1}^{(n/2) + 1} {{\tilde B}_k}{\rho ^{2k - 1}}.} \hfill \cr }

Since f (ρ) is a polynomial generated by a linear combination of the monomials {ρ, ρ3, . . ., ρn+1}. Using Descartes theorem it follows that the polynomial f (ρ) have at most n/2 simple positive zeros, and consequently from subsection 2.1 we get using the averaging theory of first-order that for ɛ > 0 sufficiently small, the differential system (3) has at most n/2 limit cycles bifurcating from the periodic solutions of the linear centre (3) with ɛ = 0.

Proof of Theorem 1 if n, l and m are odd

Again working as in case 1, we compute the averaged function and we obtain $f(ρ)=12π∫02πℱ(ρ,θ)dθ=∑s+q=1(n−1)/2b2s+1,2q−1α2s+1+m,2qρ2s+2q+∑s+q=1(n+1)/2b2s,2q−1α2s,2qρ2s+2q−1 +∑s+q=1(n−1)/2a2s+1,2qα2s+2,2qρ2s+1+2q+∑s+q=1(n+1)/2a2s,2qα2s+l+1,2qρ2s+2q=∑k=1n+1Ckρk.$ \matrix{ {f(\rho )} \hfill & { = {1 \over {2\pi }}\int_0^{2\pi } {\cal F}(\rho ,\theta )d\theta } \hfill \cr {} \hfill & { = \sum\limits_{s + q = 1}^{(n - 1)/2} {b_{2s + 1,2q - 1}}{\alpha _{2s + 1 + m,2q}}{\rho ^{2s + 2q}} + \sum\limits_{s + q = 1}^{(n + 1)/2} {b_{2s,2q - 1}}{\alpha _{2s,2q}}{\rho ^{2s + 2q - 1}}} \hfill \cr {} \hfill & {\quad + \sum\limits_{s + q = 1}^{(n - 1)/2} {a_{2s + 1,2q}}{\alpha _{2s + 2,2q}}{\rho ^{2s + 1 + 2q}} + \sum\limits_{s + q = 1}^{(n + 1)/2} {a_{2s,2q}}{\alpha _{2s + l + 1,2q}}{\rho ^{2s + 2q}} = \sum\limits_{k = 1}^{n + 1} {C_k}{\rho ^k}.} \hfill \cr }

Since the polynomial f (ρ) has the monomials {ρ, ρ2, . . . , ρn+1}. Using Descartes theorem it follows that the polynomial f (ρ) has at most n simple zeros. Therefore for ɛ > 0 sufficiently small the differential system (3) has at most n limit cycles bifurcating from the periodic solutions of the linear centre (3) with ɛ = 0.

Proof of Theorem 1 if n is odd and l and m are one odd and the other even

We distinguish two cases with n odd.

Case 1: m is odd and l is even. Computing the averaged function we get $f(ρ)=12π∫02πℱ(ρ,θ)dθ=∑s+q=1(n−1)/2b2s+1,2q−1α2s+1+m,2qρ2s+2q+∑s+q=1(n+1)/2b2s,2q−1α2s,2qρ2s+2q−1 +∑s+q=1(n−1)/2a2s+1,2qα2s+2,2qρ2s+1+2q+∑s+q=1(n−1)/2a2s+1,2qα2s+l+2,2qρ2s+1+2q=∑k=1nC˜kρk.$ \matrix{ {f(\rho )} \hfill & { = {1 \over {2\pi }}\int_0^{2\pi } {\cal F}(\rho ,\theta )d\theta } \hfill \cr {} \hfill & { = \sum\limits_{s + q = 1}^{(n - 1)/2} {b_{2s + 1,2q - 1}}{\alpha _{2s + 1 + m,2q}}{\rho ^{2s + 2q}} + \sum\limits_{s + q = 1}^{(n + 1)/2} {b_{2s,2q - 1}}{\alpha _{2s,2q}}{\rho ^{2s + 2q - 1}}} \hfill \cr {} \hfill & {\quad + \sum\limits_{s + q = 1}^{(n - 1)/2} {a_{2s + 1,2q}}{\alpha _{2s + 2,2q}}{\rho ^{2s + 1 + 2q}} + \sum\limits_{s + q = 1}^{(n - 1)/2} {a_{2s + 1,2q}}{\alpha _{2s + l + 2,2q}}{\rho ^{2s + 1 + 2q}} = \sum\limits_{k = 1}^n {{\tilde C}_k}{\rho ^k}.} \hfill \cr }

Case 2: m is even and l is odd. Then the averaged function $f(ρ)=12π∫02πℱ(ρ,θ)dθ=∑s+q=1(n+1)/2b2s,2q−1α2s+m,2qρ2s+2q−1+∑s+q=1(n+1)/2b2s,2q−1α2s,2qρ2s+2q−1 +∑s+q=1(n−1)/2a2s+1,2qα2s+2,2qρ2s+1+2q+∑s+q=1(n−1)/2a2s,2qα2s+l+1,2qρ2s+2q=∑k=1nC^kρk.$ \matrix{ {f(\rho )} \hfill & { = {1 \over {2\pi }}\int_0^{2\pi } {\cal F}(\rho ,\theta )d\theta } \hfill \cr {} \hfill & { = \sum\limits_{s + q = 1}^{(n + 1)/2} {b_{2s,2q - 1}}{\alpha _{2s + m,2q}}{\rho ^{2s + 2q - 1}} + \sum\limits_{s + q = 1}^{(n + 1)/2} {b_{2s,2q - 1}}{\alpha _{2s,2q}}{\rho ^{2s + 2q - 1}}} \hfill \cr {} \hfill & {\quad + \sum\limits_{s + q = 1}^{(n - 1)/2} {a_{2s + 1,2q}}{\alpha _{2s + 2,2q}}{\rho ^{2s + 1 + 2q}} + \sum\limits_{s + q = 1}^{(n - 1)/2} {a_{2s,2q}}{\alpha _{2s + l + 1,2q}}{\rho ^{2s + 2q}} = \sum\limits_{k = 1}^n {{\hat C}_k}{\rho ^k}.} \hfill \cr }

Now the polynomials f (ρ) in cases 1 and 2 are generated by the monomials {ρ, ρ2, . . . , ρn}. Using Descartes theorem it follows that the polynomial f (ρ) has at most n − 1 simple zeros. Therefore for ɛ > 0 sufficiently small the differential system (3) has at most n − 1 limit cycles bifurcating from the periodic solutions of the linear centre (3) with ɛ = 0.

Proof of Theorem 1 if n is odd, and l and m are even

Computing the averaged function we obtain $f(ρ)=12π∫02πℱ(ρ,θ)dθ=∑s+q=1(n+1)/2b2s,2q−1α2s+m,2qρ2s+2q−1+∑s+q=1(n+1)/2b2s,2q−1α2s,2qρ2s+2q−1 +∑s+q=1(n−1)/2a2s+1,2qα2s+2,2qρ2s+1+2q+∑s+q=1(n−1)/2a2s+1,2qα2s+l+2,2qρ2s+1+2q=∑k=1(n+1)/2Dkρ2k−1+∑k=1(n−1)/2D^kρ2k+1.$ \matrix{ {f(\rho )} \hfill & { = {1 \over {2\pi }}\int_0^{2\pi } {\cal F}(\rho ,\theta )d\theta } \hfill \cr {} \hfill & { = \sum\limits_{s + q = 1}^{(n + 1)/2} {b_{2s,2q - 1}}{\alpha _{2s + m,2q}}{\rho ^{2s + 2q - 1}} + \sum\limits_{s + q = 1}^{(n + 1)/2} {b_{2s,2q - 1}}{\alpha _{2s,2q}}{\rho ^{2s + 2q - 1}}} \hfill \cr {} \hfill & {\quad + \sum\limits_{s + q = 1}^{(n - 1)/2} {a_{2s + 1,2q}}{\alpha _{2s + 2,2q}}{\rho ^{2s + 1 + 2q}} + \sum\limits_{s + q = 1}^{(n - 1)/2} {a_{2s + 1,2q}}{\alpha _{2s + l + 2,2q}}{\rho ^{2s + 1 + 2q}}} \hfill \cr {} \hfill & { = \sum\limits_{k = 1}^{(n + 1)/2} {D_k}{\rho ^{2k - 1}} + \sum\limits_{k = 1}^{(n - 1)/2} {{\hat D}_k}{\rho ^{2k + 1}}.} \hfill \cr }

The polynomial f (ρ) is generated by the monomials {ρ, ρ3, . . . , ρn}. Using Descartes theorem it follows that the polynomial f (ρ) has at most (n − 1)/2 simple zeros. Therefore for ɛ > 0 sufficiently small the differential system (3) has at most (n − 1)/2 limit cycles bifurcating from the periodic solutions of the linear centre (3) with ɛ = 0.

Proof of Theorem 1 if n, l and m are odd

Again working as in case 1, we compute the averaged function and we obtain $f(ρ)=12π∫02πℱ(ρ,θ)dθ=∑s+q=1(n−1)/2b2s+1,2q−1α2s+1+m,2qρ2s+2q+∑s+q=1(n+1)/2b2s,2q−1α2s,2qρ2s+2q−1 +∑s+q=1(n−1)/2a2s+1,2qα2s+2,2qρ2s+1+2q+∑s+q=1(n+1)/2a2s,2qα2s+l+1,2qρ2s+2q=∑k=1n+1Ckρk.$ \matrix{ {f(\rho )} \hfill & { = {1 \over {2\pi }}\int_0^{2\pi } {\cal F}(\rho ,\theta )d\theta } \hfill \cr {} \hfill & { = \sum\limits_{s + q = 1}^{(n - 1)/2} {b_{2s + 1,2q - 1}}{\alpha _{2s + 1 + m,2q}}{\rho ^{2s + 2q}} + \sum\limits_{s + q = 1}^{(n + 1)/2} {b_{2s,2q - 1}}{\alpha _{2s,2q}}{\rho ^{2s + 2q - 1}}} \hfill \cr {} \hfill & {\quad + \sum\limits_{s + q = 1}^{(n - 1)/2} {a_{2s + 1,2q}}{\alpha _{2s + 2,2q}}{\rho ^{2s + 1 + 2q}} + \sum\limits_{s + q = 1}^{(n + 1)/2} {a_{2s,2q}}{\alpha _{2s + l + 1,2q}}{\rho ^{2s + 2q}} = \sum\limits_{k = 1}^{n + 1} {C_k}{\rho ^k}.} \hfill \cr }

Since the polynomial f (ρ) has the monomials {ρ, ρ2, ⋯, ρn+1}. Using Descartes theorem it follows that the polynomial f (ρ) has at most n simple zeros. Therefore for ɛ > 0 sufficiently small the differential system (3) has at most n limit cycles bifurcating from the periodic solutions of the linear centre (3) with ɛ = 0.

I.S. Berezin, N.P. Zhidkov, Computing Methods, vol. II, Pergamon Press, Oxford, 1964. BerezinI.S. ZhidkovN.P. Computing Methods II Pergamon Press Oxford 1964 Search in Google Scholar

P. A. Brandão, Stability boundaries of a Mathieu equation having PT symmetry. Phys. Lett. A 383 (2019), no. 25, 3043–3046. BrandãoP. A. Stability boundaries of a Mathieu equation having PT symmetry Phys. Lett. A 383 2019 25 3043 3046 10.1016/j.physleta.2019.07.003 Search in Google Scholar

A. Buică, J. Giné, J. Llibre, Bifurcation of limit cycles from a polynomial degenerate centre, Adv. Nonlinear Stud. 10 (2010), 597–609. BuicăA. GinéJ. LlibreJ. Bifurcation of limit cycles from a polynomial degenerate centre Adv. Nonlinear Stud. 10 2010 597 609 10.1515/ans-2010-0305 Search in Google Scholar

T. Chen, J. Llibre, Limit cycles of a second-order differential equation, Appl. Math. Lett. 88 (2019), 111–117. ChenT. LlibreJ. Limit cycles of a second-order differential equation Appl. Math. Lett. 88 2019 111 117 10.1016/j.aml.2018.08.015 Search in Google Scholar

C. Christopher, C. Li, Limit cycles of differential equations, Advanced Courses in Mathematics. CRM Barcelona. Birkhäuser Verlag, Basel, 2007. ChristopherC. LiC. Limit cycles of differential equations, Advanced Courses in Mathematics. CRM Barcelona Birkhäuser Verlag Basel 2007 Search in Google Scholar

D. Hilbert, Mathematische Probleme, Lecture, Second Internat. Congr. Math. (Paris, 1900), Nachr. Ges Wiss. Gottingen Math. Phys. KL. (1900), 253–297; English transl., Bull. Amer. Math. Soc. 8 (1902), 437–479; Bull. (New Series) Amer. Math. Soc. 37 (2000), 407–436. HilbertD. Mathematische Probleme, Lecture, Second Internat Congr. Math. (Paris, 1900), Nachr. Ges Wiss. Gottingen Math. Phys. KL 1900 253 297 English transl., Bull. Amer. Math. Soc. 8 (1902), 437–479; Bull. (New Series) Amer. Math. Soc. 37 (2000), 407–436. Search in Google Scholar

K. Imaizumi, Exact WKB analysis and TBA equations for the Mathieu equation. Phys. Lett. B 806 (2020), 135500, 7 pp. ImaizumiK. Exact WKB analysis and TBA equations for the Mathieu equation Phys. Lett. B 806 2020 135500 7 pp. 10.1016/j.physletb.2020.135500 Search in Google Scholar

J. Llibre, R. Moeckel, C. Simó, Central Configurations, Periodic Orbits and Hamiltonian Systems, in: Advanced Courses in Mathematics, CRM Barcelona, Birkhäuser, 2015. LlibreJ. MoeckelR. SimóC. Central Configurations, Periodic Orbits and Hamiltonian Systems in: Advanced Courses in Mathematics, CRM Barcelona Birkhäuser 2015 10.1007/978-3-0348-0933-7 Search in Google Scholar

E. Mathieu, Mémoire sur le mouvement vibratoire d’une membrane de forme elliptique, J. Math. Pures Appl. 13 (1868), 137–203. MathieuE. Mémoire sur le mouvement vibratoire d’une membrane de forme elliptique J. Math. Pures Appl. 13 1868 137 203 Search in Google Scholar

H. Poincaré, Mémoire sur les courbes définies par une équation différentielle I, II, J. Math. Pures Appl. 7 (1881), 375–422; 8 (1882), 251–296. PoincaréH. Mémoire sur les courbes définies par une équation différentielle I, II J. Math. Pures Appl. 7 1881 375 422 8 1882 251 296 Search in Google Scholar

L.S. Pontrjagin, Über Autoschwingungssysteme, die den hamiltonschen nahe liegen, Phys. Z. Sowjetunion 6 (1934), 25–28. PontrjaginL.S. Über Autoschwingungssysteme, die den hamiltonschen nahe liegen Phys. Z. Sowjetunion 6 1934 25 28 Search in Google Scholar

L. Ruby, Applications of the Mathieu Equation, Am. J. Phys., 64 (1) (1996), 39–44. RubyL. Applications of the Mathieu Equation Am. J. Phys. 64 1 1996 39 44 10.1119/1.18290 Search in Google Scholar

J. J. Stoker, Nonlinear vibrations in mechanical and electrical systems. Vol. 2. New York: Interscience Publishers, 1950. StokerJ. J. Nonlinear vibrations in mechanical and electrical systems 2 New York Interscience Publishers 1950 Search in Google Scholar

A. C. Tzemos, G. Contopoulos, Integrals of Motion in Time-periodic Hamiltonian Systems: The Case of the Mathieu Equation, Regul. Chaotic Dyn. 26 (2021), no. 1, 89–104. TzemosA. C. ContopoulosG. Integrals of Motion in Time-periodic Hamiltonian Systems: The Case of the Mathieu Equation Regul. Chaotic Dyn. 26 2021 1 89 104 10.1134/S1560354721010056 Search in Google Scholar

F. Verhulst, Nonlinear differential equations and dynamical systems, Universitext, Springer, Berlin, Second Edition, 2000. VerhulstF. Nonlinear differential equations and dynamical systems Universitext, Springer Berlin Second Edition 2000 Search in Google Scholar

M. Zamora, A note on the periodic solutions of a Mathieu-Duffing type equations, Math. Nachr. 290 (2017), no. 7, 1113–1118. ZamoraM. A note on the periodic solutions of a Mathieu-Duffing type equations Math. Nachr. 290 2017 7 1113 1118 10.1002/mana.201400122 Search in Google Scholar

D. Zwillinger, Table of Integrals, Series, and Products, ISBN: 978-0-12-384933-5, 2014. ZwillingerD. Table of Integrals, Series, and Products ISBN: 978-0-12-384933-5, 2014 Search in Google Scholar

Recommended articles from Trend MD