3rd orderCanonical form of nonlinear dynamical system

94 Views Asked by At

I am have been solving this problem since a month. I solved a more difficult ones but I do not know why I stuck at this one. There is a clue I can not understand. I solved the first point and stuck with the second one.

The problem statement mentions **Theorem 1.2 ** in Chapter 3, Section 3.1 in the book Local Bifurcations, Center Manifolds, and Normal Forms in Infinite-Dimensional Dynamical Systems

Let $f: \mathbb{R}^n \to \mathbb{R}^n $ be a smooth vector field and let $T: \mathbb{R}^n \to \mathbb{R}^n $ be a isometric linear transform. $f$ is called to be commutative with $T$ if $\forall x \in \mathbb{R}^n $ we have $f(Tx) = Tf(x)$.

  1. Show that if $f$ is commutative with $T$ then for any solution for the ODE $\dot{x} = f(x), \, x(t) \in \mathbb{R}^n $, we have $y(t) := T x(t)$ also solve the ODE.

  2. Show that if $f$ is commutative with $T$, then you can choose polynomial $\Phi$ and $N$ in Theorem 1.2 such that $\Phi$ and $N$ are commutative with $T$.

  3. Let $\omega >0$. Consider the ODE with respect to $z = (z_1,z_2) \in \mathbb{C}^2$

\begin{align} \label{complex-sys-1} \dot{z_1} & = i \omega z_1 + f_1(z_1,\overline{z_1},z_2,\overline{z_2}) \\ \dot{z_2} & = i \omega z_2 + f_2(z_1,\overline{z_1},z_2,\overline{z_2}),\nonumber \end{align} where $f_1,f_2$ are complex-valued functions and the right hand side of (\ref{complex-sys-1}) is commutative with $R$ and $T_\theta \,\, (\forall \theta \in \mathbb{R})$ wiht

\begin{align*} R(z_1,z_2) &= (z_2,z_1)\\ T_\theta (z_1,z_2) &= (e^{i \theta} z_1, e^{-i \theta} z_2), \,\, \theta \in \mathbb{R}. \end{align*}

Then, what is the 3-order canonical form? And what is the higher order canonical form?

[(1)]- Let $f$ is commutative with $T$, and let $s(t) \in \mathbb{R}^n $ be a solution for $\dot{x}=f(x)$, i.e. $\dot{s}(t)=f(s(t)),\, \forall x \in \mathbb{R}^n ,\,t >0$. Let $t>0$ and let $x \in \mathbb{R}^n $. Differentiating the given $ y(t) = T x(t)$ w.r.t time we get; \begin{align*} \dot{y}(t) & = T \dot{x}(t)\\ &= Tf(x(t))\\ & = f(Tx(t))\\ & = f(y(t)).\\ & = f(Tx(t)). \end{align*} Thus, $\dot{y} = f(y)$ is a solution for the system $\dot{x}=f(x)$.

[(2)-] According to Theorem 1.2, there is a polynomial $\phi: \mathbb{R}^n \to \mathbb{R}^n $ of degree $2 \geq p$ with

$$\phi(0)=0,\,\,\, \text{ and }\,\,\, D\phi(0)=0.$$

2

There are 2 best solutions below

0
On BEST ANSWER

First, let us define the operator $T^*$ on $H_k$;

$$T^* h_k (x) = T^{-1} h_k (Tx),$$

Now, knowing that $L_{A^{(k)}(h_k (y))} = - (Dh_k (y)) A_y -A h_k (y)$. We show that $T^*$ commutes with $L^{(k)}_A$ as the following;

\begin{align} L_A^{(k)} (T^* h_k (x)) & = L_A (T^{-1} h_k (Tx))\\ &= A T^{-1} h_k (Tx) - D(T^{-1} h_k (Tx)) A x\\ & =A T^{-1} h_k (Tx) - T^{-1} Dh_k (Tx) T Ax\\ &= T^{-1} (Ah_k (Tx)) - T^{-1} D h_k (Tx) A Tx\\ &= T^{-1} (Ah_k (Tx) - Dh_k (Tx)) A Tx\\ &= T^{-1} L^{(k)} (h_k(Tx))\\ &= T^{*} L^{(k)}_A (h_k (x)). \end{align}

From above we can see that the image and the kernal of the operator $L^{(k)}_A$ are invariant under the influence of $T^*$. Now, we will show that the image and the kernal of the operator $L^{(k)_{A^*}}$ are invariant under $T^*$.

\begin{align} L_{A^*}^{(k)} (T^* h_k (x)) &= L^{(k)}_{A^*} (T^{-1} h_k (Tx))\\ &= A^* T^{-1} h_k (Tx) - D (T^{-1} h_k (Tx)) A^* x\\ &= A^* T^{-1} h_k (Tx) - T^{-1} Dh_k (Tx) T A^* x\\ &= T^{-1} A^* h_k (Tx) - T^{-1} Dh_k (Tx) A^* Tx\\ &= T^{-1} L^{(k)}_{A^*} (h_k (Tx))\\ &= T^* L_{A^*}^{(k)} (h_k (x)). \end{align}

Using the above facts we have

\begin{align*} \dot{x} &= T^{-1} A Tx + T^{-1} F_2(Tx) + T^{-1} F_3(Tx)+\dots\\ &= Ax + F_2(x) + F_3 (x) + \dots. \end{align*}

Considering simplifying the second order terms by making the change of variables

$$x \mapsto y + h_2(y),$$

gives

$$\dot{x} = A x + L^{(k)}_A (h_2 (x)) + F_2 (x) + O(3).$$

Now we let $y \mapsto Ty$ we get

$$\dot{y} = T^{-1} A T y + T^{-1} L_A^{(2)} (h_2 (Ty)) + T^{(k)} P_{Im (L^{(2)}_A)} + T^{-1} P_{\ker L^{(2)}_{A^*}} F_2 (Ty)+ O(3),$$

where $P_{Im(L^{(2)}_{A^{*}})}$ and $P_{\ker L^{(2)}_{A^*}}$ are the projections onto $Im L_A^{(k)}$ and $ \ker L_{A^*}^{(2)} $ respectively. Since $T^*$ commutes with the projection operators, the we can write

$$\dot{y} = Ay + T^* \left( L^{(k)_A} (h_2 (x)) + P_{Im (L_A^{(k)})} F_2(y) \right) + P_{\ker L_{A^*}^{(2)}} T^{-1} F_2 (Ty) + O(3).$$

Since $h_2 (y)$ can be chosen such that

$$L_A^{(2)} (h_2(y)) + P_{\ker L_{A^*}^{(2)}}=0,$$ it follows that

$$\dot{y} = A y + P_{\ker L_{A^*}^{(2)}} T^{-1} F_2 (T x) + O(3).$$

We can do the same steps for higher orders too.

[(3)-] Considering the system

\begin{align} \dot{z_1} & = i \omega z_1 + f_1(z_1,\overline{z_1},z_2,\overline{z_2}) \\ \dot{z_2} & = i \omega z_2 + f_2(z_1,\overline{z_1},z_2,\overline{z_2}),\nonumber \end{align}

Let's define the change of variable

$$ \begin{pmatrix} z_1 \\ z_2 \end{pmatrix} = \begin{pmatrix} 1 &&0 \\ 0&&-1 \end{pmatrix} \begin{pmatrix} z\\ \overline{z}\\ \end{pmatrix}$$

$$ \begin{pmatrix} z \\ \overline{z} \end{pmatrix} = \begin{pmatrix} 1 &&0 \\ 0&&-1 \end{pmatrix} \begin{pmatrix} z_1\\ z_2\\ \end{pmatrix}$$

$$ \begin{pmatrix} \dot{z} \\ \overline{\dot{z}} \end{pmatrix} = \begin{pmatrix} 1 &&0 \\ 0&&-1 \end{pmatrix} \begin{pmatrix} \dot{z_1}\\ \dot{z_2}\\ \end{pmatrix}$$

$$ \begin{pmatrix} \dot{z} \\ \overline{\dot{z}} \end{pmatrix} = i \omega \begin{pmatrix} 1 &&0 \\ 0&&-1 \end{pmatrix} \begin{pmatrix} z_1\\ z_2 \end{pmatrix} + \begin{pmatrix} 1 &&0 \\ 0&&-1 \end{pmatrix} \begin{pmatrix} f_1(z_1,\overline{z_1},z_2,\overline{z_2})\\ f_2(z_1,\overline{z_1},z_2,\overline{z_2}) \end{pmatrix}$$

\begin{align*} \dot{z}&=i \omega z + F_1(z,\overline{z})\\ \overline{\dot{z}}&=- i \omega \overline{z} + F_2(z, \overline{z}). \end{align*}

From theorem 5.1, and considering $Q: \mathbb{R} \to \mathbb{C}$ such the standard form

\begin{align*} \dot{z}&=i \omega z +zQ (|z|^2) + r(z,\overline{z})\\ \overline{\dot{z}}&=- i \omega \overline{z} + \overline{z} \overline{Q (|z|^2)} + \overline{r(z, \overline{z})}. \end{align*}

1
On

This is my solution of point two, I hope it is right.

Consider the system

\begin{equation} \tag{1} \dot{x} = Ax + f(x). \end{equation} where $f(x)$ commutes with $T$. Having (1) satisfies the hypotheses of Theorem 1.2 then for any $2 \leq p \leq k$, there is a polynomial $h_p:\mathbb{R}^N \to \mathbb{R}^N$ of degree $p$ with

$$h_p(0)=0,\,\,\,\, \text{ and } \,\,\,\, Dh_p(0)=0.$$

and such that the change of variable $$x = y + h_p(y), \,\,\,\, \text{ where }\, y=Tx$$

defined on a neighbourhood of the origin $U_p \subset \mathbb{R}^N$ transforms (1) into the normal form

\begin{equation} \label{nform1} \dot{y} = Ay + g_p(y) + r_p(y) \end{equation} Now, we apply the theorem again on the system

$$\dot{y}= Ay + f(y).$$

where $y- Tx$. For any $2 \leq q \leq k$, there is a polynomial $h_q:\mathbb{R}^N \to \mathbb{R}^N$ of degree $p$ with

$$h_q(0)=0,\,\,\,\, \text{ and } \,\,\,\, Dh_q(0)=0.$$

and such that the change of variable $$T^{-1}x = x + h_q(x), \,\,\,\, \text{ where }\, y=Tx$$

defined on a neighbourhood of the origin $U_p \subset \mathbb{R}^N$ transforms (2) into the normal form

\begin{equation} \tag{2} \dot{x} = Ax + g_q(x) + r_q(x). \end{equation} Choosing the neighbourhoods small enough such that $p=q$ and denote $h=h_p=h_q$ and $g_p = g_q=g$. From $x= y+ h(y)$ and $ x =T^{-1}(Tx) = Tx + T h(x)$ we have

$$Th(x) = h(Tx).$$

\begin{align*} \dot{y} & = Ay + g(y) +r(y)\\ T \dot{x}&= T(Ax) + T g(x) + T r(x). \end{align*}

Then we get

$$Tg(x) = g(Tx).$$