1: \chapter[\protect\uppercase{Bifurcation Analysis of C$_2$H$_2$ Stretch-Bend}]{Bifurcation Analysis of C$_2$H$_2$ Stretch-Bend}\label{ch.ch5}%
2: \addtocontents{toc}{\protect\vspace{0.25in}}%
3:
4: \section{Introduction}
5: \addtocontents{toc}{\protect\vspace*{7pt}}
6:
7: Beyond the pure bending subsystem, the next step is extending the critical point analysis to all 7 vibrational DOF (4 bending and 3 stretching) of C$_2$H$_2$ explicitly. In such a multidimensional system, an analytic detection method would be superior to both visual inspection and numerical search.
8:
9: \subsection{5.1.1 The Effective Hamiltonian}
10: \addtocontents{toc}{\protect\vspace*{5pt}}
11:
12: Three stretch-bend effective Hamiltonians of C$_2$H$_2$ have been published up to date \cite{Herman1995,Herman1999,Field2001}.
13:
14: In this chapter, we use the first Hamiltonian of Herman \et, because none of the other Hamiltonians has been studied theoretically \cite{Gaspard,DerivState}. The Hamiltonian in \cite{Herman1995} contains 8 resonances: one Darling-Dennison type $K_{11/33}$ resonance coupling between the two normal C-H stretch modes, the $K_{44/55}$ (known as DD-I in Chapter 4) between the two normal bending modes, the $\ell$ resonance $r_{45}$, and $K_{3/245}$, $K_{1/244}$, $K_{1/255}$, $K_{14/35}$, $K_{33/1244}$ resonances which couple between the stretch and bend DOF. This coupling structure is illustrated in Fig.~\ref{sbcoupling}. In a typical stretch-bend polyad, the normal ZOS $\vert n_1, n_2, n_3, n_4^{\ell_4},n_5^{\ell_5} \rangle$ are all coupled by a complex web of resonances except for two special cases. The pure bending polyads analyzed in Chapter 4 form an isolated subsystem with polyad number $N_s=0$. The C-C stretch overtones $\vert 0,n_2,0,0^0,0^0 \rangle$ are not coupled by any resonance; therefore they are eigenstates of the Hamiltonian.
15:
16: \newpage \begin{figure}[hbtp]
17: %==\vspace{3.59in}
18: \begin{center}\includegraphics[width=4.756in]{sbcoupling.eps}\end{center}
19: \cpn{Resonance couplings among the stretch-bend normal modes} {Resonance couplings among the stretch-bend normal modes, according to the Hamiltonian of \cite{Herman1995}. On the left are the normal modes of Fig.~\ref{normalmodes}. \label{sbcoupling}}
20: \end{figure} \newpage
21:
22: In producing this Hamiltonians (as well as the other later ones), the fitting is based on only a small part of the states predicted by the polyad model. A high-lying polyad typically contains dozens of states, out of which only a few have been experimentally detected and included in the fit. This allows for some uncertainty in the high-order coefficients in the fit. Especially lacking are states that contain both stretching ($N_s \geq 1$) excitation and high bending excitation ($n_4+n_5 \geq 12$).
23:
24: Using Heisenberg's Correspondence Principle in eqn. (\ref{Heisenberg}), the 7 DOF quantum Hamiltonian in \cite{Herman1995} is transformed to a classical one with 14 action-angle variables: ($\tau_i, \phi_i$) for $i=1-5$ and ($\kappa_j, \chi_j$) for $j=4-5$. The actions are related to the zero-order quantum numbers by
25: \begin{align}
26: \tau_i &=n_i +\frac{1}{2} & \mbox { \,\,\,\,\,\,for\,\,} i&=1, 2, 3 \label{eqn5.1} \\
27: \tau_i &=n_i +1 & \mbox { \,\,\,\,\,\,for\,\,} i&=4, 5 \label{eqn5.2} \\
28: \kappa_j & = \ell_j & \mbox {\,\,\,\,\,\,for\,\,} j&=4,5 \label{eqn5.3}
29: \end{align}
30:
31: To simplify our analysis, the last five high-order parameters in Table~\ref{sbparam}, $y_{244}, K_{33/1244}, k_4, r_{445}, r_{545}$, are ignored in the rest of this Chapter. The classical Hamiltonian \cite{Gaspard} then has the following form:
32: \begin{align}
33: H_{sb} = &H_0+H_v \label{sbhamil}
34: \end{align}
35: with
36: {\small \begin{align}
37: H_0 = \sum_{i=1}^{5} \omega_i \tau_i+\sum_{i,j=1; i\le j}^5 x_{ij}\tau_i \tau_j + \sum_{i,j=1;i \le j}^5 g_{ij}\kappa_i \kappa_j \label{sbH0}\end{align} }
38: {\small \begin{align}
39: H_v= & \frac{K_{1133}}{2} \tau_1 \tau_3 \cos \lbrack 2 (\phi_1- \phi_3) \rbrack \nonumber\\
40: & +\frac{K_{3/245}}{4}\sqrt{\tau_2\tau_3}\{\sqrt{(\tau_4-\kappa_4)(\tau_5+\kappa_5)}\cos[\phi_2-\phi_3+\phi_4-\chi_4+\phi_5+\chi_5] \nonumber\\
41: & \quad +\sqrt{(\tau_4+\kappa_4)(\tau_5-\kappa_5)}\cos[\phi_2-\phi_3+\phi_4+\chi_4+\phi_5-\chi_5]\} \nonumber\\
42: & +\frac{K_{1/244}}{2}\sqrt{\tau_1\tau_2(\tau_4^2-\kappa_4^2)}\cos[\phi_1-\phi_2-2\phi_4] \nonumber\\
43: & +\frac{K_{1/255}}{2}\sqrt{\tau_1\tau_2(\tau_5^2-\kappa_5^2)}\cos[\phi_1-\phi_2-2\phi_5] \nonumber\\
44: & +\frac{K_{14/35}}{2}\sqrt{\tau_1\tau_3}\{\sqrt{(\tau_4-\kappa_4)(\tau_5-\kappa_5)}\cos[\phi_1-\phi_3+\phi_4-\chi_4-\phi_5+\chi_5] \nonumber\\
45: & \quad +\sqrt{(\tau_4+\kappa_4)(\tau_5+\kappa_5)}\cos[\phi_1-\phi_3+\phi_4+\chi_4-\phi_5-\chi_5] \} \nonumber\\
46: & +\frac{K_{44/55}}{2}\sqrt{(\tau_4^2-\kappa_4^2)(\tau_5^2-\kappa_5^2)}\cos[2(\phi_4-\phi_5)] \nonumber\\
47: & +\frac{r_{45}}{2}\sqrt{(\tau_4^2-\kappa_4^2)(\tau_5^2-\kappa_5^2)}\cos[2(\chi_4-\chi_5)] \label{sbclassicalH}
48: \end{align} }
49:
50: The values of the parameters in $H_{sb}$ are listed in Table~\ref{sbparam} \fn{It was found later that the $K_{14/35}$ value appears as $29.044$ in \cite{Herman1995} and as $29.944$ in \cite{Gaspard}. This appears as a minor discrepancy compared to its later revision to $15.66$ in \cite{Herman1999} and then to $16.614$ in \cite{Field2001}.}. The 3 polyad numbers correspond to 3 classical constants of motion:
51: \begin{align}
52: P= & 5 \tau_1 + 3\tau_2 + 5\tau_3 + \tau_4 + \tau_5 = N_t + \frac{15}{2} \label{ClassicalP} \\
53: R= & \tau_1 + \tau_2+ \tau_3 = N_s+\frac{3}{2} \label{ClassicalR} \\
54: L= & \kappa_4+\kappa_5 = \ell \label{ClassicalL}
55: \end{align}
56:
57: \begin{table}[hbt]
58: \cpn{C$_2$H$_2$ stretch-bend effective Hamiltonian} {C$_2$H$_2$ stretch-bend effective Hamiltonian from \cite{Herman1995} and published in \cite{Gaspard} for the classical Hamiltonian. The parameters are in units of cm$^{-1}$. \label{sbparam}}
59: \begin{center} \vspace{0.2in}
60: \begin{tabular}{|c|r|c|r|c|r|c|r|} \hline \hline \, & \, & \, & \, & \, & \, & \, & \, \\
61: $\omega_1$ & 3501.537 & $x_{15}$ & -10.09 & $x_{45}$ & -2.311 & $K_{14/35}$ & 29.944 \\ \, & \, & \, & \, & \, & \, & \, & \, \\
62: $\omega_2$ & 2013.425 & $x_{22}$ & -7.802 & $x_{55}$ & -2.492 & $K_{44/55}$ & -12.909 \\ \, & \, & \, & \, & \, & \, & \, & \, \\
63: $\omega_3$ & 3417.644 & $x_{23}$ & -5.882 & $g_{44}$ & 0.4181 & $r_{45}$ & -6.09 \\ \, & \, & \, & \, & \, & \, & \, & \, \\
64: $\omega_4$ & 621.692 & $x_{24}$ & -12.841 & $g_{45}$ & 6.603 & --- & --- \\ \, & \, & \, & \, & \, & \, & \, & \, \\
65: $\omega_5$ & 746.773 & $x_{25}$ & -1.829 & $g_{55}$ & 3.676 & $y_{244}$ & 0.1522 \\ \, & \, & \, & \, & \, & \, & \, & \, \\
66: $x_{11}$ & -24.758 & $x_{33}$ & -27.483 & $K_{11/33}$ & -102.816 & $K_{33/1244}$ & 6.38 \\ \, & \, & \, & \, & \, & \, & \, & \, \\
67: $x_{12}$ & -11.199 & $x_{34}$ & -10.617 & $K_{3/245}$ & -16.698 & $k_{4}$ & -1.315 \\ \, & \, & \, & \, & \, & \, & \, & \, \\
68: $x_{13}$ & -103.386 & $x_{35}$ & -8.676 & $K_{1/244}$ & 6.379 & $r_{445}$ & 0.1255 \\ \, & \, & \, & \, & \, & \, & \, & \, \\
69: $x_{14}$ & -12.98 & $x_{44}$ & 3.595 & $K_{1/255}$ & 6.379 & $r_{455}$ & -0.225 \\ \, & \, & \, & \, & \, & \, & \, & \, \\ \hline \hline
70: \end{tabular} \end{center} \end{table}
71:
72: \subsection{5.1.2 Overview of Existing Studies}
73: \addtocontents{toc}{\protect\vspace*{5pt}}
74:
75: To our knowledge, the stretch-bend effective Hamiltonians has never been analyzed with all the resonances. We are aware of only three relevant studies:
76:
77: \begin{itemize}
78: \item Pals and Gaspard \cite{Gaspard} investigated the recurrences of the classical trajectories. Although the study did include the stretching DOF, the authors were mostly focused on the pure bending subsystem.
79:
80: \item Hasegawa and Someda \cite{DerivState} analyzed the quantum dynamics using a perturbative method. The focus was the short-time evolution of 3 types of quantum ZOS: $|0,0,0,n_4^0,0^0 \rangle$, $|1,3,0,6^0,0^0 \rangle$ and $|0,0,n_3,0^0,0^0 \rangle$.
81:
82: \item Kellman \et investigated the planar system ($\ell_4=\ell_5=0$) with 3 resonances: $K_{11/33}, K_{3/245}$ and $K_{44/55}$ in a diabatic correlation approach \cite{KellmanEPJD}. A series of states termed the ``primary subpolyad" are identified, which carries most of the intensity in the experimental spectra. The states are then fit to a single-resonance Hamiltonian in an effort to interpret the spectral patterns.
83: \end{itemize}
84:
85: Due to the dimensionality of the problem, none of these studies explicitly considered all the major resonances in the effective Hamiltonian. Since the approach presented in this thesis is designed to be dimensionality-independent, we believe it is more suitable for analyzing the stretch-bend acetylene system.
86:
87: \section{Preliminary Considerations}
88: \addtocontents{toc}{\protect\vspace*{7pt}}
89:
90: In Chapter 4 it was shown that the normal bending modes are destabilized at increased internal energy (polyad numbers) by the resonance couplings, and new stable bending modes are born in the bifurcations. In the stretch-bend system, the PO search by Prosmiti and Farantos also indicates that the highly excited stretch-bend system retains some regularity \cite{C2H2Farantos1}. Therefore one can expect a similar to the pure bending case to exist in the stretch-bend dynamics, provided there are sufficient excitation and coupling.
91:
92: \subsection{5.2.1 The Stretch Overtone Polyads}
93: \addtocontents{toc}{\protect\vspace*{5pt}}
94:
95: The critical points analysis starts with solving for all critical points in the 4 DOF (7 DOF - 3 polyad numbers) reduced classical Hamiltonian. In obtaining the preliminary results, we focus on a single series of polyads of interest, as opposed to varying all three polyad numbers independently.
96:
97: Of the five normal modes of C$_2$H$_2$ depicted in Fig.~\ref{normalmodes}, the pure bending subsystem was analyzed in Chapter 4, and the C-C stretch ($\tau_2$) overtones form an isolated subsystem. This leaves the obvious question: what could happen to the two C-H normal stretches, as they are excited to higher energy? Although the C-H stretching dynamics has been actively studied using two-mode models \cite{C2H2HCAO}, their coupling to the other vibrational modes is complex and poorly understood. This motivates us to investigate the following polyad series, which includes the symmetric and antisymmetric C-H stretch overtones \fn{Here in the purely classical analysis, the quantization requirement that the action in each mode exceed the zero-order energy (see eqns.~\ref{eqn5.1}, \ref{eqn5.2}) will be ignored.}:
98: \begin{eqnarray}
99: \{ \tau_1, \tau_2, \tau_3, \tau_4, \tau_5, \kappa_4, \kappa_5 \} = \{ \tau_1, 0, 0, 0, 0, 0, 0 \}, \{0, 0, \tau_3, 0, 0, 0, 0 \} \label{CHovertone}
100: \end{eqnarray}
101: \noindent The polyads containing them are found by substituting eqn. (\ref{CHovertone}) into eqns. (\ref{ClassicalP}-\ref{ClassicalL}) to give
102: \begin{align}
103: P &=5 R, & L &=\ell=0
104: \end{align}
105: \noindent These $\lbrack P, R, \ell \rbrack=\lbrack 5 R, R, 0 \rbrack$ polyads are henceforward referred to as the {\it stretch overtone polyads}. The upper limit of $R$ in this study is set at 8, as the effective Hamiltonian in \cite{Herman1999} includes up to 6 quanta of C-H overtones excitation ($R=7.5$).
106:
107: \subsection{5.2.2 Stability of the Normal C-H Stretch Overtones}
108: \addtocontents{toc}{\protect\vspace*{5pt}}
109:
110: The first consideration is the stability of the normal C-H stretch overtone states. For this purpose, the classical trajectories very close to the normal mode overtones in eqn. (\ref{CHovertone}) are integrated. The overtone is classically stable if the deviation of nearby trajectories remains small. The onset of large-amplitude oscillations indicates the destabilization of the corresponding overtone trajectory.
111:
112: The results are displayed in Fig.~\ref{OvertoneStability}. In panel (a), the symmetric stretch overtone $\tau_1$ remains stable to at least $\tau_1=6.5$ ($n_1=6$). In contrast, the antisymmetric $\tau_3$ overtone becomes unstable at as low as $\tau_3=1.5$ ($n_3=1$). Two resonances may be responsible for the destabilization of $\tau_3$ overtones: $K_{11/33}$ and $K_{3/245}$.
113:
114: \newpage \begin{figure}[hbtp]
115: %==\vspace{2.93in}
116: \begin{center}\includegraphics[width=4.92in]{overtonestab.eps}\end{center}
117: \cpn{Classical stability of C-H normal stretch overtones}{Classical stability of C-H normal stretch overtones. Panel (a) and (b) are the $\tau_1$ and $\tau_3$ values of trajectories very close to the respective overtone conditions in eqn. (\ref{CHovertone}). \label{OvertoneStability}}
118: \end{figure} \newpage
119:
120: \noindent \underline{Single $K_{11/33}$ Resonance } \,\,\, The single $K_{11/33}$ resonance system
121: \begin{align}
122: H_{11/33} = H_0+\frac{K_{1133}}{2} \tau_1 \tau_3 \cos \lbrack 2(\phi_1-\phi_3) \rbrack
123: \end{align}
124: \noindent is analyzed for the stretch overtone polyads using the formulation of $\S$ 3.1, with $m:n=2:2$. After the other actions $\tau_2, \tau_4, \tau_5, \kappa_4, \kappa_5$ are set to be zero, ($\tau_1, \tau_3$) form a two-mode system. The 2 DOF reduced phase space is described be canonical variables $I_z, \psi$:
125: \begin{align}
126: I_z & = \frac{\tau_1-\tau_3}{2}, & \psi & = \phi_1- \phi_3 \label{K1133Canonical}
127: \end{align}
128:
129: The constant energy contours (at arbitrary energy) of $H_{11/33}$ in the ($I_z, \psi$) space are plotted in Fig.~\ref{PPS1133}. Each of these panels is the Mecartor projection of the corresponding PPS, as in panel (a) of Fig.~\ref{phasesphere}. The top and bottom of $I_z$ represent the $\tau_1$ and $\tau_3$ overtones, respectively. The contours display the same qualitative features as semiclassical trajectories of the eigenstates. The critical points (dark dots in panel a) at $\psi=0, \pi$ and $2\pi$ are the stable (E) local C-H stretch critical points \fn{The points at $0$ and $2\pi$ are identified with each other. They are also energetically degenerate with the one at $\pi$, as the two local C-H stretches are classically equivalent.}. The local model of C-H stretch has been previously used in studying the two coupled C-H stretch oscillators \cite{ChildLawtonC2H2,Klemperer}.
130:
131: The curves at the top of all panels of Fig.~\ref{PPS1133} that run across $\lbrack 0, 2\pi \rbrack$ indicate the $\tau_1$ overtone is a stable critical point. The $\tau_3$ overtone, in contrast, becomes {\it unstable} on the PPS from $\tau_3=2$ onward. These qualitative observations verify the stabilities shown in Fig.~\ref{OvertoneStability}.
132:
133: \newpage \begin{figure}[hbtp]
134: %==\vspace{4.5in}
135: \begin{center}\includegraphics[width=4.5in]{pps1133.eps}\end{center}
136: \cpn{Reduced phase space of $H_{11/33}$ Hamiltonian}{Reduced phase space of $H_{11/33}$ Hamiltonian, and the constant energy contours. In panels (a)-(d) $R=2, 4, 6$ and $8$ respectively. The dashed line in panel (a) labels the separatrix crossing the unstable $\tau_3$ overtones (the $I_z=-1$ line). The black dots at $\psi=0, \pi$ and $2\pi$ indicate the stable local mode critical point. \label{PPS1133}}
137: \end{figure} \newpage
138:
139: \noindent \underline{Single $K_{3/245}$ Resonance } \,\,\, The single $K_{3/245}$ resonance system
140: {\small \begin{align}
141: H_{3/245} & = H_0+\frac{K_{3/245}}{4}\sqrt{\tau_2\tau_3}\{\sqrt{(\tau_4-\kappa_4)(\tau_5+\kappa_5)}\cos[\phi_2-\phi_3+\phi_4-\chi_4+\phi_5+\chi_5] \nonumber\\
142: & \quad +\sqrt{(\tau_4+\kappa_4)(\tau_5-\kappa_5)}\cos[\phi_2-\phi_3+\phi_4+\chi_4+\phi_5-\chi_5]\} \label{k3245}
143: \end{align} }
144: \noindent was formally analyzed by Rose and Kellman in \cite{Rose2345}. However, they failed in performing a calculation using actual parameters of C$_2$H$_2$. Here we perform the calculation for the stretch overtone polyads, with actions $\tau_1=\kappa_4=\kappa_5=0$ \fn{In \cite{Rose2345}, the elimination of $\kappa_i$ from consideration is not rigorous, since the $K_{3/245}$ resonance does couple $\ell_4, \ell_5$, and therefore $\kappa_4, \kappa_5$.}. The $\tau_3$ overtones are then contained in the action space with
145: \begin{align}
146: \tau_2 = \tau_4 = \tau_5 = R-\tau_3
147: \end{align}
148: \noindent Because the derivation in \cite{Rose2345} failed to use the proper $d_i$ for the doubly-degenerate bends, the canonical variables ($J, \varphi$) \fn{The notations are intentionally different from those used in $H_{11/33}$ above to avoid confusion.} of the 1 DOF reduced phase space are redefined as:
149: \begin{align}
150: J &= 14 \tau_2 + \tau_3 - 7 \tau_4 - 7 \tau_5, & \varphi &= \phi_3-\phi_2-\phi_4-\phi_5
151: \end{align}
152:
153: Fig.~\ref{PPS3245} displays the arbitrary constant energy contours in ($J, \varphi$) space. The top of each panel (maximum $J$) corresponds to the $\tau_3$ overtones and the bottom to
154: \begin{align}
155: \tau_2 & =\tau_4=\tau_5 = R, & \tau_3 &=0\nonumber\\
156: \end{align}
157: \noindent In panel (a), the $K_{3/245}$ resonance induces a bifurcation of the $\tau_3$ overtone. In this process, a stable critical point is created at $\varphi=\pi$. However, the overtone itself remains stable at higher $R$ (panels b-d).
158:
159: \newpage \begin{figure}[hbtp]
160: %==\vspace{4.5in}
161: \begin{center}\includegraphics[width=4.5in]{pps3245.eps}\end{center}
162: \cpn{Reduced phase space of $H_{3/245}$ Hamiltonian}{Reduced phase space of $H_{3/245}$ Hamiltonian, and the constant energy contours. In panels (a)-(d) $R=$ 2, 4, 6 and 8. \label{PPS3245}} \end{figure} \newpage
163:
164: From the above separate consideration of $H_{11/33}$ and $H_{3/245}$, the $\tau_3$ overtones are first destabilized by the $K_{11/33}$ resonance below $R=2$, and the {\it local} C-H stretch becomes the stable mode.
165:
166: \subsection{5.2.3 Effect of Stretch-Bend Resonances}
167: \addtocontents{toc}{\protect\vspace*{7pt}}
168:
169: After its creation, the local C-H stretch may be further perturbed by the stretch-bend resonances to couple with the bending DOF. Of the four stretch-bend resonances ($K_{3/245}$, $K_{1/244}$, $K_{1/255}$ and $K_{14/35}$), it is unlikely that they are equally important at all parts of a stretch overtone polyad. We use a simplified version of Chirikov analysis \cite{Chirikov} in order to estimate of which resonances are important in affecting the local C-H stretch. When $\kappa_4, \kappa_5$ are left out of consideration, the ``strength" of each of these 4 resonances as well as $K_{11/33}$ can be measured by how far the zero-order frequencies $\omega_i= \partial H_0 / \partial \tau_i$ are tuned towards the integer ratio $m_i$ corresponding to an exact resonance condition. At the exact integer ratio, $\omega_i$ satisfy \cite{OxtobyRice}:
170: \begin{align}
171: \sum_i m_i \frac{\partial H_0}{\partial \tau_i} = 0 \label{resonanceplane}
172: \end{align}
173: \noindent The vector with integer components $\{ m_1, m_2, m_3, m_4, m_5 \}$ corresponds to the resonance vectors mentioned in $\S$ 2.1. In this case they are:
174: \begin{align}
175: K_{11/33}: & \{ 2,0,-2,0,0 \}, & K_{3/245}: & \{ 0,-1,1,-1,-1 \} \nonumber\\
176: K_{1/244}: & \{ 1,-1,0,-2,0 \}, & K_{1/255}: & \{ 1,-1,0,0,-2 \} \nonumber\\
177: K_{14/35}: & \{ 1,0,-1,1,-1 \} & \, & \, \nonumber
178: \end{align}
179:
180: For each resonance, eqn. (\ref{resonanceplane}) defines a 4-dimensional hypersurface in the $\{ \tau_1, \tau_2, \tau_3, \tau_4, \tau_5 \}$ space. Because of the conservation of $P$ and $R$, only 3 of the actions can be independently varied within a polyad. It is therefore possible to represent these hypersurfaces as 2-dimensional surfaces, which are referred to as {\it resonance planes} in a 3-dimensional volume ($J_1$, $J_2$, $J_3$)
181: \begin{align}
182: J_1= & \tau_1-\tau_3 & \in & [ -R, R ] \\
183: J_2= & \tau_2 & \in & [ 0, R ] \\
184: J_3= & \tau_4-\tau_5 & \in & [ -2 R, 2 R ]
185: \end{align}
186: \noindent This volume is further constrained by the requirement that all the $\tau_i$ should be non-negative.
187:
188: Fig.~\ref{SBChirikov} depicts the resonance planes in ($J_1, J_2, J_3$). In panel (a), at $R=4$ the C-H stretch system (thick line) is separately in contact with $K_{11/33}$ (red) and $K_{1/244}$ (magenta) resonance planes at two places. The $K_{1/244}$ resonance only perturbs but does not destabilize the $\tau_1$ overtones, according to Fig.~\ref{OvertoneStability}. In panels (b) and (c), the C-H stretch system first interacts with the $K_{11/33}$ resonance plane, which is next intersected by the $K_{1/244}$ plane. In all 3 panels the $K_{3/245}$ and $K_{1/255}$ planes are located at another side of the ($J_1, J_2, J_3$) space, and can only interact with the C-H stretch system via $K_{11/33}$ and other resonance planes. Because the $K_{14/35}$ resonance term vanishes in the absence of bending excitation, the $K_{14/35}$ its plane (navy) also cannot directly interact with the C-H stretch system.
189:
190: \newpage \begin{figure}[hbtp]
191: %==\vspace{6.37in}
192: \begin{center}\includegraphics[width=5.747in]{resonanceplanes.eps}\end{center}
193: \cpn{Frequency resonance planes in the stretch overtone polyads}{Frequency resonance planes in the stretch overtone polyads. Panels (a), (b) and (c) display the cases with $R=$ 4, 6 and 8, respectively. The color coding is: $K_{11/33}$ (red), $K_{3/245}$ (turquoise), $K_{1/244}$ (magenta), $K_{1/255}$ (green), $K_{14/35}$ (navy). The C-H stretch system with $\tau_2=\tau_4=\tau_5=0$ is indicated by the thick black line in each panel. \label{SBChirikov}}
194: \end{figure} \newpage
195:
196: \section{Critical Points Analysis}
197: \addtocontents{toc}{\protect\vspace*{7pt}}
198:
199: Based on the results of $\S$ 5.2.2-5.2.3, in this subsection we compute critical points in the following three cases:
200: \begin{enumerate}
201: \item The ($\tau_1, \tau_3$) subsystem with $K_{11/33}$ resonance;
202: \item The ($\tau_1, \tau_2, \tau_3, \tau_4$) subsystem with both $K_{11/33}$ and $K_{1/244}$ resonances;
203: \item The full $H_{sb}$ of eqn. (\ref{sbclassicalH}) with all 7 resonance couplings.
204: \end{enumerate}
205:
206: \subsection{5.3.1 Computational Details}
207: \addtocontents{toc}{\protect\vspace*{5pt}}
208:
209: Using the procedure described in Appendix A, a canonical transformation is selected that expresses $H_{sb}$ in the following new action-angle variables: 3 pairs of trivial action-angle variables ($P, R, L, \theta_P, \theta_R, \theta_L$) for the polyad numbers (defined in \ref{ClassicalP}-\ref{ClassicalL}) and their conjugate angles, and 4 pairs of non-trivial ones ($I_1, I_2, I_3, I_4, \Psi_1, \Psi_2, \Psi_3, \Psi_4$) which span the 8-dimensional reduced phase space. The angles $\Psi_1-\Psi_4$ are chosen to correspond to the $K_{11/33}$, $K_{1/244}$, $K_{44/55}$ and $\ell$ resonances, respectively. The simplest canonical transformation we have found so far is:
210: \begin{align}
211: I_1=& 12 \tau_1 + 8 \tau_2 + 11 \tau_3 + 2 \tau_4 + 2 \tau_5 + 2 \kappa_4 + 2 \kappa_5, & \Psi_1=& \phi_1-\phi_3 \\
212: I_2=& 6 \tau_1 + 3 \tau_+ 6 \tau_3+ \tau_4+ \tau_5 + \kappa_4 + \kappa_5, & \Psi_2=& \phi_1-\phi_2-2\phi_4 \\
213: I_3=& 6 \tau_1 + 4 \tau_2 + 6 \tau_3 + \tau_4 + \kappa_4 + \kappa_5, & \Psi_3=& \phi_4-\phi_5 \\
214: I_4=& -12 \tau_1 - 8 \tau_2 - 12 \tau_3 - 2 \tau_4 - 2 \tau_5 - \kappa_4 - 2 \kappa_5, & \Psi_4=& \chi_4-\chi_5
215: \end{align}
216: \noindent $\Psi_1$ is the relative phase angle between the symmetric $\tau_1$ and antisymmetric $\tau_3$ oscillators. $\Psi_3$ and $\Psi_4$ are identical to $\psi_a$ and $\psi_b$ in Chapter 4, respectively.
217:
218: Critical points in the reduced phase space are defined by:
219: \begin{align}
220: \frac{\partial H_{sb} }{\partial \Psi_1} =\frac{\partial H_{sb}}{\partial \Psi_2 }=\frac{\partial H_{sb}}{\partial \Psi_3}=\frac{\partial H_{sb}}{\partial \Psi_4} & =0 \label{SBFP2} \\
221: \frac{\partial H_{sb}}{\partial I_1} =\frac{\partial H_{sb}}{\partial I_2}=\frac{\partial H_{sb}}{\partial I_3}=\frac{\partial H_{sb}}{\partial I_4} & =0 \label{SBFP1}
222: \end{align}
223:
224: In eqns. (\ref{SBFP2}), $\Psi_i$ only appear in the form of $\cos$ functions. Similarly to the treatment of the pure bending subsystem ($\S$ 4.3.1), a sufficient condition for them to be simultaneously satisfied is:
225:
226: {\footnotesize \begin{align}
227: & \sin[2\Psi_1]=\sin[\Psi_2]=\sin[2\Psi_3]=\sin[2\Psi_4]=\sin[\Psi_2-2\Psi_3] \nonumber\\
228: & \quad = \cos[\Psi_1+\Psi_3 \pm \Psi_4]\sin[\Psi_1+\Psi_3 \mp \Psi_4] =\sin[\Psi_1-\Psi_2+\Psi_3 \pm \Psi_4]=0 \label{sbangles}
229: \end{align}}
230:
231: One can therefore fix $\Psi_i$ to the discrete values satisfying eqn. (\ref{sbangles}), and only solve the remaining 4 equations in (\ref{SBFP1}). The latter are transformed by the following substitutions:
232: \begin{eqnarray}
233: &u_1 =\sqrt{\tau_1}, u_2 =\sqrt{\tau_2}, u_3 =\sqrt{\tau_3}, \nonumber\\
234: &u_4 =\sqrt{\tau_4+\kappa_4}, u_5 =\sqrt{\tau_4-\kappa_4}, u_6 =\sqrt{\tau_5+\kappa_5}, u_7 =\sqrt{\tau_5-\kappa_5} \label{sbtransform}
235: \end{eqnarray}
236: \noindent and then multiplied by appropriate factors to remove their denominators. The result is 7 simultaneous polynomial equations:
237: \begin{align}
238: & \frac{\partial H_{sb}(u_i)}{\partial I_1(u_i)} \cdot (u_1 u_3)^{1/2} = 0 \label{3EQ1} \\
239: & \frac{\partial H_{sb}(u_i)}{\partial I_2(u_i)} \cdot (u_1 u_2 u_4 u_5)^{1/2} = 0 \label{3EQ2} \\
240: & \frac{\partial H_{sb}(u_i)}{\partial I_3(u_i)} \cdot (u_4 u_5 u_6 u_7)^{1/2} = 0 \label{3EQ3} \\
241: & \frac{\partial H_{sb}(u_i)}{\partial I_4(u_i)} \cdot (u_4 u_5 u_6 u_7)^{1/2} = 0 \label{3EQ4} \\
242: & 5 u_1^2 +3 u_2^2+5 u_3^2 + \frac{1}{2}(u_4^2 + u_5^2+ u_6^2 + u_7^2 ) = P \label{sbconstrain1} \\
243: & u_1^2 +u_2^2 +u_3^2 = R \label{sbconstrain2} \\
244: & u_4^2 - u_5^2 + u_6^2 - u_7^2 = 0 \label{sbconstrain3}
245: \end{align}
246: \noindent The 7 unknown variables $u_1-u_7$ are then solved for in {\it PHCpack} \cite{PHC}.
247:
248: \noindent \underline{Treatment of Subsystems } \,\,\, When any of the $u_i$ vanishes, at least one of eqns. (\ref{3EQ1}-\ref{3EQ4}) becomes unphysical. This happens when there is zero action in any normal mode $\tau_i$, or when $\tau_j = \pm \kappa_j$.
249:
250: \noindent \,\,\,
251:
252: Special consideration is required for these cases. The contributions from the resonance(s) involved with the vanishing action should also vanish from the Hamiltonian. The equations defining the critical points then must be adjusted accordingly. As a pedagogical example, in the H$_2$O system the two O-H normal stretch critical point families have no bend action \cite{Zi-MinH2O1}. In deriving these families from the critical points analysis, the two stretch-bend Fermi resonances have to be removed. Critical points in the resulting 2 DOF subsystem then correspond to the normal stretch modes.
253:
254: In order to locate {\it all} the possible critical points in the stretch-bend system, all combinations of $u_i=0$ have to be considered. In each case, the relevant zero-order and resonance terms in the classical Hamiltonian are removed. A new canonical transformation (Appendix A) may be necessary so that the new angles correspond to the remaining resonances. Here we treat the two subsystems outlined in the beginning of $\S$ 5.3 using this kind of special consideration.
255:
256: \subsection{5.3.2 Results}
257: \addtocontents{toc}{\protect\vspace*{7pt}}
258:
259: The resulting critical points consist of families of curves in the ($\tau_i, \kappa_j$) 7-dimensional action space, parameterized by the polyad number $R$. Associated with each family is a discrete set of $\Psi_i$ values. All these solutions turn out to have $\kappa_4 = \kappa_5=0$. In Fig.~\ref{sbbifurfig}, we graph the results in 5 separate panels with each $\tau_i$ versus $R$ separately. In the full system, only those with significantly non-zero $\tau_i$ are displayed here. In the $K_{11/33}$ subsystem, $\tau_2= \tau_4= \tau_5=\kappa_4=\kappa_5=0$. In the $K_{11/33}+K_{1/244}$ subsystem, $\tau_5=\kappa_4=\kappa_5=0$.
260:
261: With $R \in \lbrack 2, 8 \rbrack $, we found one family for the $K_{11/33}$ subsystem corresponding to the local C-H stretch, two families for the $K_{11/33}+K_{1/244}$ subsystem, and two families for $H_{sb}$ system.
262:
263: \newpage \begin{figure}[hbtp]
264: %==\vspace{5.14in}
265: \begin{center}\includegraphics[width=5.73in]{sbbifur.eps}\end{center}
266: \cpn{Critical points in the stretch overtone polyads} {Critical points in the stretch overtone polyads $\lbrack 5R, R, 0 \rbrack$. Black, gray and red: critical points in $H_{sb}$. Navy and magenta: the $K_{11/33}+ K_{1/244}$ system with $\tau_5=0$. Green: The $K_{11/33}$ system, with $\tau_2=\tau_4=\tau_5=0$. \label{sbbifurfig}}
267: \end{figure} \newpage
268:
269: When there is only $K_{11/33}$ resonance, one family of stable (E) critical points bifurcates out of the $\tau_3$ overtone. The new family has
270: \begin{align}
271: \Psi_1 = 0 \label{K1133Family}
272: \end{align}
273: \noindent which is the local C-H stretch mode on Fig.~\ref{PPS1133}.
274:
275: When both $K_{11/33}+K_{1/244}$ are included and $\tau_5=\kappa_5=0$, the reduced phase space employs four variables ($I_1, I_2, \Psi_1, \Psi_2$). Two new families of critical points are created at:
276: \begin{align}
277: (\Psi_1, \Psi_2) = (0, 0) , (0, \pi)
278: \end{align}
279: \noindent The $(0, 0)$ family has (EE) stability and the $(0, \pi)$ family has (EH) stability. The C-H stretching motion of these critical points is qualitatively like the local C-H stretch motion, because the $\Psi_1$ value for both families is the same as in eqn. (\ref{K1133Family}). The $\Psi_2$ angle, as the phase angle of a three-mode resonance $K_{1/244}$, is yet to be given a clear physical meaning.
280:
281: In the full $H_{sb}$, solving the critical point equations (\ref{3EQ1}-\ref{sbconstrain3}) results in two more families of critical points with
282: \begin{align}
283: (\Psi_1,\Psi_2,\Psi_3,\Psi_4) = (0,0,0,\pi), (0,\pi,0,\pi)
284: \end{align}
285: \noindent Although there is no explicit constraint, both of these families have $\kappa_4=\kappa_5=0$. The $(0, 0, 0, \pi)$ family start with (EEEE) stability and switches to (EEMM) at $R=5.45$. The $(0, \pi, 0, \pi)$ family has (EEEH) stability for up to $R=$ 8. These two families also have $\Psi_1=0$, indicating a local-type stretch in the C-H DOF. The values of two other angles $\Psi_3=0, \Psi_4=\pi$, according to Table~\ref{16cond}, suggest that the bending motion is of the local ({\bf L}) type.
286:
287: A striking feature in Fig.~\ref{sbbifurfig} is that the critical points in both the $K_{11/33}+K_{1/244}$ and full Hamiltonian appear to have bifurcated out of the local C-H stretch family (green line). The latter then bifurcates out of the $\tau_3$ overtones. We believe that the new families of critical points in Fig.~\ref{sbbifurfig} correspond to novel types of dynamics in the respective stretch overtone polyads.
288:
289: These critical points in the stretch-bend Hamiltonian do not correspond to PO in the full phase space when there are up to 3 frequencies in the cyclic angles:
290: \begin{align}
291: \dot{\theta}_P =\frac{\partial H_{sb}}{\partial P}, \mbox{\,\,\,\,\,} \dot{\theta}_R =\frac{\partial H_{sb}}{\partial R}, \mbox{\,\,\,\,\,} \dot{\theta}_\ell =\frac{\partial H_{sb}}{\partial \ell}
292: \end{align}
293: \noindent The last frequency, $\dot{\theta}_\ell$, turns out to be zero for all the critical points graphed in Fig.~\ref{sbbifurfig}, in the same manner as $\dot{\theta}_b$ does in the $\lbrack N_b, 0 \rbrack$ pure bending polyads when $\ell=0$ (see Chapter 4). The first two frequencies, $\dot{\theta}_P$ and $\dot{\theta}_R$, are not uniquely defined. Instead, an alternative choice of the canonical transformation can change the definitions of $\theta_P$ and $\theta_R$ as well as their frequencies.
294:
295: \section{Summary}
296: \addtocontents{toc}{\protect\vspace*{7pt}}
297:
298: The critical points analysis is performed on the C$_2$H$_2$ stretch-bend effective Hamiltonian, for the purpose of clarifying the fate of normal C-H stretch mode under increasing excitation. Preliminary results indicate that the normal antisymmetric C-H stretch overtone (critical point) is first destabilized by $K_{11/33}$ resonance to produce the local C-H stretch. Then the local stretch bifurcates into at least 4 families of stretch-bend critical points at higher excitation after the inclusion of stretch-bend resonances. Such an observation suggests that the stretch-bend dynamics could be influences by a chain of bifurcations, each induced by perhaps the addition of one resonance.
299:
300: Currently we are working to clarify the physical meaning of these critical points, as well as their role in the classical phase space. A more complete classification, including all the subsystems (combinations of $\tau_i=0$) is also in progress.