1: \chapter{Analytic and Numerical Results}
2: \label{ch:results}
3:
4: A computer program to solve the discretized integral equations can be implemented with the help of some basic numerical algorithms. Once completed, it can be used to generate highly accurate results and analyze the three-body system. Nonetheless, attempts to extract analytic results should not be overlooked. Even some of the most general properties of the integral equations allow us to draw conclusions about the behavior of the system.
5:
6: We begin this chapter with analytic results obtained from studying the leading order integral equations. These results include statements about the cutoff dependence of bound-state energies and the phase for $f_{d0}$. A proof for the cyclic behavior of $\delta_0$ is given, which is then used to infer similar behavior for $G_3$.
7:
8: Following the analytic results is a section containing numerical solutions to the integral equations. Here we examine behavior that cannot be determined analytically. Solutions for the functions $f_{l0}$ and $f_{h0}$ are shown, and the cutoff dependence of $G_3$ is calculated. Some relations that are proven analytically are also verified numerically.
9:
10:
11: %%%%%%%%%%%%%%%%%%%%%%%%
12: %% ANALYTIC RESULTS %%
13: %%%%%%%%%%%%%%%%%%%%%%%%
14:
15: \section{Analytic Results}\label{sec:analytic}
16:
17: In Chapter \ref{ch:expand3body}, we saw that Eq.~(\ref{eqn:QRfl0}) contains no $\Lambda$ dependence, making the function $f_{l0}(p)$ cutoff-independent. However, for values of $p \gg \eta_3$, the solution must behave like $f_{d0}(p) = \cos\left(s_0 \, \ln\left(p/\Lambda\right) + \theta \right)$, which explicitly has the cutoff in it. The only way these two equations can be reconciled is if $\theta$ contains cutoff dependence of the form $s_0 \, \ln(\Lambda)$. Any remaining part of $\theta$ must be a function of $\eta_2$ and $\eta_3$. Therefore, we shall write
18: %
19: \begin{equation}
20: \theta = s_0 \, \ln\left(\Lambda/\eta_3\right) + \tilde{\theta}\left(\eta_2/\eta_3\right) \label{eqn:theta_tilde},
21: \end{equation}
22: %
23: \noindent where $\tilde{\theta}$ is a dimensionless function of the ratio $\eta_2/\eta_3$. This relation holds for any values of $\eta_2$ and $\eta_3$, including all $\eta_3$ values corresponding to multiple bound states with the same $\eta_2$. In the next chapter, we shall see that $\tilde{\theta}$ is closely related to Efimov's universal function \cite{efimov:2}. Using $\eta_3$ in the ratio with $\Lambda$ is a matter of choice. Any quantity composed of $\eta_2$ and $\eta_3$ with the same dimension as $\Lambda$ would work just as well; however, we need to allow $\eta_2 = 0$, so $\eta_2$ alone is a poor choice.
24:
25: While Eq.~(\ref{eqn:theta_tilde}) is quite simple, it has many interesting consequences. As we mentioned earlier, two different three-body bound states in the same spectrum must have the same phase $\theta$. Suppose that we choose some fixed values for $\eta_2$ and $\eta_3$ and make them cutoff-independent by choosing the appropriate $\Lambda$ dependence for $G_2$ and $G_3$. This might be desirable if we are trying to match those energies to experimental data. The phase for the solution in this case is
26: %
27: \begin{equation}
28: \theta_{\Lambda} = s_0 \, \ln\left(\Lambda/\eta_3\right) + \tilde{\theta}\left(\eta_2/\eta_3\right) ,
29: \end{equation}
30: %
31: \noindent for some given cutoff $\Lambda$. If the cutoff is changed, $G_2$, $G_3$, and $\theta_{\Lambda}$ will all change with it, but $\eta_2$, $\eta_3$, and $\tilde{\theta}(\eta_2/\eta_3)$ will not.
32:
33: Imagine now that we find a second three-body bound-state solution with the same phase. Let us call its energy $\bar{\eta}_3$. The phase for this solution is
34: %
35: \begin{equation}
36: \bar{\theta}_{\Lambda} = s_0 \, \ln\left(\Lambda/\bar{\eta}_3\right) + \tilde{\theta}\left(\eta_2/\bar{\eta}_3\right) ,
37: \end{equation}
38: %
39: \noindent which must be equal to $\theta_{\Lambda}$ by assumption. This results in the relation
40: %
41: \begin{equation}
42: s_0 \, \ln\left(\bar{\eta}_3/\eta_3\right) = \tilde{\theta}\left(\eta_2/\bar{\eta}_3\right) - \tilde{\theta}\left(\eta_2/\eta_3\right) .
43: \end{equation}
44: %
45: \noindent If the cutoff is now changed to a new value $\Lambda'$, the original data gives a phase of
46: %
47: \begin{equation}
48: \theta_{\Lambda'} = s_0 \, \ln\left(\Lambda'/\eta_3\right) + \tilde{\theta}\left(\eta_2/\eta_3\right) .
49: \end{equation}
50: %
51: \noindent The question is whether $\bar{\eta}_3$ is still a valid solution. Its new phase is
52: %
53: \begin{eqnarray}
54: \bar{\theta}_{\Lambda'} & = & s_0 \, \ln\left(\Lambda'/\bar{\eta}_3\right) + \tilde{\theta}\left(\eta_2/\bar{\eta}_3\right) \nonumber
55: \\
56: & = & s_0 \, \ln\left(\Lambda'/\bar{\eta}_3\right) + \left[ s_0 \, \ln\left(\bar{\eta}_3/\eta_3\right) + \tilde{\theta}\left(\eta_2/\eta_3\right) \right] \nonumber
57: \\
58: & = & s_0 \, \ln\left(\Lambda'/\eta_3\right) + \tilde{\theta}\left(\eta_2/\eta_3\right) \nonumber
59: \\
60: & = & \theta_{\Lambda'} .
61: \end{eqnarray}
62: %
63: \noindent Since the phases still match, $\bar{\eta}_3$ is still a bound-state solution. This remains true for any cutoff, implying that $\bar{\eta}_3$ is also cutoff-independent like $\eta_3$. Of course, the same statement applies to any other three-body bound state {\em making the entire spectrum completely independent of $\Lambda$}. Such behavior should come as no surprise since the leading order equations represent the $\Lambda \rightarrow \infty$ limit. Keep in mind that this is true for any other physical quantity, but does not apply to the couplings. Obviously, it is the cutoff dependence of the couplings that enables the bound states to be cutoff independent. We have shown that a single three-body contact interaction allows us to renormalize the entire three-body bound-state spectrum.
64:
65: Just as $f_{l0}$ has no dependence on $\Lambda$, neither does it have any dependence on $h_2$. As a consequence, $h_2$ has no effect at leading order on the binding energies or other physical quantities. It does appear in the equation for $f_{l1}$ however, showing that it is needed when considering first order corrections. Because we are working only to leading order, we shall use $h_2 = 0$ in our calculations from now on unless otherwise specified.
66:
67: We now turn to the equation for $f_{h0}(p)$:
68: %
69: \begin{eqnarray}
70: f_{h0}(p, \Lambda) = \frac{p^2}{2 \pi^2 D_{h0}(p,\Lambda)} \int_0^{\infty} dq \, \int_{-1}^{1} dz \, \frac{U_2\left(\vec{q} + \frac{1}{2}\vec{p}\right) U_2\left(\vec{p} + \frac{1}{2}\vec{q}\right)}{2p^2 + 2q^2 + 2pqz} f_{h0}(q, \Lambda) \nonumber
71: \\
72: \hspace{0.5in} -~\delta_0 \, \frac{3 h_2^2 + 8 h_2 + 16}{128 \sqrt{2} \, \pi^{3/2}} \left( \frac{D_{1h0}(p,\Lambda)}{D_{h0}(p,\Lambda)} \right) \left( \frac{p^2}{\Lambda^2} \right) \label{eqn:thisfh0}.
73: \end{eqnarray}
74: %
75: \noindent The dependence on $p$ and $\Lambda$ has been explicitly shown. Since $f_{h0}$ is dimensionless, only the ratio $p/\Lambda$ can occur in the function. Furthermore, there is a dependence upon $h_2$ and $\theta$. The $h_2$ dependence comes from its appearance in the function $U_2(p)$, either directly in the integral or indirectly in $D_{h0}$ and $D_{1h0}$. The $\theta$ dependence is a result of $f_{h0}$ approaching $f_{d0}$ for $p \ll \Lambda$.
76:
77: Assuming that $\Lambda$ is held fixed, choosing a value for $\delta_0$ will uniquely determine the phase. Thus, we can view the phase as a function of the coupling, $\theta(\delta_0)$. Conversely, choosing a phase determines the coupling, so it is equally valid to treat the coupling as a function of the phase, $\delta_0(\theta)$. The coupling $\delta_0$ can also have a dependence upon $h_2$ but not upon $\Lambda$. The reason is that $\delta_0$ is dimensionless and there is no other quantity available to form a dimensionless ratio with $\Lambda$.
78:
79: With this in mind, consider a solution to Eq.~(\ref{eqn:thisfh0}) with a coupling of $\delta_0(\theta)$ and a function $f_{h0}(p)$ that behaves asymptotically as $\cos\left(s_0 \, \ln\left(p/\Lambda\right) + \theta \right)$ for small $p$. For a phase of $\theta + 2 \pi$, the cosine behavior will remain unchanged. We may therefore conclude that
80: %
81: \begin{equation}
82: \delta_0(\theta + 2 \pi) = \delta_0(\theta) ,
83: \end{equation}
84: %
85: \noindent showing that the coupling exhibits periodic behavior. If we apply the operator $-\partial^2/\partial\theta^2$ to both sides of (\ref{eqn:thisfh0}), we find once again that the asymptotic cosine behavior is the same. This implies
86: %
87: \begin{equation}
88: -\frac{\partial^2 \delta_0}{\partial\theta^2} = \delta_0(\theta) .
89: \end{equation}
90: %
91: \noindent The conclusion to be drawn is that the coupling may be written as a cosine function with some amplitude $\mathcal{A}$ and phase $\phi$. These two parameters will contain any $h_2$ dependence that $\theta$ may possess, so we shall write
92: %
93: \begin{equation}
94: \delta_0(\theta) = \mathcal{A}(h_2) \, \cos\left( \theta + \phi(h_2) \right) \label{eqn:PseudoCosine}.
95: \end{equation}
96:
97: The equation relating $\delta_0$ and $G_3$,
98: %
99: \begin{equation}
100: G_{3,0} = \frac{\delta_0}{G_{2,0} \, \mathcal{I}_0 - D_{2,0}\,\delta_0/\Lambda^4 } ,
101: \end{equation}
102: %
103: \noindent shows that whenever $\delta_0$ is zero, so is $G_3$.\footnote{The only problem that might arise is if $\mathcal{I}_0 = 0$ in such a way that the ratio is non-zero, but we have found no parameters for which this is true.} Since two adjacent zeros of $\delta_0$ occur when the cutoffs satisfy
104: %
105: \begin{equation}
106: \theta_{\Lambda'} - \theta_{\Lambda} = s_0 \, \ln\left(\frac{\Lambda'}{\Lambda}\right) = \pi ,
107: \end{equation}
108: %
109: \noindent the adjacent zeros of $G_3(\Lambda)$ should be spaced by a cutoff factor of $\Lambda'/\Lambda = \exp(\pi/s_0) \simeq 22.69438259536$. This suggests that $G_3(\Lambda)$ may also possess cyclic behavior, but this must ultimately be verified numerically.
110:
111:
112: %%%%%%%%%%%%%%%%%%%%%%%
113: %% NUMERIC RESULTS %%
114: %%%%%%%%%%%%%%%%%%%%%%%
115:
116: \section{Numerical Results}
117:
118: We begin our numerical investigation by considering solutions for the functions $f_{l0}$ and $f_{h0}$ which lead to an approximation for the complete function $f(p)$. Next, the cutoff independence of the three-body spectrum is verified, followed by an analysis of the coupling constants. These numerical results are confirmed by Wilson's calculations \cite{Wilson:notes}.
119:
120:
121: %%%%%%%%%%%%%%%%%%%%%
122: %% WAVEFUNCTIONS %%
123: %%%%%%%%%%%%%%%%%%%%%
124: \subsection{Low- and High-Momentum Functions}
125:
126: Figure \ref{fig:FL0-2} shows the numerical solution for $f_{l0}(p)$ in the case of $B_2 = 0.1$, $B_3 = 1.0$, and $\Lambda = 10^8$. Notice that it is constant for small values of momentum and then takes on the cosine behavior as $p$ becomes large. This behavior is typical of all solutions for $f_{l0}$. For reference, we have included numerical solutions for the additional cases of $B_2 = 0.0, 0.5,$ and $1.0$ in Figs.~\ref{fig:FL0-1}, \ref{fig:FL0-3}, and \ref{fig:FL0-4} respectively. Since the behavior of the solution to the low-momentum equation is determined by the ratio $B_2/B_3$, these figures provide a representative sample of all solutions. We can also study the effects of changing the cutoff while keeping the energies fixed. Equation (\ref{eqn:theta_tilde}) shows that changing the value of $\Lambda$ will only result in a shift in the value of $\theta$, which is exactly the behavior seen in Fig.~\ref{fig:FL0-256}. Note that we follow the convention of restricting all phases to be within the range 0 to $2 \pi$, which explains why one of the phase values is close to zero.
127:
128: Several bound states may exist for the same values of $B_2$ and $\Lambda$. In the case of $B_2 = 1.0$ and $B_3 = 1.0$ for $\Lambda = 10^8$, there exist states of energy $B_3 = 6.7502901502599$ and $B_3 = 1406.130393204$. The solutions of $f_{l0}$ for these energies are shown in Fig.~\ref{fig:FL0-Spectrum2}. All of the functions are very similar. The only real difference is in the length of the initial ``plateau.'' As the bound-state energies become larger, the flat region becomes longer, joining the cosine part near later peaks. An additional illustration of this behavior is shown in Fig.~\ref{fig:FL0-Spectrum1} using different bound-state energies.
129:
130: Using the parameters $\theta = 5.684386276089572$ and $\Lambda = 10^8$, we find the numerical solution for the function $f_{h0}(p)$ shown in Fig.~\ref{fig:FH0-2}. This is the high-momentum function corresponding to the low-momentum function in Fig.~\ref{fig:FL0-2}. Figures \ref{fig:FH0-1} to \ref{fig:FH0-4} display the high-momentum counterparts of Figs.~\ref{fig:FL0-1} to \ref{fig:FL0-4}. For all $f_{h0}$ solutions, we see the cosine behavior for $p \ll \Lambda$ and a suppression of large momentum values when $p > \Lambda$. This suppression is an effect of the gaussian behavior of $U_2$. The behavior of $f_{h0}$ is determined by the phase. Different values of the cutoff for the same phase should result in the same function. An example of this is shown in Fig.~\ref{fig:FH0-567} where $\theta = 3.0$ for $\Lambda = 10^5, 10^8,$ and $10^{11}$. Notice that all three solutions coincide.
131:
132: Combining $f_{l0}$, $f_{d0}$ and $f_{h0}$, we can see the overall leading order behavior for the function $f(p)$. Plots of $f_{l0}+ f_{h0} - f_{d0}$ are shown in Figs.~\ref{fig:F0-2} to \ref{fig:F0-4} for the four low- and high-momentum solutions previously considered. These solutions span several values of $B_2/B_3$ and show the typical behavior of any leading order solution. Figure \ref{fig:F0-2LDH} shows the three individual components for $B_2 = 0.1$ on the same plot. From this, it easy to see how cancellations occur in each region making it clear why $f_l + f_h - f_d$ is a good approximation to $f$ everywhere.
133:
134:
135: %%%%%%%%%%%%%%%%
136: %% ENERGIES %%
137: %%%%%%%%%%%%%%%%
138: \subsection{Three-Body Binding Energies}
139:
140: In Sec.~\ref{sec:analytic} we proved that the three-body bound-state spectrum is cutoff independent to leading order. This is verified numerically in Fig.~\ref{fig:B3vsLAM}. Here we have chosen $B_2 = 0$ and let $G_3$ change with $\Lambda$ so that the state $B_3 = 1.0$ is held constant. Two other states, one shallower and one deeper, are calculated as the cutoff changes. Since very small fluctuations are impossible to see in the plot, we have included Table \ref{tab:B3vsLAMBDA-1}, which shows the calculated energies for several values of $\Lambda$. Using Efimov's result that the ratio of adjacent binding energies is $\exp(2 \pi/s_0)$ when $B_2 = 0$ \cite{efimov:2,efimov:4}, the relative error for each calculation can be determined and is also given in the table. This illustrates that each energy is cutoff-independent to about 12 digits and also matches the true value to the same accuracy. As an additional example, Table \ref{tab:B3vsLAMBDA-2} shows the case of $B_2 = B_3 = 1.0$ and considers the next two deeper states as $\Lambda$ changes. The binding energies are approximately 6.75029 and 1406.13. These energies have been previously calculated by Braaten, Hammer. and Kusunoki \cite{Braaten:uefes} using a method that gives at most two digits of numerical precision. Their results are 6.8 and $1.4\times 10^3$, which match the results given here to within relative errors of $\mathcal{O}(10^{-3})$. Note that our values for these states are independent of $\Lambda$ to about 12 digits.
141:
142: \begin{table}
143: \begin{center}
144: \begin{tabular}{|c|c|c|c|c|}
145: \hline
146: $\Lambda$ & $B_3$ (Shallow) & Error & $B_3$ (Deep) & Error \\
147: \hline
148: \hline
149: 100000.00000000 & 0.0019416156131338 & 6.7e-13 & 515.03500138461 & 5.2e-13 \\
150: 738905.60989306 & 0.0019416156131358 & 3.2e-13 & 515.03500138403 & 1.6e-12 \\
151: 5459815.0033144 & 0.0019416156131435 & 4.3e-12 & 515.03500138287 & 3.9e-12 \\
152: 109663315.84284 & 0.0019416156131358 & 3.2e-13 & 515.03500138520 & 6.0e-13 \\
153: 3631550267.4246 & 0.0019416156131435 & 4.3e-12 & 515.03500138520 & 6.0e-13 \\
154: \hline
155: \end{tabular}
156: \end{center}
157: \caption[Binding energies of the next shallowest and next deepest 3-body bound states for $B_2 = 0.0$, $B_3 = 1.0$ and various cutoffs.]{\label{tab:B3vsLAMBDA-1}Binding energies of the next shallowest and next deepest 3-body bound states for $B_2 = 0.0$, $B_3 = 1.0$ and various cutoffs. Several data points from Fig.~\ref{fig:B3vsLAM} are shown along with the relative error of each result.}
158: \end{table}
159:
160: We have shown that a three-body interaction is capable of fixing a three-body binding energy. However, this does not prove that such an interaction is necessary. To investigate this claim, let us assume that $G_3 = 0$ for any value of $\Lambda$, so that the three-body interaction is removed from our equations. Using $B_2 = 0$, we calculate two values of $B_3$ for a cutoff of $\Lambda = 10^5$. The energy of each state is tracked as the value of $\Lambda$ changes. The results are shown in Fig.~\ref{fig:B3diverge}.
161:
162: \begin{table}
163: \begin{center}
164: \begin{tabular}{|c|c|c|}
165: \hline
166: $\Lambda$ & $B_3$ \#1 & $B_3$ \#2 \\
167: \hline
168: \hline
169: 100000.00000000 & 6.750290150257678 & 1406.13039320296 \\
170: 738905.60989306 & 6.750290150257678 & 1406.13039320593 \\
171: 2008553.6923187 & 6.750290150255419 & 1406.13039320345 \\
172: 14841315.910257 & 6.750290150268966 & 1406.13039320345 \\
173: 298095798.70417 & 6.750290150257678 & 1406.13039320593 \\
174: 5987414171.5197 & 6.750290150259935 & 1406.13039320345 \\
175: 44241339200.892 & 6.750290150257678 & 1406.13039320296 \\
176: \hline
177: \end{tabular}
178: \end{center}
179: \caption{\label{tab:B3vsLAMBDA-2}Binding energies of the two next deeper 3-body bound states for $B_2 = B_3 = 1.0$ and various cutoffs.}
180: \end{table}
181:
182: Both binding energies diverge like $\Lambda^2$. This is easily seen using equation (\ref{eqn:theta_tilde}). Since $G_3$ = 0, the phase must be fixed. This means
183: %
184: \begin{equation}
185: \theta = s_0 \, \ln\left(\frac{\Lambda}{\sqrt{B_3}}\right) + \tilde{\theta}(0) = \mathrm{constant},
186: \end{equation}
187: %
188: \noindent from which we can easily derive the relation $B_3 \propto \Lambda^2$. This divergent behavior is an example of the unbounded three-body spectrum predicted by L.~H.~Thomas for short-ranged interactions \cite{Thomas:shorta}. While this alone is not a proof that a three-body contact interaction is required, it should be possible to show that any other local interaction does not enter the leading order equation. This would imply $G_3$ is the only coupling capable of making any binding energy independent of the cutoff.
189:
190:
191:
192: %%%%%%%%%%%%%%%%%
193: %% COUPLINGS %%
194: %%%%%%%%%%%%%%%%%
195: \subsection{Couplings}
196:
197: Equation (\ref{eqn:PseudoCosine}) shows that the coupling $\delta_0$ should have a cosine dependence on the phase $\theta$, which is defined by $f_{d0} = \cos(s_0\ln(p/\Lambda) + \theta)$. Numerical data for $\delta_0$ as a function of the phase is shown in Fig.~\ref{fig:PSEUDOvsTHETA-2}, along with a best-fit cosine curve to verify the behavior. This behavior is independent of $B_2$ and $B_3$. Figure \ref{fig:PSEUDOvsTHETA-123} proves this by plotting the same relation for different energies. Keep in mind that $h_2$ can affect the amplitude and/or phase shift of the cosine behavior.
198:
199: We suggested in Sec.~\ref{sec:analytic} that this periodic behavior should carry over to $G_3$. Figure \ref{fig:G3vsLAMBDA-2} displays a plot of $G_3$ as a function of $\Lambda$. This data exhibits the limit-cycle behavior of the three-body coupling. As the cutoff increases, $G_3$ becomes larger and larger, eventually diverging to infinity. It then jumps to negative infinity and continues increasing. The data for $\delta_0$ and $G_3$ are combined in Fig.~\ref{fig:G3andPSEUDOvsLAMBDA-2} to show that $\delta_0 = 0$ implies $G_3 = 0$, a claim also made in Sec.~\ref{sec:analytic}. Figure \ref{fig:G3vsLAMBDA-13} uses two different sets of energies to show that the limit-cycle behavior is not dependent upon any specific bound-state values, but the positioning of the cycle is dependent upon the energies.
200:
201: If the scattering length $a$ is large compared to the effective range, the zero angular momentum sector of the three-body problem with short-ranged interactions reduces to a one-dimensional problem with a potential of $1/r^2$ in the region $r_e \ll r \ll a$ \cite{Nielsen:review}. The $1/r^2$ potential is studied in \cite{Beane:2001} where a short-range interaction is used to regulate the divergences from the long-range $1/r^2$ interaction. While one might expect to find cutoff-dependent behavior similar to what we see, they instead find a monotonically increasing coupling with no divergences. They suggest that the limit-cycle behavior is not a universal aspect of the renormalization group flow but rather an artifact of the regularization method used.
202:
203: However, this cyclic behavior has been previously observed \cite{hammer:orig} using a sharp cutoff that simply discards all momenta above $\Lambda$. Our method does not discard such momenta. It merely suppresses them by a gaussian factor. Yet the limit cycle still remains. This evidence seems to indicate that the limit-cycle behavior is universal, and the issue deserves further investigation.
204:
205: Another coupling constant we can study is $h_2$. It is shown above that $h_2$ has no effect on the binding energies to leading order, but it may affect other quantities like $G_3$ and $\delta_0$. In fact, Eq.~(\ref{eqn:PseudoCosine}) explicitly exhibits such dependence. In Figs.~\ref{fig:PSEUDOvsH2} and \ref{fig:G3vsH2}, we plot the effect of changing $h_2$ upon $\delta_0$ and $G_3$ respectively. The range of $h_2$ values is determined from Eq.~(\ref{eqn:ar2h}), reproduced here for convenience:
206: %
207: \begin{equation}
208: \frac{1}{4 \sqrt{2 \pi} \Lambda} h_2^2 + \left( \frac{2}{\sqrt{2 \pi} \Lambda} - \frac{2}{a \Lambda^2} \right) h_2 + \left( \frac{2}{a \Lambda^2} + \frac{1}{2} r_e - \frac{4}{\sqrt{2 \pi} \Lambda} \right) = 0
209: \end{equation}
210: %
211: \noindent For very large cutoffs, the effective range as computed from $h_2$ takes the form
212: %
213: \begin{equation}
214: r_e = - \frac{1}{2 \sqrt{2 \pi} \Lambda} \left( h_2^2 + 8 h_2 - 16 \right) .
215: \end{equation}
216: %
217: \noindent For illustrative purposes, we restrict ourselves to the case $r_e > 0$ so that the values of $h_2$ are limited to the range $-4(1 + \sqrt{2}) \le h_2 \le -4(1 - \sqrt{2})$. This example is chosen simply to provide a window of $h_2$ values to concentrate on.
218:
219: The cosine behavior of $\delta_0(\theta)$ should remain unchanged if $h_2$ is changed, but the amplitude and phase of the curve may differ. This is exactly what we find in Fig.~\ref{fig:PSEUDOvsTHETA-245} for the cases $h_2 = 0.0, -2.0,$ and $-5.0$. It appears as though all curves intersect at the same two points. However, we do not know if this can be explicitly proven from the leading order equations or what its significance may be.
220:
221: The general limit-cycle behavior of $G_3$ also stays the same for $h_2 = 0.0, -2.0,$ and $-5.0$. These curves are shown together in Fig.~\ref{fig:G3vsLAMBDA-245}. Changing $h_2$ results in a shift and/or flattening of the limit cycles, but the periodic behavior is still apparent.
222:
223:
224:
225:
226: %%%%%%%%%%%%%%%
227: %% FIGURES %%
228: %%%%%%%%%%%%%%%
229:
230: \begin{figure}
231: \begin{center}
232: \includegraphics[height=4.0in]{figures/FL0-2.ps}
233: \end{center}
234: \caption[Numerical solution of $f_{l0}(p)$ for the case of $B_2 = 0.1$, $B_3 = 1.0$, and $\Lambda = 10^8$.]{\label{fig:FL0-2}Numerical solution of $f_{l0}(p)$ for the case of $B_2 = 0.1$, $B_3 = 1.0$, and $\Lambda = 10^8$. The dashed line is the best-fit cosine curve that matches the high-momentum behavior.}
235: \end{figure}
236:
237: \begin{figure}
238: \begin{center}
239: \includegraphics[height=4.0in]{figures/FL0-1.ps}
240: \end{center}
241: \caption[Numerical solution of $f_{l0}(p)$ for the case of $B_2 = 0.0$, $B_3 = 1.0$, and $\Lambda = 10^8$.]{\label{fig:FL0-1}Numerical solution of $f_{l0}(p)$ for the case of $B_2 = 0.0$, $B_3 = 1.0$, and $\Lambda = 10^8$. The dashed line is the best-fit cosine curve that matches the high-momentum behavior.}
242: \end{figure}
243:
244: \begin{figure}
245: \begin{center}
246: \includegraphics[height=4.0in]{figures/FL0-3.ps}
247: \end{center}
248: \caption[Numerical solution of $f_{l0}(p)$ for the case of $B_2 = 0.5$, $B_3 = 1.0$, and $\Lambda = 10^8$.]{\label{fig:FL0-3}Numerical solution of $f_{l0}(p)$ for the case of $B_2 = 0.5$, $B_3 = 1.0$, and $\Lambda = 10^8$. The dashed line is the best-fit cosine curve that matches the high-momentum behavior.}
249: \end{figure}
250:
251: \begin{figure}
252: \begin{center}
253: \includegraphics[height=4.0in]{figures/FL0-4.ps}
254: \end{center}
255: \caption[Numerical solution of $f_{l0}(p)$ for the case of $B_2 = 1.0$, $B_3 = 1.0$, and $\Lambda = 10^8$.]{\label{fig:FL0-4}Numerical solution of $f_{l0}(p)$ for the case of $B_2 = 1.0$, $B_3 = 1.0$, and $\Lambda = 10^8$. The dashed line is the best-fit cosine curve that matches the high-momentum behavior.}
256: \end{figure}
257:
258: \begin{figure}
259: \begin{center}
260: \includegraphics[height=4.0in]{figures/FL0-256.ps}
261: \end{center}
262: \caption[Numerical solution of $f_{l0}(p)$ for the cutoffs $\Lambda = 10^5, 10^8,$ and $10^{11}$ with energies $B_2 = 0.1$ and $B_3 = 1.0$.]{\label{fig:FL0-256}Numerical solution of $f_{l0}(p)$ for the cutoffs $\Lambda = 10^5, 10^8,$ and $10^{11}$ with energies $B_2 = 0.1$ and $B_3 = 1.0$. The phases corresponding to these cutoffs are 5.016726935003937, 5.684386276089572, and 0.06886030999563049 respectively.}
263: \end{figure}
264:
265: \begin{figure}
266: \begin{center}
267: \includegraphics[height=4.0in]{figures/FL0-Spectrum2.ps}
268: \end{center}
269: \caption[Numerical solutions of $f_{l0}$ for several bound states using $B_2 = 1.0$ and $\Lambda = 10^8$.]{\label{fig:FL0-Spectrum2}Numerical solutions of $f_{l0}$ for several bound states using $B_2 = 1.0$ and $\Lambda = 10^8$. The three-body bound-state energies are 1.0, 6.7502901502599, and 1406.1303932044.}
270: \end{figure}
271:
272: \begin{figure}
273: \begin{center}
274: \includegraphics[height=4.0in]{figures/FL0-Spectrum1.ps}
275: \end{center}
276: \caption[Numerical solutions of $f_{l0}$ for several bound states using $B_2 = 0.01$ and $\Lambda = 10^5$.]{\label{fig:FL0-Spectrum1}Numerical solutions of $f_{l0}$ for several bound states using $B_2 = 0.01$ and $\Lambda = 10^5$. The three-body bound-state energies are 0.2, 63.033762419242, and 31645.59444559555.}
277: \end{figure}
278:
279: \clearpage
280:
281: \begin{figure}
282: \begin{center}
283: \includegraphics[height=4.0in]{figures/FH0-2.ps}
284: \end{center}
285: \caption[Numerical solution of $f_{h0}(p)$ for the case $\theta = 5.684386276089572$ and $\Lambda = 10^8$.]{\label{fig:FH0-2}Numerical solution of $f_{h0}(p)$ for the case $\theta = 5.684386276089572$ and $\Lambda = 10^8$. The dashed line is the best-fit cosine curve that matches the low-momentum behavior.}
286: \end{figure}
287:
288: \begin{figure}
289: \begin{center}
290: \includegraphics[height=4.0in]{figures/FH0-1.ps}
291: \end{center}
292: \caption[Numerical solution of $f_{h0}(p)$ for the case $\theta = 5.299886863265019$ and $\Lambda = 10^8$.]{\label{fig:FH0-1}Numerical solution of $f_{h0}(p)$ for the case $\theta = 5.299886863265019$ and $\Lambda = 10^8$. The dashed line is the best-fit cosine curve that matches the low-momentum behavior.}
293: \end{figure}
294:
295: \begin{figure}
296: \begin{center}
297: \includegraphics[height=4.0in]{figures/FH0-3.ps}
298: \end{center}
299: \caption[Numerical solution of $f_{h0}(p)$ for the case $\theta = 0.1031320389575355$ and $\Lambda = 10^8$.]{\label{fig:FH0-3}Numerical solution of $f_{h0}(p)$ for the case $\theta = 0.1031320389575355$ and $\Lambda = 10^8$. The dashed line is the best-fit cosine curve that matches the low-momentum behavior.}
300: \end{figure}
301:
302: \begin{figure}
303: \begin{center}
304: \includegraphics[height=4.0in]{figures/FH0-4.ps}
305: \end{center}
306: \caption[Numerical solution of $f_{h0}(p)$ for the case $\theta = 1.681619491361092$ and $\Lambda = 10^8$.]{\label{fig:FH0-4}Numerical solution of $f_{h0}(p)$ for the case $\theta = 1.681619491361092$ and $\Lambda = 10^8$. The dashed line is the best-fit cosine curve that matches the low-momentum behavior.}
307: \end{figure}
308:
309: \begin{figure}
310: \begin{center}
311: \includegraphics[height=4.0in]{figures/FH0-567.ps}
312: \end{center}
313: \caption{\label{fig:FH0-567}Numerical solution of $f_{h0}(p)$ with $\theta = 3.0$ for cutoffs $\Lambda = 10^5, 10^8,$ and $10^{11}$.}
314: \end{figure}
315:
316: \begin{figure}
317: \begin{center}
318: \includegraphics[height=4.0in]{figures/F0-2.ps}
319: \end{center}
320: \caption{\label{fig:F0-2}Numerical solution of $f_0(p)$ for the case of $B_2 = 0.1$, $B_3 = 1.0$, and $\Lambda = 10^8$.}
321: \end{figure}
322:
323: \begin{figure}
324: \begin{center}
325: \includegraphics[height=4.0in]{figures/F0-1.ps}
326: \end{center}
327: \caption{\label{fig:F0-1}Numerical solution of $f_0(p)$ for the case of $B_2 = 0.0$, $B_3 = 1.0$, and $\Lambda = 10^8$.}
328: \end{figure}
329:
330: \begin{figure}
331: \begin{center}
332: \includegraphics[height=4.0in]{figures/F0-3.ps}
333: \end{center}
334: \caption{\label{fig:F0-3}Numerical solution of $f_0(p)$ for the case of $B_2 = 0.5$, $B_3 = 1.0$, and $\Lambda = 10^8$.}
335: \end{figure}
336:
337: \begin{figure}
338: \begin{center}
339: \includegraphics[height=4.0in]{figures/F0-4.ps}
340: \end{center}
341: \caption{\label{fig:F0-4}Numerical solution of $f_0(p)$ for the case of $B_2 = 1.0$, $B_3 = 1.0$, and $\Lambda = 10^8$.}
342: \end{figure}
343:
344: \begin{figure}
345: \begin{center}
346: \includegraphics[height=4.0in]{figures/F0-2LDH.ps}
347: \end{center}
348: \caption{\label{fig:F0-2LDH}Numerical solutions of $f_{l0}$, $f_{d0}$, and $f_{h0}$ for the case of $B_2 = 0.1$, $B_3 = 1.0$, and $\Lambda = 10^8$.}
349: \end{figure}
350:
351: \begin{figure}
352: \begin{center}
353: \includegraphics[height=4.0in]{figures/B3vsLAM.ps}
354: \end{center}
355: \caption{\label{fig:B3vsLAM}Binding energies of the next shallowest and next deepest 3-body bound states as a function of the cutoff $\Lambda$ for $B_2 = 0$ and $B_3 = 1.0$.}
356: \end{figure}
357:
358: \begin{figure}
359: \begin{center}
360: \includegraphics[height=4.0in]{figures/B3diverge.ps}
361: \end{center}
362: \caption[Binding energies of two three-body bound states for the case of $B_2 = 0$ and $G_3 = 0$.]{\label{fig:B3diverge}Binding energies of two three-body bound states for the case of $B_2 = 0$ and $G_3 = 0$. Notice that without a three-body interaction, all the binding energies diverge as the cutoff is increased.}
363: \end{figure}
364:
365: \begin{figure}
366: \begin{center}
367: \includegraphics[height=4.0in]{figures/PSEUDOvsTHETA-2.ps}
368: \end{center}
369: \caption[The coupling $\delta_0$ as a function of the phase $\theta$ for the case $B_2 = 0.5$ and $B_3 = 1.0$.]{\label{fig:PSEUDOvsTHETA-2}The coupling $\delta_0$ as a function of the phase $\theta$ for the case $B_2 = 0.5$ and $B_3 = 1.0$. The dashed line is the best-fit cosine curve. The data matches to within an error of about $10^{-10}$.}
370: \end{figure}
371:
372: \begin{figure}
373: \begin{center}
374: \includegraphics[height=4.0in]{figures/PSEUDOvsTHETA-123.ps}
375: \end{center}
376: \caption[The coupling $\delta_0$ as a function of the phase $\theta$ for three different sets of $B_2$ and $B_3$.]{\label{fig:PSEUDOvsTHETA-123}The coupling $\delta_0$ as a function of the phase $\theta$ for three different sets of $B_2$ and $B_3$. Notice that all curves are identical to within numerical error.}
377: \end{figure}
378:
379: \clearpage
380:
381: \begin{figure}
382: \begin{center}
383: \includegraphics[height=4.0in]{figures/G3vsLAMBDA-2.ps}
384: \end{center}
385: \caption[The dimensionless three-body coupling $G_3$ as a function of the cutoff $\Lambda$ for $B_2 = 0.5$ and $B_3 = 1.0$.]{\label{fig:G3vsLAMBDA-2}The dimensionless three-body coupling $G_3$ as a function of the cutoff $\Lambda$ for $B_2 = 0.5$ and $B_3 = 1.0$. The limit-cycle behavior of $G_3$ is evident from the fact that it is periodic in $\ln(\Lambda)$.}
386: \end{figure}
387:
388: \begin{figure}
389: \begin{center}
390: \includegraphics[height=4.0in]{figures/G3andPSEUDOvsLAMBDA-2.ps}
391: \end{center}
392: \caption[Dependence of $G_3$ and $\delta_0$ on the cutoff $\Lambda$ for the case of $B_2 = 0.5$ and $B_3 = 1.0$.]{\label{fig:G3andPSEUDOvsLAMBDA-2}Dependence of $G_3$ and $\delta_0$ on the cutoff $\Lambda$ for the case of $B_2 = 0.5$ and $B_3 = 1.0$. The values for $\delta_0$ have been multiplied by a constant factor of 150 to make the cosine behavior visible. Both couplings are shown to emphasize that they equal zero at the same points.}
393: \end{figure}
394:
395: \begin{figure}
396: \begin{center}
397: \includegraphics[height=4.0in]{figures/G3vsLAMBDA-13.ps}
398: \end{center}
399: \caption[$G_3$ as a function of the cutoff $\Lambda$ for two different sets of $B_2$ and $B_3$.]{\label{fig:G3vsLAMBDA-13}$G_3$ as a function of the cutoff $\Lambda$ for two different sets of $B_2$ and $B_3$. Both curves have the same limit-cycle behavior, but the difference in energies causes one to be shifted relative to the other.}
400: \end{figure}
401:
402: \begin{figure}
403: \begin{center}
404: \includegraphics[height=4.0in]{figures/PSEUDOvsH2.ps}
405: \end{center}
406: \caption{\label{fig:PSEUDOvsH2}The ratio $\delta_0(h_2)/\delta_0(0)$ as a function of $h_2$ for $\theta = 1.0$ and $\Lambda = 10^5$.}
407: \end{figure}
408:
409: \begin{figure}
410: \begin{center}
411: \includegraphics[height=4.0in]{figures/G3vsH2.ps}
412: \end{center}
413: \caption{\label{fig:G3vsH2}The ratio $G_3(h_2)/G_3(0)$ as a function of $h_2$ for $\theta = 1.0$ and $\Lambda = 10^5$.}
414: \end{figure}
415:
416: \clearpage
417:
418: \begin{figure}
419: \begin{center}
420: \includegraphics[height=4.0in]{figures/PSEUDOvsTHETA-245.ps}
421: \end{center}
422: \caption[The coupling $\delta_0$ as a function of $\theta$ for several values of $h_2$.]{\label{fig:PSEUDOvsTHETA-245}The coupling $\delta_0$ as a function of $\theta$ for several values of $h_2$. The curves are all cosines with different amplitudes and phases.}
423: \end{figure}
424:
425: \begin{figure}
426: \begin{center}
427: \includegraphics[height=4.0in]{figures/G3vsLAMBDA-245.ps}
428: \end{center}
429: \caption[$G_3$ as a function of $\Lambda$ for several values of $h_2$.]{\label{fig:G3vsLAMBDA-245}$G_3$ as a function of $\Lambda$ for several values of $h_2$. A change in $h_2$ results in a shift and/or flattening of the limit-cycle curve, but the general periodic form is the same.}
430: \end{figure}