1:
2: \documentclass[prc,aps,twocolumn,floatfix]{revtex4}
3:
4: \usepackage[dvips]{graphicx}
5: \usepackage{epsfig}
6: \usepackage{pst-plot}
7: \usepackage{bm}
8: \usepackage{amsfonts}
9:
10: \draft
11: \begin{document}
12: \title{Generalized contour deformation method in momentum space:
13: two-body spectral structures and scattering amplitudes.}
14: \author{G. Hagen}
15: \affiliation{Department of physics, University of Bergen, N-5000 Bergen, Norway}
16: \author{M.~Hjorth-Jensen}
17: \affiliation{Department of Physics and Centre of Mathematics for Applications,
18: University of Oslo, Norway}
19: \author{J. S. Vaagen}\affiliation{Department of Physics, University of Bergen,
20: N-5000 Bergen, Norway}
21: \date{\today}
22:
23: \begin{abstract}
24: A generalized contour deformation method (GCDM) which combines complex rotation
25: and translation in momentum space, is discussed. GCDM gives accurate results
26: for bound, virtual (antibound), resonant and
27: scattering states starting with a realistic nucleon-nucleon interaction.
28: It provides a basis for full off-shell $t$-matrix calculations
29: both for real and complex input energies. GCDM competes favorably with analytic
30: continuation. Results for both spectral structures and scattering amplitudes
31: compare perfectly well with exact values for the separable Yamaguchi potential.
32: Accurate calculation of virtual states in the Malfliet-Tjon and the realistic
33: CD-Bonn nucleon-nucleon interactions are presented.
34:
35: GCDM is also a promising method for the computation of in-medium properties such as
36: the resummation of particle-particle and particle-hole diagrams
37: in infinite nuclear matter. Implications for in-medium scattering are discussed.
38: \end{abstract}
39:
40: \pacs{PACS number(s): 13.75.Cs, 24.10.Cn, 24.30.Gd}
41: \maketitle
42:
43:
44:
45:
46:
47: \section{Introduction}
48: \label{sec:introduction}
49: The study of two-body resonant structures has a long history in
50: theoretical physics, and there exists a variety of methods, described in
51: textbooks such as \cite{newton,kukulin}. Among the more popular methods
52: we have the complex scaling method (CSM) and the method based on
53: analytic continuation in the coupling constant (ACCC).
54:
55: In this work we consider a new approach formulated for integral equations
56: in momentum space. The method is based on
57: deforming the contour integrals in momentum space, known as
58: the contour deformation or distortion
59: method (CDM). It has been shown in Ref.~\cite{afnan1} that a \emph{contour rotation} in
60: momentum space is equivalent to a rotation of the corresponding differential equation in
61: coordinate space. The coordinate space analog is often referred to as the
62: \emph{dilation group transformation}, or \emph{complex scaling}. The
63: \emph{dilation group transformation} was first discussed and formulated in
64: Refs.~\cite{abc,abc1}, and was developed to examine the spectrum of the Green's
65: function on the second energy sheet.
66:
67: Complex scaling in coordinate space has for a long time
68: been used extensively in atomic and molecular physics, see
69: Ref.~\cite{moise}.
70: During the last decade it has also been applied in nuclear physics,
71: as interest in loosely bound nuclear halo systems has grown, see for
72: example Refs.~\cite{csoto, garrido, imante}. Complex scaling
73: in coordinate space is usually based on a variational method \cite{moise}, and an
74: optimal variational basis and scaling parameters have to be searched for. One of the disadvantages
75: of the coordinate space approach is that the boundary conditions have to be
76: built into the equations, and convergence may be slow if the basis does not
77: mirror the physical outgoing boundary conditions well.
78:
79: There are several advantages in considering the contour deformation method
80: in momentum space. First, most realistic potentials derived from field theoretical
81: considerations are given explicitly in momentum space. Secondly, the boundary conditions
82: are automatically built into the integral equations. Moreover,
83: the Gamow states \cite{kukulin} in momentum space
84: are non-oscillating and rapidly decreasing, even for Gamow
85: states with large widths far from the real
86: energy axis,
87: as opposed to the complex scaled coordinate space counterpart. These states
88: are represented by
89: strongly oscillating and exponentially decaying functions.
90: Finally, numerical procedures are
91: often easier to implement and check. Convergence is easily obtained by just increasing
92: the number of integration points in the numerical integration.
93:
94: The contour deformation method (CDM) formulated in momentum space is not new
95: in nuclear physics. It was studied and applied in the 1960`s and 1970`s,
96: see for example Refs.~\cite{brayshaw,nuttal,stelbovics}, especially in the field of
97: three-body systems. These references
98: applied a \emph{contour rotation} method in momentum space. By restricting
99: oneself to a rotated contour certain limitations and restrictions however appear in the
100: equations, determined by the analytical structure of the integral kernels and potentials.
101: We will study an alternative approach, by considering an extended deformation of the integration
102: contour based on rotation followed by translation in the complex momentum plane. This choice
103: of contour can be regarded as a special case of the \emph{Berggren class} of contours
104: \cite{berggren}. Berggren \cite{berggren} and later Lind \cite{lind} studied various
105: completeness relations derived by analytic continuation of the Newton completeness
106: relation \cite{newton} to the complex plane. The Berggren completeness
107: includes discrete summation
108: over resonant as well as bound states. Our choice of contour differ from the
109: \emph{Berggren class} of contours in that the contour approaches infinity along complex
110: rays in the complex $k$-plane
111: as opposed to the various contours studied by Lind \cite{lind} which approach
112: infinity along the real $k$-axis.
113: By transforming the momentum space Schr\"odinger
114: equation onto the rotated followed by translated contour,
115: we will show that we are able to expose and explore more of the physical
116: interesting area
117: on the second energy sheet, i.e., the choice of contour enables us to study both
118: bound, virtual and resonant states.
119: If one restricts the deformation to a rotation of the contour,
120: as studied in Refs.~\cite{brayshaw,nuttal,stelbovics,nuttal1,tikto},
121: we are not able to
122: expose virtual states in the
123: energy spectrum, since the maximum rotation angle does not allow rotation into the
124: third quadrant of the complex momentum plane. This limitation is sometimes used as an
125: argument for advocating different approaches, such as the ACCC method, see the recent work
126: of Aoyama \cite{aoyama}.
127: By distorting the contour by
128: rotation and translation we
129: are able to introduce a new feature to the complex scaling method, namely
130: \emph{accurate calculation of virtual states as well as bound and resonant states}.
131: Our method represents also an alternative to the \emph{exterior complex scaling} method.
132: The \emph{exterior complex scaling} method
133: was formulated to avoid intrinsic non-analyticities of the potential, and in this way
134: calculation of resonances in \emph{non-dilation} analytic potentials are possible, see
135: Ref.~\cite{moise} and references therein.
136: By the
137: rotated and translated contour choice the CDM is a preferable method compared to
138: all other known methods, such as the ACCC method.
139:
140: The contour deformation method has also been applied to the solution of the
141: full off-shell scattering amplitude ($t$-matrix), see Refs.~\cite{afnan1,nuttal, stelbovics,afnan}.
142: By rotating the integration contour, an integral equation was obtained with a
143: compact integral kernel. This has numerical advantages as the kernel is no longer
144: singular. As discussed in Ref.~\cite{nuttal}, a rotation of the contour gives certain
145: restrictions on the rotation angle and maximum incoming/outgoing momentum in
146: the scattering amplitude. We will again show that our extended choice of contour in momentum
147: space avoids all these limitations and that an accurate calculation of the
148: scattering amplitude can be obtained.
149:
150: Thus, the method we will advocate allows us to give
151: an accurate calculation of the full energy spectrum. Moreover, it yields
152: a powerful method for calculating the full off-shell complex scattering amplitude ($t$-matrix).
153: It is also rather straightforward to extend
154: this scheme to in-medium scattering in e.g., infinite nuclear matter.
155:
156: In section \ref{sec:formalism} we outline the contour deformation method in momentum space, and
157: in section \ref{sec:tmatrix} the spectral representation of the full off-shell $t$-matrix is
158: given along with the deformed integration contour. Section \ref{sec:yama} presents
159: as a test case
160: a simple separable interaction which admits analytical solutions for both the energy spectrum
161: and the $t$-matrix. Section \ref{sec:results} gives numerical calculations of energy spectra and
162: the $t$-matrix by the contour deformation method. We present numerical results for
163: the Yamaguchi \cite{yamaguchi}, Malfliet-Tjon \cite{malfliet}
164: and the charge-dependent Bonn (CD-Bonn) \cite{machleidt} interactions.
165: Calculation of virtual states of the CD-Bonn interaction is not known by us to have
166: been performed previously.
167:
168:
169:
170: \section{Theoretical framework}
171: \label{sec:formalism}
172: We will in the following use natural units $\hbar = c = 1$.
173: The two-body momentum space Schr\" odinger equation in a partial wave decomposition reads
174: \begin{equation}
175: \label{eq:eq1}
176: {k^{2}\over 2\mu}\psi_{nl}(k) + {2\over\pi}\int_{0}^{\infty}
177: dq q^{2}V_{l}(k,q)\psi_{nl}(q) = E_{nl}\psi_{nl}(k).
178: \end{equation}
179: For the sake of simplicity we assume here that
180: the interaction is spherically symmetric and local in coordinate space and
181: without tensor components
182: and/or spin-orbit coupling. When solving the correspondning equations for a more realistic
183: nucleon-nucleon interaction below, these degrees of freedom will be accounted for.
184:
185: The Fourier-Bessel transform of the potential $V_{l}(r)$ in coordinate space is given by
186: \begin{equation}
187: \label{eq:eq2}
188: V_{l}(k,k') = \int_{0}^{\infty}dr r^{2} j_{l}(kr)j_{l}(k'r)V_{l}(r).
189: \end{equation}
190: The momentum space Schr\"odinger equation in Eq.~(\ref{eq:eq1}) (with real momenta)
191: corresponds to a hermitian Hamiltonian. The eigenvalues will in this case
192: always be real, corresponding to discrete bound states ($E_{nl} < 0 $) and
193: a continuum of scattering
194: states ($E_{nl} > 0$). The eigenstates form a complete set, and for
195: a given partial wave $l$ the completeness relation, more precisely known
196: as \emph{resolution of unity},
197: can be written \cite{newton}
198: \begin{equation}
199: \label{eq:unity1}
200: {\bf 1} = \sum _{n}\vert\psi_{nl}\rangle\langle\psi_{nl}\vert +
201: {1\over 2}\int_{-\infty}^{\infty} dk \vert\psi_{l}(k)\rangle\langle\psi_{l}(k)\vert,
202: \end{equation}
203: The infinite space spanned by this basis is given by all square integrable functions on the real
204: energy axis, known as the $L^{2}$ space, which forms a Hilbert space.
205: Resonant and virtual states can never be obtained
206: by directly solving Eq.~(\ref{eq:eq1}), as it stands. In a sense, one can say that
207: the spectrum of a hermitian Hamiltonian does not display all information
208: about the physical system.
209:
210: In this paper we study and explore the resonant and virtual state spectra by the
211: contour deformation method. This is essentially a transformation of
212: Eq.~(\ref{eq:eq1}) into the lower-half complex $k$-plane. Such a transformation of
213: Eq.~(\ref{eq:eq1}) can be obtained by an analytic continuation of the completeness
214: relation of Eq.~(\ref{eq:unity1}) to the complex $k$-plane. One can consider the integral in
215: Eq.~(\ref{eq:unity1}) as an integral over the contour $\Gamma = S + C $, where
216: the contour $C$ is defined on the real $k$-axis from $-\infty$ to $ +\infty$ and
217: the contour $S$ is given by an
218: infinite semicircle in the upper-half complex $k$-plane closing the contour $\Gamma$.
219: In Ref.~\cite{lind} completeness relations for various \emph{inversion symmetric} contours
220: in the complex $k$-plane were derived and discussed. \emph{Inversion symmetric} contours are
221: defined by the following: if $z$ is on $C$, then $-z$ is also on $C$.
222: The derivation given in Ref.~\cite{lind} was based on analytic
223: continuation by deforming the contour $C$ defined on the real $k$-axis.
224: These completeness relations can
225: be regarded as a generalization of the Berggren completeness relation \cite{berggren}.
226: We will therefore label the \emph{inversion symmetric} contours discussed in \cite{lind}
227: as the \emph{extended Berggren class} of contours.
228: The completeness relation of Berggren was an extension of the completeness relation
229: of Eq.~(\ref{eq:unity1}) through the inclusion of a finite sum over discrete resonant states.
230: By redefining the completeness relation on distorted contours in the complex $k$ plane, one
231: can show by using Cauchy's residue theorem that
232: the summation over discrete states will in general include bound, virtual and
233: resonant states \cite{lind}. The eigenfunctions will form a \emph{biorthogonal} set, and
234: the normalization follows the generalized $c$-product \cite{moise,lind}
235: \begin{equation}
236: \label{eq:norm}
237: \langle\langle \psi_{nl}\vert \psi_{n'l} \rangle\rangle \equiv
238: \langle \psi_{nl}^{*}\vert \psi_{n'l}\rangle = \delta_{n,n'}.
239: \end{equation}
240: The most general completeness relation on an arbitrary \emph{inversion symmetric} contour
241: $C = C^{+} + C^{-} $ can then be written as
242: \begin{equation}
243: \label{eq:unity2}
244: {\bf 1} = \sum _{n\in \bf{C}}\vert\psi_{nl}\rangle\langle\psi_{nl}^{*}\vert +
245: \int_{C^{+}} dz \vert\psi_{l}\rangle\langle\psi_{l}^{*}\vert,
246: \end{equation}
247: where $C^{+} $ is the distortion of the positive real $k$-axis, and $C^{-} $ is the
248: distortion of the negative real $k$-axis. The symmetry of the integrand has been
249: taken into account, that is
250: \[
251: \int_{C^{-}} dz \vert\psi_{l}\rangle\langle\psi_{l}^{*}\vert =
252: \int_{C^{+}} dz \vert\psi_{l}\rangle\langle\psi_{l}^{*}\vert .
253: \]
254: The summation is over all discrete states (bound, virtual and
255: resonant states) located in the domain $\bf{C}$,
256: defined as the area above the contour $C$,
257: and the integral is over the non-resonant complex continuum defined on $C^{+}$.
258: The space spanned by the basis given in Eq.~(\ref{eq:unity2})
259: includes all square integrable functions defined in the domain $\bf{C}$.
260: The complete basis could then be used to expand resonant and virtual states defined in
261: the region above the distorted contour. Such a complete basis is more
262: flexible than a complete basis defined for only real energies.
263: From the general completeness relation (\ref{eq:unity2})
264: one can deduce the corresponding eigenvalue problem, $H\vert\psi\rangle = E\vert\psi\rangle $.
265: The Hamilitonian will in this case be complex and non-hermitian,
266: as Gamow and virtual states are included in the spectrum.
267:
268: In close analogy with the above discussion on completeness relations, the momentum space
269: Schr\"odinger equation, see Eq.~(\ref{eq:eq1}), defined on the positive real $k$-axis
270: can be continued to the lower-half complex $k$-plane.
271: Eq.~(\ref{eq:eq1}) is a \emph{Fredholm} integral equation of the
272: second kind. Continuing Eq.~(\ref{eq:eq1}) to the lower-half $k$-plane, the general
273: rule that the moving singularities of the integral kernel must not intercept the integration
274: contour must be obeyed, see for example Ref.~\cite{kukulin}.
275: The choice of distorted contour will for each partial wave be based on the
276: \emph{a posteriori} knowledge of poles in the scattering matrix.
277: By considering the transformed momentum space version of Eq.~(\ref{eq:eq1}), the
278: choice of contour will in addition be dictated by the analytic structure of the integral kernel
279: and potential. The contour must be chosen in such a way that singularities in the
280: potential are located outside the closed integration contour.
281:
282: In the following we study the analytic continuation of Eq.~(\ref{eq:eq1}) onto
283: two distorted contours $C_{1}^{+}$ and $C_{2}^{+}$. These contours can be regarded
284: as a special case of the
285: \emph{extended Berggren class} of contours.
286: The contour $C_{1}^{+}$ is obtained
287: by a phase transformation (rotation) into the lower-half complex $k$-plane while
288: the second contour $C_{2}$ will be based on rotation
289: followed by translation in the lower-half complex $k$-plane. These contours differ in an
290: important aspect from the
291: \emph{extended Berggren class} of
292: contours, in that they approach infinity along complex rays, and not along the real $k$-axis.
293: It has previously been assumed as a requirement for the choice of distorted contours
294: that they approach infinity along the real $k$-axis, see for example Ref.~\cite{betan}.
295:
296: First we consider a contour $C^{+}_{1}$ given by
297: two line segments $L_{1}$ and $ L_{2}$. Line $L_{1}$ is
298: given by $z_{1} = k\exp{(-i \theta)}$ where $k\in [0,k_{max}]$, $L_{2}$ by
299: $z_{2} = k_{max} \exp{(-i\theta)}$ ($k$ still real).
300: One can easily show that for an exponentially bounded potential in coordinate
301: space the integral in Eq.~(\ref{eq:eq1}) along the arc $L_{2}$
302: will go to zero for $k_{max}\rightarrow \infty $. In this case the contour $C_{1}^{+} $
303: reduces to the line $L_{1}$.
304: Fig.~\ref{fig:contour1} shows the contour $C_{1}^{+}$ along with the exposed
305: and excluded two-body spectrum in the complex $k$-plane, which this contour choice implies.
306: The discrete spectrum consisting of bound, virtual and resonant states corresponds to poles of the
307: scattering matrix $S_{l}(k)$ in the complex $k$-plane.
308: The contour
309: $C_{1}^{+}$ is part of the \emph{inversion symmetric} contour $C_{1} = C_{1}^{+} +
310: C_{1}^{-}$, as can be seen in Fig.~\ref{fig:contour1}.
311: \begin{figure}[hbtp]
312: \begin{center}
313: \resizebox{8cm}{5cm}{\epsfig{file=fig1.eps}}
314: \end{center}
315: \caption{Contour $ C_{1}^{+} = L_{1} + L_{2} $ is given by the solid line, while
316: the contour $C_{1}^{-} $ is given by the dashed line. The contour $C_{1} = C_{1}^{+}+C_{1}^{-}$ is clearly
317: \emph{inversion symmetric}. The two body spectrum which is exposed by this contour is marked by
318: filled circles $ \bullet $ and the excluded spectrum by open circles $\circ $.
319: The full spectrum includes bound states (BS),
320: virtual (VS), decay (DS) and capture (CS) resonant states.}
321: \label{fig:contour1}
322: \end{figure}
323: For a potential which is analytic in the region above the contour $C_{1}^{+}$ and below
324: the real $k$-axis, we can then derive the transformed
325: Eq.~(\ref{eq:eq1}) on contour $C_{1}^{+} $
326: \begin{eqnarray}
327: \nonumber
328: \left( \exp{(-2i\theta)}{k^{2}\over 2\mu} - E_{nl}\right) \psi_{nl}(z) + \\
329: \label{eq:eq3}
330: \exp{(-3i\theta)}{2\over\pi}\int_{0}^{\infty}
331: dq q^{2}V_{l}(z,z')\psi_{nl}(z') = 0,
332: \end{eqnarray}
333: where $z=k\exp{(-i\theta)}$ and $z' = q\exp{(-i\theta)}$. Eq.~(\ref{eq:eq3}) is the
334: momentum space version of the complex scaled Schr\"odinger equation in
335: coordinate space, discussed in e.g., Ref.~\cite{afnan1}.
336: A rotation in momentum space,
337: $k\exp{(-i\theta)} $, is equivalent to the complex scaling $ r\exp{(i\theta)} $
338: in coordinate space. The phase-transformation $k\: \rightarrow\: k\exp{(-i\theta)} $
339: is formally a similarity transformation, see for example Ref.~\cite{brown}.
340: The restriction on the rotation angle $\theta$ for the phase-transformation
341: $k\: \rightarrow\: k\exp{(-i\theta)} $ is given by the region of
342: analyticity of the potential.
343: It has been shown in Refs.~\cite{abc,abc1} that such a transformation does not alter
344: the location of bound states in the
345: system; the bound state spectrum is invariant under such transformations. This is also
346: clear from the discussion above on completeness relations.
347: The continuum is shifted into the lower-half $k$-plane, while resonant states
348: will occur as long as they are located above the integration contour.
349: The transformed Eq.~(\ref{eq:eq3}) represents a non-hermitian Hamiltonian. The eigenvalues
350: are therefore in general complex, and the corresponding eigenfunctions form
351: a \emph{biorthogonal} set with the normalization condition given by Eq.~(\ref{eq:norm}).
352:
353: Next we consider the contour obtained by rotation followed by translation
354: in the lower-half complex $k$-plane. The contour $C^{+}_{2}$ consists of three line segments.
355: The line segment $L_{1}$ is given by a rotation $z_{1} =
356: k_{1}\exp{(-i\theta)}$ where
357: $k_{1} \in [0, b]$, $L_{2} $ is given by a translation
358: $z_{2} = k_{2} -ib\sin (\theta)$ where $k_{2} \in [b\cos (\theta), k_{max}]$ and
359: $b$ determines the translation into the lower-half $k$-plane and
360: $L_{3}$ by $z_{3} = k_{max} -ic$ where $c\in [b\sin (\theta) ,0]$.
361: For $k_{max} \: \rightarrow \: \infty $ the contribution to the integral in Eq.~(\ref{eq:eq1})
362: along the line segment $L_{3}$ will vanish, and the contour $C_{2}^{+}$ reduces to
363: the line segments $L_{1}$ and $L_{2}$.
364: Fig.~\ref{fig:contour2} shows the contour $C_{2}^{+} = L_{1} + L_{2} + L_{3}$
365: along with the exposed
366: and excluded two-body spectrum which this contour choice implies. The contour
367: $C_{2}^{+}$ is part of the \emph{inversion symmetric} contour $C_{2} = C_{2}^{+} +
368: C_{2}^{-}$ clearly seen in the figure.
369: \begin{figure}[hbtp]
370: \begin{center}
371: \resizebox{8cm}{5cm}{\epsfig{file=fig2.eps}}
372: \end{center}
373: \caption{Contour $ C_{2}^{+} = L_{1} + L_{2} + L_{3} $ is given by the solid line, while
374: the contour $C_{1}^{-} $ is given by the dashed line. The contour $C_{1} = C_{1}^{+}+C_{1}^{-}$ is clearly
375: \emph{inversion symmetric}. The two body spectrum which is exposed by this contour is marked by
376: filled circles
377: $ \bullet $ and the excluded spectrum by open circles $\circ $.
378: The full spectrum includes bound states (BS),
379: virtual (VS), decay (DS) and capture (CS) resonant states. }
380: \label{fig:contour2}
381: \end{figure}
382: We can now transform the momentum space Schr\"odinger equation (\ref{eq:eq1}) onto the distorted contour
383: $C_{2}^{+} $ given by the lines
384: $L_{1}$ and $L_{2}$, i.e., $k,k' \: \rightarrow \: z,z' $.
385: The integral in Eq.~(\ref{eq:eq1}) will in this case couple the
386: complex momenta $z_{1}$ and $z_{2}$. The coupled transformed Schr\"odinger equation then reads
387: \begin{widetext}
388: \begin{equation}
389: \label{eq:eq4}
390: {1\over2\mu}
391: \left( \begin{array}{cc}
392: z_{1}^{2} & 0 \\
393: 0 & z_{2}^{2}
394: \end{array} \right)
395: \left( \begin{array}{c}
396: \psi_{nl}(z_{1}) \\
397: \psi_{nl}(z_{2})
398: \end{array}\right)
399: +
400: {2\over\pi}
401: \left( \begin{array}{cc}
402: \int_{L_{1}} d{z'}_{1}\:{z'}_{1}^{2}V_{l}(z_{1},{z'}_{1}) &
403: \int_{L_{2}} d{z'}_{2}\:{z'}_{2}^{2}V_{l}(z_{1},{z'}_{2}) \\
404: \int_{L_{1}} d{z'}_{1}\:{z'}_{1}^{2}V_{l}(z_{2},{z'}_{1}) &
405: \int_{L_{2}} d{z'}_{2}\:{z'}_{2}^{2}V_{l}(z_{2},{z'}_{2})
406: \end{array} \right)
407: \left( \begin{array}{c}
408: \psi_{nl}({z'}_{1}) \\
409: \psi_{nl}({z'}_{2})
410: \end{array}\right)
411: =
412: E_{nl}
413: \left( \begin{array}{c}
414: \psi_{nl}(z_{1}) \\
415: \psi_{nl}(z_{2})
416: \end{array}\right).
417: \end{equation}
418: \end{widetext}
419: This equation is again a non-hermitian Hamiltonian. The basis of eigenstates
420: forms a \emph{biorthogonal} set, and the normalization is again given by
421: Eq.~(\ref{eq:norm}). The completeness relation given in Eq.~(\ref{eq:unity2}), will
422: include a discrete sum over bound, virtual and resonant states, and the integration along
423: $L = L_{1}+L_{2}$ is over the complex non-resonant energy continuum.
424:
425: Fig.~\ref{fig:fig2} shows a plot of the energy spectra of the phase-transformed
426: Eq.~(\ref{eq:eq3}) and the rotated + translated Eq.~(\ref{eq:eq4}) in the complex
427: energy plane, for a two-body potential supporting bound and resonant states.
428: \begin{figure}[hbtp]
429: \begin{center}
430: \resizebox{8cm}{4cm}{\epsfig{file=fig3.eps}}
431: \end{center}
432: \caption{General energy spectra in the complex energy plane for the
433: complex transformed
434: Schr\"odinger Eqs.~(\ref{eq:eq3}) and (\ref{eq:eq4}), Fig.a and Fig.b,
435: respectively.
436: bound states are denoted by (BS), decay resonant states by (DS) and the complex
437: continuum by (CC). }
438: \label{fig:fig2}
439: \end{figure}
440:
441: Whether one chooses to solve the Schr\"odinger equation on the contour $C_{1}^{+}$
442: or on the contour $C_{2}^{+}$ depends on the problem under consideration. For
443: potentials which are analytic in the entire lower-half $k$-plane, solving
444: along contour $C_{1}^{+}$ is numerically the most straightforward method.
445: In most cases the potential has singularities in
446: the lower-half $k$-plane, and solving on the contour $C_{2}^{+}$ enables us to
447: avoid the singularities of the potential, while still being able to study
448: resonant structures
449: in the system. Another advantage is that there is no restriction on the
450: rotation angle $\theta $ as long as the contour is chosen so that the poles of
451: the potential are
452: located outside the contour. If a potential is of such an analytic structure
453: that we are allowed to rotate into the third quadrant of the complex
454: $k$-plane (see Fig.~\ref{fig:fig3}), virtual states will
455: appear in the calculated spectra, as long as the potential supports virtual
456: states. By solving the Schr\"odinger equation on the distorted contour $C_{2}^{+}$
457: rotated into the third quadrant of the complex $k$-plane, we expose
458: a part of the negative imaginary $k$-axis where virtual states may be located,
459: while at the same time excluding
460: a part of the positive imaginary $k$-axis where bound states may be located.
461: This reminds us that the
462: contour should be chosen relative to the partial wave component under study. This means that a
463: separate analysis has to be made for each partial wave.
464:
465: \begin{figure}[hbtp]
466: \begin{center}
467: \resizebox{8cm}{4cm}{\epsfig{file=fig4.eps}}
468: \end{center}
469: \caption{Plot of integration contour $C_{2}^{+}$ where the potential singularities
470: are located in the marked area in Fig.~a. The rotation angle $\theta$ is greater
471: than $\pi /2$ and the integration contour encloses a virtual state (A) in the $k$-plane.
472: The corresponding energy spectrum in the complex energy plane in Fig.~b illustrates how the
473: virtual state (VS) is exposed in the spectrum and the complex non-resonant continuum (CC).}
474: \label{fig:fig3}
475: \end{figure}
476:
477: \section{Two-body $t$-matrix}
478: \label{sec:tmatrix}
479: In this section we will discuss different procedures for solving the full
480: off-shell $t$-matrix, and hence the full two-body scattering problem.
481: We will consider the general mathematical case where the incoming energy is
482: allowed to take non-physical values, i.e., the input energy is complex. This has
483: relevance for nuclear medium studies where the input energy is in general
484: complex. Two methods for solving the full off-shell
485: $t$-matrix for arbitrary complex energy $\omega $ are outlined.
486:
487: The $t$-matrix is defined in operator form by
488: \begin{equation}
489: \label{eq:t1}
490: t(\omega) = V + VG(\omega)V,
491: \end{equation}
492: or
493: \begin{equation}
494: \label{eq:t2}
495: t(\omega) = V + VG_{0}(\omega)t(\omega).
496: \end{equation}
497: Here $\omega$ is the incoming energy, $ G(\omega) $ is the resolvent,
498: commonly known as the Green's operator, and $G_{0}(\omega) $ the
499: corresponding free Green's operator. In operator form they are defined by
500: \begin{eqnarray}
501: G_{0}(\omega) & = & {1\over \omega - H_{0}}, \\
502: \label{eq:greensfunc}
503: G(\omega) & = & {1\over \omega - H }.
504: \end{eqnarray}
505: The term $H_{0}$ is the kinetic energy operator and $H$ the full two-body
506: Hamiltonian. By expanding the unit operator on a complete set of
507: physical eigenstates of $H$ given in Eq.~(\ref{eq:unity1}),
508: we can write the Green's operator as
509: \begin{equation}
510: \label{eq:spectral1}
511: G(\omega) = \sum _{b} {\vert\psi _{b}\rangle\langle \psi _{b}\vert\over \omega - E_{b}} + \int _{0}^{\infty }dE_{c}{\vert\psi_{c}\rangle\langle\psi_{c}\vert\over \omega - E_{c}}.
512: \end{equation}
513: This is the spectral decomposition of the Greens's function.
514: Here $b$ denotes the discrete bound state spectrum and $c$ the positive
515: energy continuum. By projecting $t(\omega)$ on momentum states, and decomposing
516: into partial waves, the $t$-matrix elements $t_{l}(k,k';\omega) $ can be
517: expressed as 1-dimensional integral equations. Depending on whether we start
518: from (\ref{eq:t1}) or (\ref{eq:t2}) we get the \emph{spectral}
519: or the \emph{Fredholm} representation of the $t$-matrix. The contour
520: deformation method will be applied to the \emph{spectral} representation of
521: the $t$-matrix while the solution of the \emph{Fredholm} representation of the
522: $t$-matrix will
523: be based on the standard principal value prescription.
524:
525: By using Eq.~(\ref{eq:t1}) and Eq.~(\ref{eq:spectral1}) we get the \emph{spectral} representation of the
526: $t$-matrix by inserting the expansion of the Green's function on a
527: complete set of states (\ref{eq:spectral1}), giving
528: \begin{widetext}
529: \begin{equation}
530: \nonumber
531: t_{l}(k,k';\omega) = V_{l}(k,k')
532: \label{eq:tmat1}
533: + {4\over \pi^{2}}\sum_{\alpha}\int _{0}^{\infty } dq\:\int_{0}^{\infty }dq'\: q^{2}{q'}^{2}
534: V _{l}(k,q){\psi_{\alpha}(q)\psi_{\alpha}^{*}(q')
535: \over \omega - E_{\alpha}}V_{l}(q',k').
536: \end{equation}
537: \end{widetext}
538: The sum over $\alpha$ implies a discrete sum over bound states and an integration
539: over the positive energy continuum. Eq.~(\ref{eq:tmat1}) is analytic along the
540: real energy axis, except for poles located at bound state energies and a branch cut along
541: the positive energy axis.
542: In physical two-body scattering the incoming
543: energy is defined on the positive real energy axis. In this case
544: the integrand in Eq.~(\ref{eq:tmat1}) is singular, and a numerical solution of
545: Eq.~(\ref{eq:tmat1}) is highly non-trivial. However, for negative real input energies it
546: can be solved by standard numerical procedures.
547:
548: We are however interested in the $t$-matrix for arbitrary complex energies, since an obvious
549: extension of this work is to consider in-medium scattering of two nucleons or
550: to study the resummation of large classes of many-body diagrams.
551: In a nuclear medium calculation the self-consistently
552: determined quasiparticle energies are in general complex.
553:
554: We can achieve this by an analytic continuation of the contour integrals into the complex
555: $k$-plane. This represents a deformation of the
556: integration contour in the complex $k$-plane. See Refs.~\cite{kukulin,nuttal1}
557: for validity and mathematical proofs concerning analytic continuation of
558: integral equations.
559:
560: We will consider the solution of
561: Eq.~(\ref{eq:tmat1}) by integrating over the two contours, $C_{1}^{+}$ or $C_{2}^{+}$,
562: discussed in Sec.~\ref{sec:formalism}, resulting in the following
563: expression for $t_{l}(k,k';\omega) $
564: \begin{widetext}
565: \begin{equation}
566: \nonumber
567: t_{l}(k,k',\omega) = V_{l}(k,k')
568: \label{eq:tmat2}
569: + {4\over \pi^{2}}\sum_{\alpha}\int _{C}\int _{C'} dz\:d{z'}
570: \: z^{2}{z'}^{2}
571: V _{l}(k,z){\psi_{\alpha}(z)\psi_{\alpha}(z')
572: \over \omega - E_{\alpha}}V_{l}(z',k'),
573: \end{equation}
574: \end{widetext}
575: Here $C$ is either $C_{1}^{+}$ or $C_{2}^{+}$. This applies to $C'$ as well.
576: We have expanded the Green's operator (\ref{eq:greensfunc}) on the basis given
577: in Eq.~(\ref{eq:unity2}), this
578: implies that $\alpha $ represents discrete bound, virtual and resonant states and
579: a non-resonant complex energy continuum.
580: From the mathematical structure of Eq.~(\ref{eq:tmat2}) it is easily seen that
581: the $t$-matrix is analytic in $\omega $ in the entire complex energy-plane
582: except for poles
583: located at bound, virtual and resonant state energies. In addition there could
584: also be singularities in the potential itself. The numerical method for solving
585: Eq.~(\ref{eq:tmat2}) is based on matrix diagonalization. We first have to
586: diagonalize the corresponding Schr\"odinger equation, and then expand the
587: Green's function on the discretized basis obtained.
588:
589: Applying this method enables us to
590: obtain $t_{l}(k,k';\omega) $ for both real and complex energies $\omega $.
591: The integral becomes non-singular on the deformed contour for real and positive
592: input energies $\omega $, resulting in numerically stable solutions for physical
593: two-body scattering. Eq.~(\ref{eq:tmat2}) can also be considered as an
594: analytic continuation of Eq.~(\ref{eq:tmat1}) for complex input energies $\omega$,
595: and stable numerical solution can be obtained for complex energies above the
596: distorted contour.
597: The limitation of this method is due to that most potentials
598: in momentum space have singularities in the complex plane when one argument
599: is real and the other is complex.
600: By applying contour $C_{1}^{+}$, which is based
601: on rotation into the complex plane, in most cases there will be restrictions
602: on both rotation angle ($\theta$) and maximum incoming and outgoing momentum
603: ($k,k'$), see for example Ref.~\cite{nuttal}.
604:
605: Using contour $C_{2}^{+} $ we can avoid these limitations by
606: choosing the integration contour in such a way that the potential
607: singularities always will lie outside the integration contour, and therefore
608: do not give any restriction on rotation angle and maximum incoming and outgoing
609: momentum.
610:
611: The partial wave decomposition of Eq.~(\ref{eq:t2}) gives the \emph{Fredholm}
612: representation, commonly known as the Lippmann-Schwinger equation
613: \begin{equation}
614: \label{eq:tmat3}
615: t_{l}(k,k';\omega) = V_{l}(k,k')+
616: {2\over \pi}\int _{0}^{\infty } {dq q^{2}V_{l}(k,q)t_{l}(q,k')\over \omega - E(q) }.
617: \end{equation}
618: In physical two-body scattering the input energy is real and positive. In this
619: case the \emph{Fredholm} integral Eq.~(\ref{eq:tmat3}) has a singular
620: kernel of Cauchy type. Solving singular integrals can be done
621: by Cauchy's Residue theorem, where we
622: integrate over a closed contour enclosing the poles. There are two ways of
623: doing this, either by letting $ z $ lie an infinitesimal distance above the
624: real axis, i.e., $ \omega \rightarrow z + i\epsilon $, or by letting $ \omega $
625: lie on the real axis. In both cases we must choose a suitable contour enclosing the
626: singularity. If we choose the latter position of the singularity, we get
627: a Cauchy \emph{principal-value} integral where we integrate up to - but not
628: through - the singularity, and a second contour integral, where the
629: contour can be chosen as a semicircle around the singularity.
630: Eq.~(\ref{eq:tmat3}) can thus be given in terms of a principal value part and a
631: second term coming from integration over the semicircle around the pole. The result is
632: \begin{widetext}
633: \begin{equation}
634: \label{eq:pv1}
635: t_{l}(k,k';\omega) = V_{l}(k,k') + {2\over \pi}
636: \mathcal{P}\int _{0}^{\infty } {dq q^{2}V _{l}(k,q)t_{l}(q,k')\over \omega - E (q) }
637: \:- \: 2i\mu k_{0}V_{l}(k,k_{0})t_{l}(k_{0},k';\omega).
638: \end{equation}
639: \end{widetext}
640: By rewriting the principal value integral using the relation
641: \begin{equation}
642: \mathcal{P}\int_{0}^{\infty}dk\: {f(k)\over k_{0}^{2} - k^{2}} =
643: \int_{0}^{\infty}dk\: {[f(k)-f(k_{0})]\over k_{0}^{2} - k^{2}},
644: \end{equation}
645: we obtain an equation suitable for numerical evaluation.
646: Eq.~(\ref{eq:pv1}) can be converted into a set of linear
647: equations by approximating the integral by a sum over $N$
648: Gaussian quadrature points $( k_{j}; j = 1,...N ) $, each weighted by $ w_{j} $
649: The full off-shell $t$-matrix is then obtained by matrix inversion. This
650: method for solving integral equations is known as the Nystrom method.
651: It is numerically effective and stable, except for the rare case when
652: the incoming energy $\omega$ coincides with or is very close to one of the
653: integration points.
654:
655: So far we have only considered physical input energies in
656: Eq.~(\ref{eq:tmat3}), but it has been shown in Ref.~\cite{kukulin}
657: that the analytically
658: continued Lippmann-Schwinger equation to complex energies takes the same form
659: as Eq.~(\ref{eq:pv1}). By solving the full off-shell $t$-matrix for arbitrary
660: complex input energy, we do not have to alter the set of linear equations
661: obtained for physical energy, the only modification is that the energy is
662: complex. It should be noted that the Lippmann-Schwinger equation (\ref{eq:tmat2})
663: can be solved by the contour deformation method as well, giving a compact integral
664: kernel for positive incoming energies. By contour distortion the principal value
665: prescription is avoided, and a numerical solution of Eq.~(\ref{eq:tmat2}) is stable
666: even for incoming energies coinciding with the integration points.
667:
668: \section{An analytically solvable two-body potential}
669: \label{sec:yama}
670: We now consider a separable potential given by Yamaguchi \cite{yamaguchi}.
671: It models $s$- and
672: $p$-waves.
673: A separable interaction in momentum space
674: is analytically solvable, see for example Ref.~\cite{newton} for a demonstration.
675: The Yamaguchi interaction therefore admits analytical solution of the
676: full off-shell
677: $t$-matrix and the $t$-matrix poles, corresponding to the
678: energy spectrum. The Yamaguchi $s$-wave potential supports bound and
679: virtual states,
680: while the Yamaguchi $p$-wave potential supports bound,
681: virtual and resonant states. The
682: Yamaguchi potential is therefore very useful in
683: modelling loosely bound two-body systems
684: which may have a rich resonant state structure, and for checking the
685: numerics in our
686: calculations of the $t$-matrix and the energy spectrum.
687:
688: The $s$-wave Yamaguchi
689: potential has the form
690: \begin{equation}
691: V_{0}(k,q) = -\lambda g_{0}(k) g_{0}(q),
692: \end{equation}
693: where
694: \begin{equation}
695: g_{0}(k) = {1\over k^{2} + \beta ^{2}}.
696: \end{equation}
697: In natural units and mass $2\mu = 1 $ MeV,
698: the full off-shell $t$-matrix for the
699: $s$-wave potential reads \cite{newton}
700: \begin{equation}
701: t_{0}(k,q;z) = -\lambda {g_{0}(k)g_{0}(q)\over 1 - \lambda {2\over \pi}
702: \int_{0}^{\infty}dk\: {k^{2}\over k^{2} - E}{1\over (k^{2}+\beta ^{2})} }.
703: \end{equation}
704: The integral in the denominator can be evaluated, giving
705: \begin{equation}
706: \lambda{2\over \pi}\int_{0}^{\infty}dk\: {k^{2}\over k^{2} - E}
707: {1\over (k^{2}+\beta ^{2})} = {\lambda\over 2}{(\beta+i \sqrt{E})^{2}\over
708: \beta (\beta^{2}+E)^{2}}.
709: \end{equation}
710: The $s$-wave $t$-matrix in closed form reads
711: \begin{equation}
712: t_{0}(k,q;E) = -\lambda {g_{0}(k)g_{0}(q)\over 1 - {\lambda\over 2}
713: {(\beta+i \sqrt{E})^{2}\over \beta (\beta^{2}+E)^{2}}}.
714: \end{equation}
715: Writing $\kappa = -ik$ where $E = k^{2}$ we can solve for the $t$-matrix poles
716: as zeroes of the denominator
717: \begin{equation}
718: 1 - {\lambda\over 2} {(\beta -\kappa)^{2}\over \beta (\beta^{2}-\kappa^{2})^{2}} = 0.
719: \end{equation}
720: Solving for $\kappa$ gives
721: \begin{equation}
722: \kappa = -\beta \pm\sqrt{{\lambda\over 2\beta}},
723: \end{equation}
724: and we see that the poles are located along the imaginary $k$-axis. Poles of the $t$-matrix
725: located on the positive imaginary $k$-axis represents bound states, while poles located on
726: the negative imaginary $k$-axis represents virtual states, giving
727: a bound $(\kappa > 0 )$ and a virtual state $( \kappa \leq 0 )$ for $\lambda > 2\beta^{3} $ and for
728: $\lambda \leq 2\beta^{3} $ two virtual states.
729:
730: The separable $p$-wave interaction is given by
731: \begin{equation}
732: V_{1}(k,q) = -\lambda g_{1}(k)g_{1}(q),
733: \end{equation}
734: where
735: \begin{equation}
736: g_{1}(k) = {k\over k^{2} + \beta ^{2} },
737: \end{equation}
738: and the $t$-matrix now becomes \cite{newton}
739: \begin{equation}
740: t_{1}(k,q;E) = -\lambda {g_{1}(k)g_{1}(q)\over 1 - \lambda {2\over\pi}
741: \int_{0}^{\infty}dk\: {k^{2}\over k^{2} -E }{k^{2}\over (k^{2}+\beta^{2})^{2}}}.
742: \end{equation}
743: Solving again the integral in the denominator gives the $t$-matrix in closed form
744: \begin{equation}
745: t_{1}(k,q;E) = -\lambda {g_{1}(k)g_{1}(q)\over 1 - {\lambda\over 2}
746: {\beta ^{3} + E(3\beta+2i\sqrt{E})\over ( \beta^{2}+ E)^{2}}},
747: \end{equation}
748: and solving for the poles gives in terms of $\kappa = -ik$
749: \begin{equation}
750: \kappa = {\lambda\over 2} - \beta \pm{1\over4}\sqrt{4\lambda^{2}-8\lambda\beta}.
751: \end{equation}
752: We see that for $p$-waves the interaction supports bound, virtual and resonant
753: states. The bound state condition is
754: \begin{equation}
755: \lambda > {2\beta}.
756: \end{equation}
757: giving in addition a virtual state.
758: The interaction has a branchpoint at $k = 0$ where the bound and virtual state
759: meet and move symmetrically from the imaginary axis into the lower-half $k$-plane
760: giving capture and decay resonant states. Fig.~\ref{fig:fig4} shows the pole
761: trajectory for the Yamaguchi $p$-wave interaction.
762: \begin{figure}[h]
763: \begin{center}
764: \resizebox{6cm}{5cm}{\epsfig{file=fig5.eps}}
765: \end{center}
766: \caption{Pole trajectory for the $p$-wave Yamaguchi interaction in the complex
767: $k$-plane.}
768: \label{fig:fig4}
769: \end{figure}
770:
771:
772:
773:
774:
775: \section{Numerical results}
776: \label{sec:results}
777: Below we present numerical calculations for the complete energy spectrum of
778: the separable Yamaguchi potential \cite{yamaguchi}, for the Malfliet-Tjon
779: \cite{malfliet} and the realistic
780: charge dependent Bonn (CD-Bonn) \cite{machleidt} nucleon-nucleon
781: interactions. We also present calculations of the $t$-matrix for the given
782: interactions within the formalism outlined above. The analytic structure of the
783: various interactions
784: is of importance for the choice of the integration contour in the complex plane, as
785: will be discussed.
786: \subsection{Results for the Yamaguchi potential}
787: The Yamaguchi potential is analytic in the entire
788: complex $k$-plane except for singularities located at $k,k' = \pm i\beta$, arising
789: from the separable part of the potential $ 1 /(k^{2}+\beta^{2})$.
790: This singularity gives a restriction on the integration contour;
791: if we consider the rotated contour $C_{1}^{+}$, the rotation angle $\theta $ has to be less than
792: $\pi /2$, i.e., $\theta < \pi/2$, in the complex $k$-plane. This contour choice
793: will not be able to give the virtual states in the calculated energy spectrum.
794: The contour $C_{2}^{+}$ could on the other hand be chosen to lie above the singularity, i.e.,
795: $ b\sin(\theta) < \beta $. By this choice there is no restriction on the
796: rotation angle $\theta$ and we can rotate into the third quadrant
797: of the complex $k$-plane. By this procedure virtual states can be included in
798: the calculated energy spectrum.
799:
800: In Table~\ref{tab:yama1} we present numerical versus exact values for the
801: $s$-wave virtual states in the Yamaguchi potential, using contour $C_{2}^{+}$.
802: The contour $C_{2}^{+} $ was rotated $\theta = 2\pi /3 $ into the third
803: quadrant and translated $c = 5\sin(2\pi/3) \approx 4.3301 $ MeV in the lower-half
804: complex $k$-plane. Fig.~\ref{fig:fig6} shows the contour along with the excluded spectrum and
805: the singularities of the $s$-wave Yamaguchi potential giving restrictions on the
806: contour choice. The parameter $\beta $ was held fixed, and equal to $\beta = 6$ MeV.
807: The potential depth $\lambda $ was chosen to give only virtual states. The
808: contour choice exposes one virtual state while one virtual state will be excluded
809: from the calculated spectrum. The calculations used $N1 = 50 $
810: integration points along line $L_{1} $ and $N2 = 50 $
811: integration points along $L_{2}$
812: \begin{figure}[hbtp]
813: \begin{center}
814: \resizebox{6cm}{5cm}{\epsfig{file=fig6.eps}}
815: \end{center}
816: \caption{Excluded ($\circ $) and included ($\bullet $) spectrum for the $s$-wave Yamaguchi potential, with
817: $\beta = 6$ MeV and $\lambda $ chosen to give only virtual states (VS) in the full spectrum.
818: The calculation of the spectrum for this contour choice is given in Table~\ref{tab:yama1}.
819: Potential singularites are given by ($\times$) in the figure.}
820: \label{fig:fig6}
821: \end{figure}
822:
823: In Table ~\ref{tab:tab1} and Table~\ref{tab:tab2} we present numerical versus
824: exact values of the $p$-wave energy spectra, calculated using contour
825: $C_{2}^{+}$ and $C_{1}^{+}$. The calculations used $N1 = 50 $
826: integration points along line $L_{1} $ and $N2 = 50 $
827: integration points along $L_{2}$ for contour $C_{2}^{+}$. For the contour
828: $C_{1}^{+}$ a number of $N = 100 $ integration points was used. The rotation angle is the same for
829: both contours and equal to $ \theta = \pi /6 $. The parameter $\beta $ in the
830: potential was held fixed, and equal to $\beta = 6$ MeV. The translation parameter of
831: contour $C_{2}^{+}$ is given by $ b = 3.5$ MeV.
832: Fig.~\ref{fig:yama2} shows the contour along with the excluded and included spectrum
833: for the contour $C_{1}^{+}$ and the contour $C_{2}^{+}$,
834: along with the singularities of the $p$-wave Yamaguchi potential giving restrictions on the
835: contour choice.
836:
837: \begin{figure}[hbtp]
838: \begin{center}
839: \resizebox{6cm}{5cm}{\epsfig{file=fig7.eps}}
840: \end{center}
841: \caption{Excluded ($\circ $) and included ($\bullet $) spectrum for the $p$-wave Yamaguchi potential, with
842: $\beta = 6$MeV and $\lambda $ chosen to give bound (BS), virtual (VS), capture (CS) and decay (DS) states
843: in the full spectrum.
844: The calculations of the spectrum for contour $C_{1}^{+}$ ($C1$) and contour $C_{2}^{+} $ ($C2$) are
845: given in Tables~\ref{tab:tab1} and \ref{tab:tab2}, respectively.
846: Potential singularites are given by ($\times$) in the figure.}
847: \label{fig:yama2}
848: \end{figure}
849:
850: \begin{table}[hbtp]
851: \begin{tabular}{rrr}
852: &\multicolumn{1}{c}{Exact} & \multicolumn{1}{c}{Contour} $ C_{2}^{+} $ \\
853: \hline
854: $\lambda $ & \multicolumn{1}{c}{Real} & \multicolumn{1}{c}{Real} \\
855: \hline
856: 200 & -3.67687178 & -3.67687178 \\
857: 250 & -2.06107759 & -2.06107759 \\
858: 300 & -1.00000000 & -1.00000000 \\
859: 350 & -0.35925969 & -0.35925969 \\
860: 390 & -0.08947449 & -0.08947449 \\
861: \hline
862: \end{tabular}
863: \caption{Numerical calculations for the Yamaguchi $s$-wave
864: virtual states versus exact values.
865: Energies are given in units of MeV.
866: The numerical calculations were obtained using the
867: contour deformation method (CDM) along contour $C_{2}^{+}$. }
868: \label{tab:yama1}
869: \end{table}
870: \begin{table}[hbtp]
871: \begin{tabular}{rrrrr}\hline
872: & \multicolumn{2}{c}{Exact} & \multicolumn{2}{c}{Contour $ C_{2}^{+} $} \\
873: \hline
874: $\lambda $ & \multicolumn{1}{c}{Real} & \multicolumn{1}{c}{Imag.}& \multicolumn{1}{c}{Real} & \multicolumn{1}{c}{Imag.} \\
875: \hline
876: 13 & -5.30277586 & 0 & -5.30277586 & 0. \\
877: 12.6 & -2.80486369 & 0 & -2.80486369 & 0. \\
878: 12.2 & -0.77620500 & 0 & -0.77620500 & 0. \\
879: 11.8 & 0.57999998 & -0.15362291 & 0.57999998 & -0.15362291 \\
880: 11.7 & 0.85500001 & -0.28102490 & 0.85500001 & -0.28102490 \\
881: 11.5 & 1.37500000 & -0.59947896 & 1.37500000 & -0.59947896 \\
882: 11.0 & 2.50000000 & -1.65831244 & 2.50000000 & -1.65831244 \\
883: \hline
884: \end{tabular}
885: \caption{Numerical calculations for the Yamaguchi $p$-wave
886: bound and resonant energy spectra versus exact values.
887: Energies are given in units of MeV.
888: The numerical calculations were obtained using the
889: contour deformation method (CDM) along contour $C_{2}^{+}$. }
890: \label{tab:tab1}
891: \end{table}
892:
893: \begin{table}[hbtp]
894: \begin{tabular}{rrrrr}\hline
895: & \multicolumn{2}{c}{Exact} & \multicolumn{2}{c}{Contour $ C_{1}^{+} $} \\
896: \hline
897: $\lambda $ & \multicolumn{1}{c}{Real} & \multicolumn{1}{c}{Imag.}& \multicolumn{1}{c}{Real} & \multicolumn{1}{c}{Imag.} \\
898: \hline
899: 13 & -5.30277586 & 0 & -5.30277563 & 0. \\
900: 12.6 & -2.80486369 & 0 & -2.80486362 & 0. \\
901: 12.2 & -0.77620500 & 0 & -0.77620499 & 0. \\
902: 11.8 & 0.57999998 & -0.15362291 & 0.57999998 & -0.15362291 \\
903: 11.7 & 0.85500001 & -0.28102490 & 0.85500001 & -0.28102490 \\
904: 11.5 & 1.37500000 & -0.59947896 & 1.37500000 & -0.59947896 \\
905: 11.0 & 2.50000000 & -1.65831244 & 2.50000000 & -1.65831244 \\
906: \hline
907: \end{tabular}
908: \caption{Numerical calculations for the Yamaguchi $p$-wave
909: bound and resonant state energy spectra versus exact values.
910: Energies are given in units of MeV. The numerical calculations were
911: obtained using the
912: contour deformation method (CDM) along contour $C_{1}^{+}$. }
913: \label{tab:tab2}
914: \end{table}
915:
916: The results for the virtual states for the $p$-wave Yamaguchi potential
917: using contour $C_{2}^{+} $,
918: are given in Table~\ref{tab:tab3}.
919: \begin{table}[hbtp]
920: \begin{tabular}{rrr}
921: &\multicolumn{1}{c}{Exact} & \multicolumn{1}{c}{Contour} $ C_{2}^{+} $ \\
922: \hline
923: $\lambda $ & \multicolumn{1}{c}{Real} & \multicolumn{1}{c}{Real} \\
924: \hline
925: 13 & -1.69722438 & -1.69722438 \\
926: 12.6 & -1.15513635 & -1.15513635 \\
927: 12.2 & -0.46379500 & -0.46379500 \\
928: 12.1 & -0.25000000 & -0.25000000 \\
929: \hline
930: \end{tabular}
931: \caption{Numerical calculations for the Yamaguchi $p$-wave
932: virtual state energy spectra versus exact values.
933: Energies are given in units of MeV.
934: The numerical calculations were obtained using the
935: contour deformation method (CDM) along contour $C_{1}^{+}$ with $\theta > \pi /2$.
936: See Fig.~\ref{fig:fig6} for a plot of the contour $C_{2}^{+}$. }
937: \label{tab:tab3}
938: \end{table} \\
939: We conclude that there is no difference between the numerically calculated and the
940: exact bound, virtual and resonant energy spectra. To see a difference
941: for the number of integration points used in the calculations we have to include
942: up two 12 significant digits. Using double precision we get for the $p$-wave
943: ($\lambda = 13 $) virtual state the exact value $E = -1.697224362268005 $ MeV while
944: the numerical result in double precision is $ E = -1.697224362276472 $ MeV.
945: The calculated results for resonant states with large widths, i.e., far from
946: the
947: real energy axis agree
948: perfectly with exact values. This illustrates the advantage of the contour
949: deformation method in momentum space as opposed to the complex scaling
950: analog in coordinate space, where convergence of these states will
951: expectedly be slower due to the slowly decreasing oscillating wavefunctions.
952: Fig.~\ref{fig:fig7} gives a plot of the calculated energy spectrum for the
953: $p$-wave Yamaguchi potential,
954: using contours $C_{1}^{+}$ and $C_{2}^{+}$. Contour $C_{1}^{+}$ is rotated by $\theta =
955: 2\pi/5 $ in the energy plane while contour $C_{2}^{+} $ is rotated into the third
956: quadrant of the complex $k$-plane by $\theta = 3\pi/5 $ giving a rotation of $\theta = 6\pi/5 $
957: in the energy plane. The resonant state appears at the same
958: location for both contours.
959: \begin{figure}[hbtp]
960: \begin{center}
961: \resizebox{7cm}{7cm}{\epsfig{file=fig8.eps}}
962: \end{center}
963: \caption{Calculated energy spectra for the $p$-wave Yamaguchi potential with
964: parameters $\lambda = 11$, $\beta = 6$ using the contours
965: $C_{1}^{+}\: (C1),C_{2}^{+}\: (C2)$. The discretized non-resonant continuum lies along the contours,
966: and a resonant state (RS) appears in the energy spectra. }
967: \label{fig:fig7}
968: \end{figure}
969:
970: Next we present a calculation of the $t$-matrix elements $t_{l}(k,k;E)$ for
971: both real and complex input energy
972: (see Table ~\ref{tab:tab4} and Table~\ref{tab:tab5} ) for the $p$-wave
973: Yamaguchi potential.
974: The $t$-matrix is calculated by applying the
975: spectral representation on the deformed contour $C_{2}^{+}$. The contour is
976: rotated into the third quadrant of the complex $k$-plane and the difference
977: between numerical calculations and exact values is of the same order as for the
978: calculations of the energy spectrum. The rotation, translation and number
979: of integration points were the same as above.
980: \begin{table}[htbp]
981: \begin{tabular}{rrrrr}\hline
982: & \multicolumn{2}{c}{Exact} & \multicolumn{2}{c}{Contour $ C_{2}^{+} $} \\
983: \hline
984: $k $ & \multicolumn{1}{c}{Real} & \multicolumn{1}{c}{Imag.}& \multicolumn{1}{c}{Real} & \multicolumn{1}{c}{Imag.} \\
985: \hline
986: 1.5 & 0.102303714 & -0.061759732 & 0.102303714 & -0.061759732 \\
987: 3 & 0.295657724 & -0.178485632 & 0.295657724 & -0.178485632 \\
988: 4 & 0.425747126 & -0.257019311 & 0.425747126 & -0.257019311 \\
989: 6 & 0.461965203 & -0.278883785 & 0.461965203 & -0.278883785 \\
990: 7.5 & 0.439705133 & -0.265445620 & 0.439705133 & -0.265445620 \\
991: 9 & 0.393627137 & -0.237628803 & 0.393627137 & -0.237628803 \\
992: 10 & 0.342892975 & -0.207001090 & 0.342892975 & -0.207001090 \\
993: \hline
994: \end{tabular}
995: \caption{Calculation of the $p$-wave ($l=1$)
996: $t_l(k,k,E) $ for real momenta $k$ and input energy
997: $E = 5 $ MeV. $t_l(k,k,E) $ is given in units of MeV$^{-2}$.
998: The interaction parameters are $\lambda = 12.5$, $\beta = 6$ MeV, the rotation
999: angle is $\theta = 2\pi/3 $ while the translation is given by $b= 3.5$ MeV.
1000: The number of integration points was $N1 = N2 = 50 $}
1001: \label{tab:tab4}
1002: \end{table}
1003: \begin{table}[htbp]
1004: \begin{tabular}{rrrrr}\hline
1005: & \multicolumn{2}{c}{Exact} & \multicolumn{2}{c}{Contour $ C_{2}^{+} $} \\
1006: \hline
1007: $k $ & \multicolumn{1}{c}{Real} & \multicolumn{1}{c}{Imag.}& \multicolumn{1}{c}{Real} & \multicolumn{1}{c}{Imag.} \\
1008: \hline
1009: 1.5 & 0.094043918 & -0.030566057 & 0.094043918 & -0.030566057 \\
1010: 3 & 0.271786928 &-0.088335901 &0.271786928 &-0.088335901\\
1011: 4.5 & 0.391373158 &-0.127203703 &0.391373158 &-0.127203703\\
1012: 6 & 0.424667060 &-0.138024852 &0.424667060 & -0.138024852\\
1013: 7.5 & 0.404204220 & -0.131374046 &0.404204220 &-0.131374046\\
1014: 9 & 0.361846477 & -0.117606975 & 0.361846477 & -0.117606975\\
1015: 10.5& 0.315208495 & -0.102448739 & 0.315208495 &-0.102448739\\
1016: \hline
1017: \end{tabular}
1018: \caption{Calculation of the $p$-wave
1019: $t_{l}(k,k,E) $ for real momenta $k$ and complex input
1020: energy $E = 5 - 2i $ MeV. $t_l(k,k,E) $ is given in units of MeV$^{-2}$.
1021: Interaction parameters are $\lambda = 12.5$, $\beta = 6$ MeV ,
1022: the rotation angle is $\theta = 2\pi/3 $ while the translation is given
1023: by $b= 3.5$ MeV. The number of integration points was $N1 = N2 = 50 $.}
1024: \label{tab:tab5}
1025: \end{table}
1026: \subsection{Results for the Malfliet-Tjon interaction}
1027: \label{subsec:malfliet}
1028: The Malfliet-Tjon interaction \cite{malfliet} is a
1029: superposition of Yukawa terms. This interaction resembles the form of a
1030: realistic nucleon-nucleon interaction with attractive and repulsive parts in the
1031: ${}^{1}S_{0}$ channel. It
1032: can be fitted to reproduce the ${}^{1}S_{0}$ phase shift in nucleon-
1033: nucleon scattering rather well.
1034: The interaction in coordinate representation is given by
1035: \begin{equation}
1036: \label{eq:MF1}
1037: V(r) = V_{A}{\exp{(-\mu_{A}r)}\over r} + V_{B}{\exp{(-\mu_{B}r})\over r}.
1038: \end{equation}
1039: This interaction will support bound and virtual states for $s$-waves, while for
1040: higher angular momentum resonances will appear. It
1041: is known that the ${}^{1}S_{0}$ channel in the nucleon-nucleon
1042: interaction supports a virtual state near
1043: the scattering threshold.
1044:
1045: We will only consider $s$-waves and calculate the
1046: energy spectrum using contour $C_{2}^{+}$ rotated into the third quadrant of the
1047: complex $k$-plane.
1048: First we consider the analytic structure of
1049: the Malfliet-Tjon interaction in momentum space. The Fourier-Bessel transform of
1050: (\ref{eq:MF1}), for the $s$-wave, is
1051: \begin{eqnarray}
1052: \nonumber
1053: V_{0}(k,k') & = & V_{A}{1\over 4 kk'}\ln \left( {(k+k')^{2}+\mu_{A}^{2}\over
1054: (k-k')^{2}+\mu_{A}^{2}}\right) \\
1055: \label{eq:MF2}
1056: & + & V_{B}{1\over 4 kk'}\ln
1057: \left( {(k+k')^{2}+\mu_{B}^{2}\over (k-k')^{2}+\mu_{B}^{2}}\right).
1058: \end{eqnarray}
1059: By analytic continuation of the interaction (\ref{eq:MF2}) to the complex
1060: $k$-plane, there will be singularities in the interaction for
1061: \begin{equation}
1062: \label{eq:sing}
1063: (z - z')^{2}+\mu_{A,B}^{2} = 0,
1064: \end{equation}
1065: and
1066: \begin{equation}
1067: \label{eq:sing1}
1068: (z + z')^{2}+\mu_{A,B}^{2} = 0.
1069: \end{equation}
1070: Eq.~(\ref{eq:sing}) is only satisfied if one of the arguments is real and the other
1071: complex. For $z' = k' $ real, we have singularities for $\Re[z] = \mp k' \:
1072: \wedge \: \Im[z] = \pm\mu$.
1073: Eq.~(\ref{eq:sing1}) is satisfied for $ ( z + z' ) = \pm i \mu_{A,B} $, which gives
1074: $\Re[z] = \Re[z'] \: \wedge \: \Im[z]= \pm\mu_{A,B} -\Im[z'] $.
1075: Solving the eigenvalue problem by the contour
1076: deformation method on the purely rotated contour $C_{1}^{+}$, singularities appear along the
1077: imaginary axis given by $\Im[z] = \pm\mu_{A,B} -\Im[z'] $.
1078: This gives a restriction on rotation angle, $\theta < \pi/2 $.
1079: This does not allow a rotation into the third quadrant of the complex
1080: $k$-plane, and calculation of virtual states is not possible by this contour choice.
1081: This problem is resolved by solving the eigenvalue problem on the rotated + translated
1082: contour $C_{2}^{+}$. For a rotation angle $\theta \geq \pi/2 $ singularities appear on the contour $C_{2}^{+} $ for
1083: $\Re[z] = \Re[z']\: \wedge \: \Im[z] + \Im [z'] = \pm\mu_{A,B} $. By imposing a lower bound
1084: on the translated line segment $L_{2}$, given by $c < \mu_{A,B} / 2$, we avoid
1085: all singularities. By this choice, rotation into the third quadrant of the complex $k$-plane
1086: is allowed, and an exploration of virtual states is possible.
1087:
1088: If we want to extract the $t$-matrix along the real $k$-axis by the contour
1089: deformation method, see Sec.~\ref{sec:tmatrix}, singularities appear when Eq.~(\ref{eq:sing}) is satisfied.
1090: If we integrate along contour $C_{1}^{+} $ there will always be a singularity on the
1091: contour given by
1092: \begin{equation}
1093: z = k_{max} - i\mu,
1094: \end{equation}
1095: where
1096: \begin{equation}
1097: k_{max} = \mu /\tan(\theta ),
1098: \end{equation}
1099: and $\mu = min[\mu_{A},\mu_{B}] $.
1100: For $k,k' > k_{max} $ the contour $C_{1}^{+}$ will
1101: pass through the singularity of the interaction and Cauchy's integral theorem cannot
1102: be applied. However, if the interaction, $V_{l}(k,k') $, is approximately zero for $k,k' > k_{max} $
1103: integrating along contour $C_{1}^{+} $ can be done as long as
1104: \begin{equation}
1105: \theta < \arctan \left( \mu/ k_{max} \right ).
1106: \end{equation}
1107: In this sense we may call $k_{max} $ the cutoff momentum.
1108: This choice may cause numerical unstable solutions for small values of momenta,
1109: since the rotated contour may lie very close to the real $k$-axis where the
1110: integral kernel is singular.
1111: The same conclusion has already been pointed out in Ref.~\cite{nuttal}.
1112:
1113: Here we see the advantage of integrating along contour $C_{2}^{+}$; Not only
1114: will it be able to reproduce virtual states in the spectra, but it will also give
1115: accurate calculations of the $t$-matrix for real incoming and outgoing momentum.
1116: It will always be possible to choose a contour $C_{2}^{+}$ lying above the nearest singularity $z = \Re[k] -
1117: i\mu $, implying no restriction on rotation angle $\theta $ irrespective of cutoff
1118: momentum $k_{max} $. Fig.~\ref{fig:fig8} gives an illustration.
1119: \begin{figure}[hbtp]
1120: \begin{center}
1121: \resizebox{7cm}{5cm}{\epsfig{file=fig9.eps}}
1122: \end{center}
1123: \caption{Illustration of potential singularities for the Malfliet-Tjon interaction $V_{l}(k,z')$,
1124: where $k$ is real and $z'$ complex. Restriction on the integration contours $C_{1}^{+} \: (C1)$
1125: and $C_{2}^{+} \: (C2)$
1126: is clearly seen for given cutoff momentum $k_{max } $ ($V_{l}(k,k') \approx 0 $ for
1127: $k,k' > k_{max} $). }
1128: \label{fig:fig8}
1129: \end{figure}
1130:
1131:
1132: A calculation of resonant and virtual states in the Malfliet-Tjon interaction is
1133: given in Ref.~\cite{elster}. The $t$-matrix poles were calculated by analytic continuation of the
1134: $t$-matrix to the second energy sheet, and thereby searching for poles. In Table~\ref{tab:tab6}
1135: a calculation of the virtual neutron-proton $s$-wave states with reduced mass
1136: $M_{np} = 938.926$ MeV and
1137: $ \hbar = c = 1 $ for interaction parameters $ V_{A} = 7.291$ MeV, $\mu_{A} = 613.69$ MeV,
1138: $\mu_{B} = 305.86$ MeV held fixed while $V_{B} $ was varied as given in Table~\ref{tab:tab6}.
1139: The rotation is $\theta =
1140: 2\pi/3 $ and the translation $c = 100\sin(2\pi/3)$ MeV. The convergence is illustrated by
1141: increasing the number of integration points.
1142: \begin{table}[hbtp]
1143: \begin{tabular}{rrrrr}
1144: \hline
1145: \multicolumn{1}{c}{} & \multicolumn{3}{c}{CDM} & \multicolumn{1}{c}{Ref.\cite{elster}} \\
1146: \hline
1147: \multicolumn{1}{c}{} &\multicolumn{1}{c}{A} &\multicolumn{1}{c}{B}
1148: &\multicolumn{1}{c}{C} &\multicolumn{1}{c}{D} \\
1149: \hline
1150: \multicolumn{1}{c}{$V_{B} $} &\multicolumn{1}{c}{Energy} &\multicolumn{1}{c}{Energy}
1151: &\multicolumn{1}{c}{Energy} &\multicolumn{1}{c}{Energy} \\
1152: \hline
1153: -2.6047 & -0.066674 & -0.066653 & -0.066653 & -0.06663 \\
1154: -2.5 & -0.310114 & -0.310115 & -0.310115 & -0.31004 \\
1155: -2.3 & -1.229845 & -1.229845 & -1.229845 & -1.22970 \\
1156: -2.1 & -2.679069 & -2.678979 & -2.678979 & -2.67878 \\
1157: \hline
1158: \end{tabular}
1159: \caption{Calculation of the neutron-proton virtual state
1160: as function of interaction parameter
1161: $\lambda $ for the $s$-wave Malfliet-Tjon interaction using the deformed
1162: integration contour $C_{2}^{+}$. The convergence is illustrated
1163: by increasing the number of integration points.
1164: Column A used N1 = 30, N2 = 50 integration points, column B used N1 = 100,
1165: N2 = 100 integration points and
1166: column C used N1 = 150, N2 = 250 integration points.
1167: Comparison is made with the calculations of Ref.~\cite{elster}, given in column D.}
1168: \label{tab:tab6}
1169: \end{table}
1170: We see a small difference in the calculated values for the virtual state in the
1171: Malfliet-Tjon interaction.
1172: As we obtained convergence of the virtual state energies by increasing the number
1173: of integration points, this suggests that our results are stable.
1174: Fig.~\ref{fig:fig9} shows a plot of the calculated $t$-matrix elements
1175: $t_{l}(k,k;E) $ for incoming energy $E = 100$ MeV for the $s$ -wave Malfliet-Tjon
1176: interaction, using the contours $C_{1}^{+} $ and $C_{2}^{+}$. In this case we used a
1177: rotation angle $\theta = \pi/6 $ for both contours, and a translation $b =
1178: 300$ MeV for contour $C_{2}^{+}$. The Malfliet-Tjon interaction has a
1179: singularity along contour $C_{1}^{+}$, given by $k_{max} = \mu /\tan(\theta ) =
1180: 529.77$ MeV. In this case the interaction
1181: is not sufficiently neglible for $k,k' > k_{max}$,
1182: and using contour $C_{1}^{+}$ will not give accurate calculation of the $t$-matrix,
1183: this is clearly seen in the plot of the calculated results.
1184: \begin{figure}[hbtp]
1185: \begin{center}
1186: \resizebox{7cm}{6cm}{\epsfig{file=fig10.eps}}
1187: \end{center}
1188: \caption{Plot of $t$-matrix elements for the $s$-wave Malfliet-Tjon interaction
1189: using the contour $C_{1}^{+}$ (dashed line) and the contour $C_{2}^{+}$ (solid line).
1190: The potential singularity along contour $C_{1}^{+}$ is clearly displayed, and
1191: located at $k_{max} = 529.77$ MeV.}
1192: \label{fig:fig9}
1193: \end{figure}
1194: This illustrates clearly the advantage of integrating along the contour $C_{2}^{+}$
1195: instead of $C_{1}^{+}$.
1196:
1197: A useful check
1198: of the numerics is the on-shell unitarity for the $S$-matrix. Table ~\ref{tab:tab7} reports
1199: calculations done by the principal-value prescription and the
1200: contour deformation method $C_{2}^{+}$. We used $N=50 $ integration points for the principal value
1201: integration, for the contour deformation method we used $N1 = 30$ and $N2 =
1202: 100,200$. We observe that we need a higher number of integration points
1203: along line $L_{2}$ for obtaining convergence.
1204: \begin{table}[htbp]
1205: \begin{tabular}{rrrrr}
1206: \hline
1207: \multicolumn{1}{c}{} & \multicolumn{1}{c}{PV} & \multicolumn{2}{c}{CDM} \\
1208: \hline
1209: \multicolumn{1}{c}{$k$[MeV]} & \multicolumn{1}{c}{$N = 50$} & \multicolumn{1}{c}{$N2 = 100 $} & \multicolumn{1}{c}{$N2 = 200 $ } \\
1210: \hline
1211: 10. & 1.00000000 & 1.00000811 & 1.00000811 \\
1212: 110. & 1.00000000 & 0.999999940 & 1.00000000 \\
1213: 210. & 1.00000000 & 0.999999940 & 1.00000000 \\
1214: 310. & 1.00000000 & 0.999999881 & 1.00000000 \\
1215: 410. & 1.00000000 & 0.999999881 & 1.00000000 \\
1216: 510. & 1.00000000 & 0.999999881 & 1.00000000 \\
1217: 610. & 1.00000000 & 0.999999881 & 1.00000000 \\
1218: 710. & 1.00000000 & 0.999999821 & 1.00000000 \\
1219: 810. & 1.00000000 & 0.999999821 & 1.00000000 \\
1220: 910. & 1.00000000 & 0.999999821 & 1.00000000 \\
1221: 1010.& 1.00000000 & 0.999999762 & 1.00000000 \\
1222: 1110.& 1.00000000& 0.999999702 & 1.00000000 \\
1223: 1210.& 1.00000000& 0.999999702 & 1.00000000 \\
1224: 1310.& 1.00000000& 0.999999642 & 1.00000000 \\
1225: 1410.& 1.00000000& 0.999999523 & 1.00000000 \\
1226: 1510.& 1.00000000& 0.999999642 & 1.00000000 \\
1227: \hline
1228: \end{tabular}
1229: \caption{Calculations of $S$-matrix norm, $\vert S(k) \vert $, for the $s$-wave
1230: Malfliet-Tjon potential with parameters $ V_{A} = 7.291$ MeV,
1231: $\mu_{A} = 613.69$ MeV, $V_{B} = -5$ MeV and $\mu_{B} = 305.86$ MeV. Column 2 gives results for the
1232: principal value prescription (PV), while columns 3,4 give results for the contour
1233: deformation method (CDM).}
1234: \label{tab:tab7}
1235: \end{table}
1236:
1237: \subsection{Results for the CD-Bonn nucleon-nucleon interaction}
1238: Finally, we present a calculation of the energy spectrum of the
1239: charge-dependent Bonn interaction (CD-Bonn) and phase shifts for a set of selected
1240: partial waves. The CD-Bonn interaction is given in Ref.~\cite{machleidt}.
1241: We illustrate the power of the contour deformation method by reproducing the deuteron bound state and
1242: the virtual
1243: states in the ${}^{1}S_{0} $ isospin triplet channel. In addition, phase
1244: shifts are calculated within the same method, and the results agree perfectly with
1245: other theoretical predictions.
1246:
1247: The tensor component of the CD-Bonn interaction
1248: couples angular momentum, $l=j-1, l = j+1$, in the spin triplet channel.
1249: It is straightforward to include this coupling in the formalism outlined
1250: in the previous sections. Instead of $N$x$N $ matrices we get $2N$x$2N$ matrices
1251: in the coupled case, for the uncoupled channels we still have $N$x$N$
1252: matrices. We should of course label the equations with quantum numbers
1253: $L,S,J,T$.
1254:
1255: The realistic nucleon-nucleon interaction does not have a
1256: resonant structure in the low energy region $ E < 350$ MeV.
1257: It does however support virtual states in the ${}^{1}S_{0} $ isospin triplet
1258: channel and a bound state in the coupled isospin singlet channel (the deuteron)
1259: $ {}^{3}S_{1} - {}^{3}D_{1} $. We compare the calculated virtual state locations in
1260: the complex $k$-plane with the values obtained by the effective range
1261: approximation, see Ref.~\cite{newton,kukulin}. In the following we do not include Coulomb
1262: effects when considering the isospin $t_{z} = -1 $ channel (proton-proton scattering).
1263:
1264: The effective range approximation for the $s$-wave poles is given by
1265: \begin{equation}
1266: \label{eq:eff_range}
1267: k = i \left[ {1\over r_{NN}} - \sqrt{ {2\over r_{NN}\vert a_{NN}\vert } + {1\over r_{NN}^{2} }}\right].
1268: \end{equation}
1269: The theoretical (see below) and experimental values \cite{machleidt}
1270: for the ${}^{1}S_{0} $ scattering lengths $a_{NN}$ and
1271: effective range $r_{NN}$ are given in Table~\ref{tab:tab8}
1272: \begin{table}[htbp]
1273: \begin{tabular}{rrr}\hline
1274: \multicolumn{1}{c}{} & \multicolumn{1}{c}{CD-Bonn} & \multicolumn{1}{c}{Experiment} \\
1275: \hline
1276: $a_{pp}$ & $-17.4602 $ & {} \\
1277: $r_{pp}$ & 2.845 & {} \\
1278: $a_{nn}$ & -18.9680 & -18.9 $\pm $0.4 \\
1279: $r_{nn}$ & 2.819 & 2.75 $\pm $0.11 \\
1280: $a_{np}$ & -23.7380 & -23.740 $\pm $0.020 \\
1281: $r_{np}$ & 2.671 & 2.77 $\pm $0.05 \\
1282: \hline
1283: \end{tabular}
1284: \caption{Scattering lengths (a) and effective ranges (r) for the ${}^{1}S_{0} $ channel,
1285: in units of fm. For the proton - proton channel Coulomb effects are not included.}
1286: \label{tab:tab8}
1287: \end{table}
1288:
1289: To apply the contour deformation method by integrating along contour $C_{2}^{+}$ we
1290: first have to locate possible singularities in the CD-Bonn interaction.
1291: The CD-Bonn interaction
1292: is given explicitly in momentum space. The derivation of the interaction is based
1293: on field theory, starting from Lagrangians describing the coupling of the various
1294: mesons of interest to nucleons \cite{machleidt}.
1295: The one-boson-exchange interaction is proportional to
1296: \begin{equation}
1297: \label{eq:cdbonn}
1298: V({\bf k,q} ) \propto \sum _{\alpha = \pi^{0},\pi^{\pm}, \rho ,\omega,
1299: \sigma_{1},\sigma_{2}}\bar{V}_{\alpha}({\bf k,q } )F_{\alpha}^{2}({\bf k,q}; \Delta_{\alpha}).
1300: \end{equation}
1301: Both $\bar{V}_{\alpha}({\bf k,q } ) $ and $F_{\alpha}^{2}({\bf k,q}; \Delta_{\alpha}) $
1302: contain terms of the form
1303: \begin{equation}
1304: {1\over ({\bf k -q } )^{2} + m_{\alpha }^{2} },
1305: \end{equation}
1306: which are of the Yukawa type.
1307: These terms determine the analyticity region of the interaction in the $k$-plane
1308: (see discussion of the Malfliet-Tjon interaction above.
1309: We see that the poles of the interaction are determined by the various meson masses
1310: $m_{\alpha}$ and cut-off masses $\Delta _{\alpha}$.
1311: Considering the solution of the eigenvalue problem by the
1312: contour deformation method using contour $C_{2}^{+}$, singularities in the interaction
1313: appear for $\theta \geq \pi/2 $, i.e., a rotation into the third quadrant of the
1314: complex $k$-plane. The maximum translation into the complex $k$-plane is then determined
1315: by the smallest meson mass entering the potential, which is the $\pi $-meson, $m_{\pi^{0}}
1316: = 134.9764$ MeV. For a given rotation into the third quadrant, i.e., $\theta \geq \pi/2 $,
1317: we get a restriction on the translation parameter; $ b < 134.9764/2\sin(\theta )$. Here $b$ is given in MeV, as
1318: we are still using natural units.
1319:
1320:
1321: Table~\ref{tab:tab9} gives results for the virtual states in the
1322: ${}^{1}S_{0} $ channel by the contour deformation method. A comparison with
1323: the effective range calculation of the virtual state poles is also shown.
1324: The contour was
1325: rotated by $\theta = 2\pi /3 $ into the complex $k$-plane and translated
1326: $c = -30\sin(5\pi/7)$ MeV or $c\approx -23.45$ MeV in the lower-half $k$-plane.
1327: This is sufficient to reproduce the virtual states in the ${}^{1}S_{0} $ channel, as they are
1328: known to lie very close to the scattering threshold, $k\approx -\hbar c 0.05i$ MeV.
1329: By this contour choice the full energy spectrum is obtainable since it is known
1330: \emph{a posteriori} that the ${}^{1}S_{0} $ channel supports only virtual states near
1331: the scattering threshold.
1332: The convergence of the
1333: numerical calculated values is demonstrated by increasing the number of integration points $N$.
1334: \begin{table}[htbp]
1335: \begin{tabular}{rrrrr}\hline
1336: \multicolumn{1}{c}{} & \multicolumn{3}{c}{CDM} &\multicolumn{1}{c}{EFR} \\
1337: \hline
1338: \multicolumn{1}{c}{} & \multicolumn{1}{c}{a} & \multicolumn{1}{c}{b} & \multicolumn{1}{c}{c}
1339: & \multicolumn{1}{c}{} \\
1340: \hline
1341: \multicolumn{1}{c}{$T_{z}$} & \multicolumn{1}{c}{Energy} &
1342: \multicolumn{1}{c}{Energy} & \multicolumn{1}{c}{Energy} &
1343: \multicolumn{1}{c}{Energy} \\
1344: \hline
1345: -1 (pp) & -0.11766 & -0.11761 & -0.11761 & -0.11763 \\
1346: 1 (nn) & -0.10070 & -0.10069 & -0.10069 & -0.10070 \\
1347: 0 (np) & -0.06632 & -0.06632 & -0.06632 & -0.06632 \\
1348: \hline
1349: \end{tabular}
1350: \caption{Calculation of virtual state energies in the ${}^{1}S_{0}$ isospin triplet channel
1351: by the effective range approximation (EFR) and the contour deformation method (CDM),
1352: in units of MeV.
1353: Convergence is obtained in each isospin channel. Column (a) used $N1 = 20, N2 = 30$ integration points,
1354: column (b) used $N1=20, N2 = 50$ integration points while column (c) used
1355: $N1 = 30, N2 = 80$ integration points }
1356: \label{tab:tab9}
1357: \end{table}
1358: Table~\ref{tab:tab10} illustrates the convergence of the deuteron binding energy as
1359: a function of integration points. The contour was rotated by $\theta = \pi /6 $ into the complex $k$-plane and translated
1360: $c = -\hbar c\sin(\pi/6)$ MeV or $ c\approx -100$ MeV into the lower-half $k$-plane.
1361: \begin{table}[htbp]
1362: \begin{tabular}{rrr}\hline
1363: \multicolumn{1}{c}{N1} & \multicolumn{1}{c}{N2} & \multicolumn{1}{c}{Energy} \\
1364: \hline
1365: 20 & 30 & -2.224581 \\
1366: 20 & 50 & -2.224573 \\
1367: 30 & 80 & -2.224574 \\
1368: 50 & 100 & -2.224575 \\
1369: 50 & 150 & -2.224575 \\
1370: \hline
1371: \end{tabular}
1372: \caption{Calculation of deuteron binding energy in the coupled channel
1373: ${}^{3}S_{1}-{}^{3}D_{1}$ by the contour deformation method (CDM), in units
1374: of MeV. See Ref.~\cite{machleidt} for further details.}
1375: \label{tab:tab10}
1376: \end{table}
1377:
1378: Figs.~\ref{fig:pp},\ref{fig:np} and \ref{fig:nn} present calculations of
1379: nucleon-nucleon phaseshifts for
1380: the uncoupled ${}^{1}S_{0}$ isospin triplet channel, by the contour
1381: deformation method. The contour was rotated by $2\pi /3 $ into the third quadrant of
1382: the complex $k$-plane, and translated $c = 50\sin(2\pi/3)$ MeV. We used $N1 = 30 $ and
1383: $ N2 = 70 $ integration points along line $L_{1}$ and line $L_{2}$, respectively.
1384: Our calculation is given by
1385: the solid line while the circles are calculations given in Ref.~\cite{machleidt}.
1386: For proton-proton scattering Coulomb effects are not included.
1387: \begin{figure}[hbtp]
1388: \begin{center}
1389: \resizebox{7cm}{7cm}{\epsfig{file=fig11.eps}}
1390: \end{center}
1391: \caption{Phaseshift for proton-proton scattering in the ${}^{1}S_{0}$ channel. Coulomb
1392: effects are not included in the calculations. }
1393: \label{fig:pp}
1394: \end{figure}
1395:
1396: \begin{figure}[hbtp]
1397: \begin{center}
1398: \resizebox{7cm}{7cm}{\epsfig{file=fig12.eps}}
1399: \end{center}
1400: \caption{Phaseshift for neutron-proton scattering in the ${}^{1}S_{0}$ channel.
1401: Calculation by the contour deformation method is given by the solid line. Circles are
1402: calculations given in Ref.~\cite{machleidt}. }
1403: \label{fig:np}
1404: \end{figure}
1405:
1406: \begin{figure}[hbtp]
1407: \begin{center}
1408: \resizebox{7cm}{7cm}{\epsfig{file=fig13.eps}}
1409: \end{center}
1410: \caption{Phaseshift for neutron-proton scattering in the ${}^{1}S_{0}$ channel.
1411: Calculation by the contour deformation method is given by the solid line. Circles are
1412: calculations given in Ref.~\cite{machleidt}. }
1413: \label{fig:nn}
1414: \end{figure}
1415:
1416:
1417: \section{Conclusions and perspectives}
1418:
1419: A generalized contour deformation method in momentum space has been presented.
1420: The deformation of the integration contour is generalized to rotation followed
1421: by translation in the complex $k$-plane. This generalization makes it
1422: possible to handle both \emph{dilation} and \emph{non-dilation} invariant
1423: potentials. We have shown this to be a powerful procedure for studying
1424: resonances and virtual states in two-body systems. The method has also
1425: been successfully applied to the calculation of the full off-shell $t$-matrix.
1426: The aim of this work has been to establish the formalism for the free scattering case,
1427: basing the analysis on schematic and realistic nucleon-nucleon interactions.
1428: Work on extension to few-body Borromean halo systems \cite{zhukov} with more than two constituents, is
1429: in progress.
1430:
1431: The exposed formalism allows for stable numerical calculations
1432: of bound states, resonances and
1433: virtual states, in addition to yielding a fully complex on-shell and off-shell scattering matrix,
1434: starting with a realistic nucleon-nucleon interaction.
1435:
1436: This allows for several interesting applications and extensions.
1437: As is well-known, one of the major challenges in the microscopic description of weakly bound nuclei is
1438: a proper treatment of both the many-body correlations and the continuum of positive energies
1439: and decay channels. Such nuclei pose a tough challenge on traditional nuclear structure methods,
1440: based essentially on the derivation of effective interactions and the nuclear shell-model, see
1441: for example Ref.~\cite{mhj95} for a review. In the traditional approaches only bound states typically
1442: enter the determination of an effective interaction, be it either based upon various many-body schemes
1443: or more phenomenologically inspired approaches. Coupled with large-scale shell model studies, several
1444: properties of stable nuclei are well reproduced. However, weakly bound nuclei may have a strong coupling to
1445: unbound states, either resonances or virtual states, as described in for example
1446: Refs.~\cite{witek1, witek2, roberto}. This implies in turn that an effective interaction should reflect
1447: such couplings with the continuum, i.e., a consistent many-body scheme should include bound states,
1448: resonances and virtual states as well.
1449:
1450: The present approach may allow for such a scheme. The simplest extension is to consider
1451: in-medium scattering of two nucleons in e.g., infinite nuclear matter or neutron star matter, as done
1452: by Dickhoff and {\em et al.} \cite{dickhoff1998} or Schulze {\em et al.} \cite{hansjosef}.
1453: This is easily accomplished by inserting an exclusion operator $Q$ in Eq.~(\ref{eq:spectral1})
1454: for the Green's operator. This exclusion operator can be constructed so as to prevent scattering
1455: into occupied states. Typically, one can then generate selected hole-hole, particle-hole
1456: and particle-particle intermediate
1457: state correlations that reflect a given nuclear medium. Combined with a self-consistent determination of the
1458: single-particle energies it is then possible to derive various classes of diagrams at the two-body level.
1459: Such work is in progress, based on the present method and the exact definition of the
1460: exclusion operator $Q$ by Suzuki {\em et al.} \cite{kenyj_rioyj}.
1461:
1462:
1463:
1464:
1465: \begin{thebibliography}{200}
1466: \bibitem{newton} R.~G.~Newton \emph{Scattering Theory of Waves and Particles}
1467: (Springer-Verlag, New York, 1966,1982).
1468: \bibitem{kukulin} V.~I.~Kukulin, V.~M.~Krasnopol'sky, and
1469: J.~Hor$\mathrm{\acute{a}\check{c}}$ek,
1470: \emph{Theory of Resonances} (Kluwer Academic publishers 1989).
1471: \bibitem{afnan1} I.~R.~Afnan, Aust.~J.~Phys.~{\bf 44 }, 201 (1991).
1472: \bibitem{abc} J.~Aguilar and J.~M.~Combes, Commun.~Math.~Phys.~{\bf 22}, 269 (1971).
1473: \bibitem{abc1} E.~Balslev and J.~M.~Combes, Commun.~Math.~Phys.~{\bf 22}, 280 (1971).
1474: \bibitem{moise} N.~Moiseyev, Phys.~Rep.~{\bf 302}, 211 (1998).
1475: \bibitem{csoto} A.~Csoto, Phys.~Rev.~C {\bf 49}, 2244 (1994).
1476: \bibitem{garrido} E.~Garrido, D.~V.~Fedorov, and A.~S.~Jensen,
1477: Nucl.~Phys.~A {\bf 708}, 277 (2002).
1478: \bibitem{imante} I. Raskinyte, \emph{Resonances in few-body systems},
1479: PhD Thesis University of Bergen 2002.
1480: \bibitem{brayshaw} D.~Brayshaw, Phys.~Rev.~{\bf 176 }, 1855 (1968).
1481: \bibitem{nuttal} J.~Nuttall and H.~L.~Cohen, Phys.~Rev.~{\bf 188}, 1542 (1969).
1482: \bibitem{stelbovics} A.~T.~Stelbovics, Nucl.~Phys.~A {\bf 288}, 461 (1978).
1483: \bibitem{berggren} T.~Berggren, Nucl.~Phys.~A {\bf 109}, 265 (1968)
1484: \bibitem{lind} P.~Lind, Phys.~Rev.~C {\bf 47}, 1903 (1993)
1485: \bibitem{nuttal1} J.~Nuttall, J.~Math.~Phys.~{\bf 8}, 873 (1967).
1486: \bibitem{tikto} G.~Tiktopoulos, Phys.~Rev.~{\bf 136}, 275 (1964).
1487: \bibitem{aoyama} S.~Aoyama, Phys.~Rev.~Lett.~{\bf 89}, 052501 (2002).
1488: \bibitem{afnan} B.~C.~Pearce and I.~R.~Afnan, Phys.~Rev.~C {\bf 30}, 2022 (1984).
1489: \bibitem{yamaguchi} K.~Yamaguchi, Phys.~Rev.~{\bf 95}, 1628 (1954).
1490: \bibitem{malfliet} R.~A.~Malfliet and J.~A.~Tjon, Nucl.~Phys.~A {\bf 127}, 161 (1969).
1491: \bibitem{machleidt} R.~Machleidt, Phys.~Rev.~C {\bf 63}, 024001 (2001).
1492: \bibitem{betan} R.~Id Betan, R.~J.~Liotta, N.~Sandulescu, and T.~Vertse,
1493: Phys.~Rev.~C {\bf 67}, 014322 (2003).
1494: \bibitem{brown} L.~Brown, D.~Fivel, B.~W.~Lee, and R.~Sawyer, Ann.~of Phys.~{\bf 23}, 167 (1963).
1495: \bibitem{elster} Ch.~Elster, J.~H.~Thomas, and W.~Gl\"ockle, Few-Body Systems {\bf 24}, 55 (1998).
1496: \bibitem{zhukov} M.~V.~Zhukov, B.~V.~Danilin, D.~V.~Fedorov, J.~M.~Bang,
1497: I.~J.~Thompson, and J.~S.~Vaagen, Phys.~Rep.~{\bf 231}, 151 (1993).
1498: \bibitem{mhj95} M.~Hjorth-Jensen, T.~T.~S.~Kuo, and E.~Osnes, Phys.~Rep.~{\bf 261},
1499: 125 (1995).
1500: \bibitem{witek1} N.~Michel, W.~Nazarewicz, M.~P{\l}oszajczak, and K.~Bennaceur,
1501: Phys.~Rev.~Lett.~{\bf 89}, 042502 (2002).
1502: \bibitem{witek2} N.~Michel, W.~Nazarewicz, M.~P{\l}oszajczak, and J.~Oko{\l}owicz, nucl-th/0302060.
1503: \bibitem{roberto}R.~Id Betan, R.~J.~Liotta, N.~Sandulescu, and T.~Vertse,
1504: Phys.~Rev.~Lett.~{\bf 89}, 042501 (2002).
1505: \bibitem{dickhoff1998} W.~H.~Dickhoff, Phys.~Rev.~C {\bf 58}, 2807 (1998); W.~H.~Dickhoff, C.~C.~Gearhart,
1506: E.~P.~Roth, A.~Polls, and A.~Ramos, Phys.~Rev.~C {\bf 60}, 064319 (1999).
1507: \bibitem{hansjosef} H.-J.~Schulze, A.~Schnell, G.~R\"opke, and U.~Lombardo, Phys.~Rev.~C {\bf 55}, 3006 (1997).
1508: \bibitem{kenyj_rioyj} K.~Suzuki, R.~Okamoto, M.~Kohno, and S.~Nagata, Nucl.~Phys.~A {\bf 665}, 92 (2000).
1509: \end{thebibliography}
1510:
1511:
1512:
1513: \end{document}
1514: