1: \chapter{ Numerical Results}
2: \section{Simplification of the Hamiltonian}
3: \hspace{0.75in} The full Hamiltonian for the general spin-Peierls
4: system is
5: shown in the previous chapter, Eq. (\ref{SH}). It is difficult to work
6: with
7: this Hamiltonian to calculate quantities of interest like single-particle
8: energies and wave functions at zero temperature, spin densities etc.
9: Hence the full Hamiltonian is simplified using the
10: physical arguments discussed below so that the quantities calculated from the
11: simplified Hamiltonian reproduce the
12: physics of the full Hamiltonian to a very good approximation.
13:
14: \subsection{Approximations involved and their justification:}
15:
16: \begin{itemize}
17: \item{ \textit{Soft phonon limit}: The time scales involved in
18: the
19: motion
20: of lattice are known to be much larger than the time scales
21: involved in
22: the
23: motion of spins arising due to the interactions between the spins and
24: phonons.
25: Hence the lattice points are initially taken to be at fixed positions
26: while
27: calculating the unperturbed
28: single-particle energies at zero temperature and zero field. At the end, the
29: lattice
30: movements are however taken into account using the perturbation theory
31: and the
32: perturbed single-particle energies due to motion of lattice points are
33: calculated to second order in bond displacements.}
34:
35: \item{\textit{XY spins}: The generalised Heisenberg spins in the
36: model are
37: replaced by the XY spins, i.e., the spin-spin interactions in the $\hat{z} $
38: direction are
39: neglected. This is done to simplify the calculations. When the spins are
40: transformed into Jordan-Wigner fermions, the interactions in the $\hat{z}$
41: direction involves four spinless fermion operators \cite{4}. It is known from
42: numerical
43: analysis that reducing the Heisenberg spins
44: to XY spins
45: does not qualitatively affect the phase diagram of a quasi 1D spin-Peierls
46: system.}
47: \end{itemize}
48:
49:
50: The simplified Hamiltonian, $H_s$, thus obtained using the above
51: approximations is given by
52:
53: \begin{eqnarray}
54: H_s\hspace{0.2 cm}=\hspace{0.2 cm}-J\sum_{n=1}^{N}
55: \left[1+\delta_{n+\frac{1}{2}}\right]\left[S_{n}^xS_{n+1}^x + S_n^yS_{n+1}^y
56: \right] \hspace{0.1 cm} - \hspace{0.1 cm} h\sum_{n=1}^{N}S_n^z
57: \hspace{0.1 cm}+\hspace{0.1 cm}
58: \frac{c}{2}\sum_{n=1}^{N}\delta_{n+\frac{1}{2}}^2
59: \label{simpSH}
60: \end{eqnarray}
61:
62: \noindent where $\delta_{n+\frac{1}{2}}=-\lambda\left(u_{n+1} - u_{n}
63: \right)$ is the bond displacement and $\lambda$ is the positive constant which
64: is a measure of the
65: coupling that exists between the lattice points and spins.
66:
67: It is to be noted that in addition to the mentioned approximations, a unitary
68: transformation of the form shown in Eqs. (\ref{T1}) is performed on the
69: Hamiltonian given in Eq.(\ref{SH}) to reverse the sign of the first term.
70:
71: \begin{eqnarray}
72: S_n^x &\rightarrow &(-1)^nS_n^x \nonumber \\
73: S_n^y &\rightarrow& (-1)^nS_n^y \nonumber \\
74: S_n^z &\rightarrow& S_n^z
75: \label{T1}
76: \end{eqnarray}
77:
78: \subsection{Jordan-Wigner transformation}
79: Jordan-Wigner transformation transforms a spin Hamiltonian into a spinless
80: fermion Hamiltonian \cite{2, 9, 10}. Here, two spinless fermion operators, the
81: creation and
82: annihilation operators, are defined as shown below
83:
84: \begin{eqnarray}
85: C_n &=&\prod_{j=1}^{n-1}(-\sigma_j^z)S_n^- \nonumber \\
86: C_n^+ &=&\prod_{j=1}^{n-1}S_n^+(-\sigma_j^z)
87: \label{JW}
88: \end{eqnarray}
89:
90: \noindent where \begin{eqnarray}
91: S_n^- &=& S_n^x-iS_n^y \nonumber \\
92: S_n^+ &=& S_n^x+iS_n^y \nonumber
93: \end{eqnarray}
94: \noindent and $\sigma$'s are the Pauli matrices. Using the properties of Pauli
95: matrices, it follows from the relations (\ref{JW}) between spin operators and
96: fermion operators that the total magnetization,\textit{m}, of the system is
97:
98: \begin{eqnarray}
99: m=\sum_{n=1}^{n=N}S_n^z=N_F-\frac{N}{2}
100: \end{eqnarray}
101:
102: \noindent where \begin{math}
103: N_F=\sum_{n=1}^{N} C_n^+C_n
104: \end{math} is the number of Jordan-Wigner fermions.
105:
106: Using the Jordan-Wigner transformation on the simplified spin Hamiltonian
107: (\ref{simpSH}), a spinless fermion Hamiltonian
108: is obtained as shown in Eq. (\ref{FH})
109:
110: \begin{eqnarray}
111: \tilde{H}\hspace{0.2 cm}&=&\hspace{0.2
112: cm}-\frac{J}{2}\left[\sum_{n=1}^{N-1}(1+\delta_{n+\frac{1}{2}})
113: ( C_n^+C_{n+1}+C_{n+1}^+C_n)\hspace{0.1 cm}
114: +\hspace{0.1cm}(-1)^{N_F+1}(C_N^+C_{1}+C_{1}^+C_N) \right]
115: \hspace{0.1 cm} \nonumber \\
116: && \hspace{0.2 cm}-
117: h\sum_{n=1}^{N}\left[C_n^+C_n-\frac{1}{2}\right]\hspace{0.1 cm}+\hspace{0.1 cm}
118: \frac{c}{2}\sum_{n=1}^{N}\delta_{n+\frac{1}{2}}^2
119: \label{FH}
120: \end {eqnarray}
121:
122: \section{Single particle energies}
123:
124: The Hamiltonian used to calculate the single-particle energies at zero
125: temperatures for all phases ( for all possible
126: magnetizations) is
127: given by
128:
129: \begin{eqnarray}
130: \textit{H}\hspace{0.2 cm}&=&\hspace{0.2
131: cm}-\frac{J}{2}\sum_{n=1}^{N-1}\left[1+\delta_{n+\frac{1}{2}}\right]\left[C_n^+
132: C
133: _{n+
134: 1}+C_{n+1}
135: ^
136: +C_n\right]\hspace{0.1 cm}
137: +\hspace{0.1cm}\left(-1\right)^{N_F}\left[C_N^+C_{1}+C_{1}^+C_N
138: \right]
139: \hspace{0.1 cm}\nonumber \\
140: & & \hspace{0.2 cm}-
141: h\sum_{n=1}^{N}\left[C_n^+C_n-\frac{1}{2}\right]\hspace{0.1
142: cm}+\hspace{0.1
143: cm}
144: \frac{c}{2}\sum_{n=1}^{N}\delta_{n+\frac{1}{2}}^2.
145: \label{NFH}
146: \end {eqnarray}
147:
148:
149: It can be seen that the \textit{boundary term} of this Hamiltonian has
150: a
151: relative minus sign compared to the Hamiltonian shown in the Eq.
152: (\ref{FH}). This appears to break the translation invariance. However, Eq.
153: (\ref{NFH})can be made translation invariant by transforming all the fermion
154: operators as
155:
156: \begin{eqnarray}
157: C_n &\rightarrow& e^{i\pi/N} C_n \nonumber \\
158: \hspace{0.3cm}C_n^+ &\rightarrow& e^{-i\pi/N} C_n^+
159: \end{eqnarray}
160:
161: \noindent which distributes the minus sign at the boundary over all the sites
162: equally.
163: As $N\rightarrow\infty$, such a transformation produces a vanishingly small
164: change in the single-particle energies.
165:
166: The system size \textit{N} is chosen to be 100 spins. The constant $c$ in the
167: final term in the Hamiltonian (\ref{NFH}) showing the harmonic energy of the
168: lattice is set equal to unity. The lattice constant
169: \textit{l} is also fixed to be unity so that all lengths which are multiples of
170: the
171: lattice
172: constant are now just integers. In order to obtain the
173: single-particle energies for any particular
174: magnetization or any particular filling of Jordan-Wigner fermions,
175: $N_F$, of the
176: ground-state, it is necessary to obtain the set of bond
177: displacements, $\delta_{n+\frac{1}{2}}$'s which minimize the Hamiltonian given
178: in
179: Eq. (\ref{NFH}). The procedure followed to achieve this is
180:
181: \begin{itemize}
182: \item{Set all $\delta_{n+\frac{1}{2}}$'s to zero initially.}
183: \item{Increment or decrement each $\delta_{n+\frac{1}{2}}$ separately in steps
184: of
185: 0.0005 and change each of the remaining spins by $0.0005/99$
186: with the reverse sign, so that for each such change the sum of all
187: displacements continues to remain zero as required for a periodic lattice.}
188: \item{Calculate the new ground-state energy for each such change and iterate in
189: a loop running over all the lattice sites, so as to minimize the Hamiltonian.}
190: \end{itemize}
191:
192: \noindent Then the eigenvalues of the Hamiltonian with these bond
193: displacements, $\delta_{n+\frac{1}{2}}$'s give the single-particle energies for
194: the
195: particular filling at zero temperature. The eigenfunctions similarly give the
196: single-particle wave functions.
197: The bond displacements and single-particle energies are obtained for all
198: possible magnetizations, few interesting ones of which are shown in the
199: following
200: sections.
201:
202: \subsection{For zero magnetization (for $N_F=50$)}
203:
204: The bond displacements for the case of zero magnetization, i.e., for
205: $N_F$=50, obtained by minimizing the Hamiltonian in Eq. (\ref{NFH}) using the
206: above procedure are shown in the lower part of Fig. (\ref{D50}). It can be
207: seen that the value of the displacement of the bond alternates in sign
208: between any two consecutive lattice spins with its magnitude remaining
209: constant. This corresponds to dimerization of the lattice.
210: The magnitude of each of the bond displacements, $\delta$, is 0.0642.
211:
212: \begin{figure}[h]
213: \centerline{\psfig{figure=D50.ps,height=10.5 cm}}
214: \caption{The bond displacements and expectation values of $S_n^z$ for zero
215: magnetization.}
216: \label{D50}
217: \end{figure}
218:
219:
220: \begin{figure}[h]
221: \centerline{\psfig{figure=E50.ps,height=10.5 cm}}
222: \caption{The single-particle energy spectrum of the zero magnetization.}
223: \label{E50}
224: \end{figure}
225:
226: The single-particle energy spectrum obtained by diagonalizing this Hamiltonian
227: with these bond displacements is shown in Fig. (\ref{E50}). It is observed
228: that
229:
230: \begin{itemize}
231: \item{There are an equal number of positive and negative energy levels.}
232: \item{For every positive energy level there is a corresponding negative energy
233: level which is equal in magnitude but opposite in sign.}
234: \item{Each level is doubly degenerate. Exceptions to this fact are the maximum
235: level (1), the minimum level (-1), the largest negative level (-0.0642) and the
236: smallest positive level (0.0642) which are non-degenerate.}
237: \item{There is a \textit{gap} between the largest negative level and
238: smallest positive level which is twice of $\delta$. It is to be noted that all
239: the quoted energies are in units of $J$.}
240: \end{itemize}
241:
242: The ground-state energy of this system is obtained by summing the lowest
243: fifty
244: levels i.e., all the states lying below the gap (negative energy states). This
245: is the state of the system at zero temperature and
246: its energy is -31.8634.
247:
248: The upper half of Fig. (\ref{D50}) gives the variation of spin
249: density
250: with the site index. It can be seen that this remains constant at half
251: throughout the chain as expected. Also the bond displacements, single-
252: particle
253: energies and spin densities remain independent of an applied magnetic field for
254: small values of the field because the
255: net magnetization in this case is zero.
256:
257:
258: \subsection{For magnetization=1 (for $N_F$=51)}
259:
260: \begin{figure}[!h]
261: \centerline{\psfig{figure=D51.ps,height=10.5 cm}}
262: \caption{The bond displacements and expectation values of $S_n^z$ for
263: magnetization=1.}
264: \label{D51}
265: \end{figure}
266:
267: \begin{figure}[!h]
268: \centerline{\psfig{figure=E51.ps,height=10.5 cm}}
269: \caption{The single-particle energy spectrum for magnetization=1.}
270: \label{E51}
271: \end{figure}
272:
273: The bond displacements for the case of 51 fermions i.e., one away from
274: dimerization, obtained by minimizing the Hamiltonian in Eq.
275: (\ref{NFH}) for the case of zero magnetic field using the same procedure, are
276: shown in the lower part of Fig. (\ref{D51}). It can be seen that the
277: displacements of the bonds oscillate around the value zero and have an
278: envelope whose
279: periodicity is 50, their maximum magnitude being 0.058. The pattern obtained is
280: similar to that of beats in
281: sound waves.
282:
283: The single-particle energy spectrum obtained in this case is shown in Fig.
284: (\ref{E51}). The features of the spectrum that get modified compared to
285: $N_F=50$
286: are
287:
288: \begin{itemize}
289:
290: \item{There is an increase in the number of non-degenerate levels by four. The
291: doubly degenerate levels lying next to the non-degenerate levels of the last
292: case become non-degenerate.}
293: \item{The \textit{size} of the energy gap existing between the greatest
294: negative level and
295: smallest positive level falls from 0.0128 to 0.02.}
296: \end{itemize}
297:
298: The ground-state energy of this system is now obtained by summing the lowest
299: fifty one levels and its value is -31.8219.
300:
301: The upper half of the Fig. (\ref{D51}) gives the variation of spin
302: density
303: with the site index. The pattern obtained is seen to be similar to
304: that of bond displacements except that the spin density maximum occurs at
305: the
306: site where the bond displacement is minimum and the oscillations within each
307: envelope remain always above the value half. Also, the area under each of these
308: envelopes, say from $n=35$ to $n=85$, is half and the sum of the areas under
309: the
310: two envelopes is equal to unity. This implies the extra ($51^{st}$) fermion
311: distributes itself into these two envelopes with equal probability. Thus, the
312: solitons with half
313: integral spin are seen in the system.
314:
315: \subsection{For magnetization=49 (for $N_F$=99)}
316:
317: \begin{figure}[!h]
318: \centerline{\psfig{figure=D99.ps,height=10.5 cm}}
319: \caption{The bond displacements and expectation values of $S_n^z$ for
320: magnetization=49.}
321: \label{D99}
322: \end{figure}
323:
324: The bond displacements for the magnetization of the system being 49 or
325: fermion filling being just one less than the maximum allowed, for the case of
326: zero magnetic field are shown in the lower part of Fig. (\ref{D99}). It can
327: be seen that the system has a single peak ($\sim 0.18$) in bond
328: displacements
329: localized over a small region chosen randomly by the system by spontaneously
330: breaking the translation symmetry.
331:
332:
333: The upper half of Fig. (\ref{D51}) gives the variation of spin
334: density
335: with the site index. It can be seen that there is a dip in the spin density
336: over the region where there is a peak in bond displacements, showing that
337: there is a localization of the missing ($100^{th}$) fermion over a small
338: region.
339:
340: \subsection{For N=102, magnetization=18 (for $N_F$=69)}
341:
342: \begin{figure}[!h]
343: \centerline{\psfig{figure=D69.ps,height=10.5 cm}}
344: \caption{The bond displacements and expectation values of $S_n^z$ for
345: magnetization=18.}
346: \label{D69}
347: \end{figure}
348:
349: The bond displacements for a system size of 102 spins, $N_F$ being one
350: away from trimerization i.e, 69, obtained by minimizing the Hamiltonian in Eq.
351: (\ref{NFH}) for the case of zero magnetic field are shown in the lower part of
352: Fig. (\ref{D69}). The bond displacements of the lattice spins are seen to
353: oscillate around the value zero, their maximum magnitude being 0.022 and have
354: an
355: envelope whose periodicity is 34. The pattern is similar to $N_F=51$ case.
356: Also,
357: the single-particle energy spectrum is seen to be gapless in this case.
358:
359: The upper half of Fig. (\ref{D51}) gives the variation of spin
360: density with the site index. The pattern obtained is similar to that of bond
361: displacements except that the spin density maximum occurs at the site where the
362: bond displacement is minimum and oscillations within each
363: envelope remain always above the value 0.667. Also, the area under each of
364: these
365: envelopes, say from $n=24$ to $n=57$, is 0.333 and sum of the areas under all
366: three envelopes is unity. This means that the extra ( $69^{th}$) fermion
367: distributes
368: itself into the three envelopes with equal probability. Thus, solitons with
369: fractional (1/3) spin are seen here in the system.
370:
371:
372: \section{Variation of ground-state magnetization with magnetic field}
373: \begin{figure}[h]
374: \centerline{\psfig{figure=em.ps,height=10.5 cm}}
375: \caption{The variation of ground-state energy with magnetization $m$ for zero
376: magnetic field $h$.}
377: \label{em}
378: \end{figure}
379:
380: The single-particle energy spectrum, the ground-state energies and the spin
381: densities at zero temperatures are calculated for each of the magnetizations (0
382: to 50) possible in the system. Fig. (\ref{em}) shows the variation of
383: ground-state energy $E_0$ with magnetization $\textit{m}$, with the magnetic
384: field term set to zero in the Hamiltonian (\ref{NFH}).
385:
386:
387:
388: \begin{figure}[h]
389: \centerline{\psfig{figure=mzm.ps,height=10.5 cm}}
390: \caption{The variation of ground-state magnetization $\textit{m}$ with
391: magnetic \mbox{field $\textit{h}$.}}
392: \label{mzm}
393: \end{figure}
394:
395:
396:
397: The variation of ground-state magnetization of the spin-Peierls system with
398: magnetic field at zero temperature is now studied numerically, by varying the
399: magnetic field $\textit{h}$ from -1.1 to 1.1 in steps of 0.001 in the Zeeman
400: term, $ h\sum_{n=1}^{N}[C_n^+C_n-\frac{1}{2}] $, of the Hamiltonian
401: (\ref{NFH}). The ground-state energies for each of the possible magnetizations
402: are
403: calculated and the magnetization which minimizes the ground-state energy is
404: identified for each value of the magnetic field. The plot in Fig. (\ref{mzm})
405: shows the variation of ground-state magnetization
406: $\textit{m}$ with magnetic field $\textit{h}$. It is seen that the saturation
407: magnetic field of this system is 1.035 and above this, the system exists only in
408: the ferromagnetic state.
409:
410:
411: \section{Dimerized to uniform phase transition}
412:
413: The dimerized to uniform structural phase transition in the spin-Peierls
414: system
415: shown in figure (\ref{PF}) is obtained numerically, for the case of zero
416: magnetic
417: field (\textit{h}=0) and zero magnetization ($\textit{m}=0$, $N_F$=50), as
418: discussed below.
419:
420: \begin{figure}[h]
421: \centerline{\psfig{figure=NP.ps,height=10.5 cm}}
422: \caption{Dimerized to uniform phase transition at zero magnetic field.}
423: \label{NP}
424: \end{figure}
425:
426: By taking the bond displacements to be of the form $\delta_n=\delta (-1)^n$,
427: which
428: is true for dimerized phase, the value of $\delta$ that minimizes the free
429: energy is
430: obtained for each non-zero temperature, by following the same procedure
431: mentioned in section (2.2), except that free energy is minimized instead of the
432: ground-state energy. The free
433: energy \textit{U} at any temperature \textit{T}, using the Fermi-Dirac
434: statistics, is given by
435:
436: \begin{eqnarray}
437: U=-\frac{1}{k_BT}\sum_{n=1}^{n=N}{\rm log}\left[1+{\rm exp}\right(\frac{E(n)
438: -h}{k_B T}\left) \right]
439: \end{eqnarray}
440:
441: \noindent where $E(n)$ are the single-particle energies at temperature
442: \textit{T}, and $k_B$ is the Boltzmann constant \cite{7}.
443:
444: The step sizes chosen in varying the temperature and the magnitude of the
445: bond displacement while minimizing the free energy
446: are both equal to 0.0001. Fig. (\ref{NP}) shows the second order phase
447: transition, the critical parameter being the magnitude of the bond
448: displacement,
449: \textit{$\delta$}, plotted as a function of temperature, \textit{T}. The
450: critical
451: temperature as obtained from the graph is 0.036. It is to be noted
452: that though the system is 1D in the Hamiltonian (\ref{NFH}) it still has a
453: phase
454: transition. This is because our calculation is equivalent to a mean-field
455: theory
456: which allows a non-zero displacement $\delta$. In a real $CuGeO_3$ system the
457: interchain couplings justify this mean-field treatment and led to a finite
458: temperature phase transition.
459:
460: \section{Calculation of the phonon frequencies }
461: \subsection{Analytical method}
462: In Section (2.2), single-particle energies were calculated assuming static
463: lattice displacements. In this section, we will calculate the phonon
464: frequencies using perturbation theory.
465:
466: The simplified spin-Peierls Hamiltonian is given by the Eq. (\ref{NFH}). The
467: $i^{th}$ bond displacement, $\delta_{i+\frac{1}{2}}$, is given by
468: \begin{center}
469: \begin{math}
470: \delta_{i+\frac{1}{2}}=-\lambda\left(u_{i+1} - u_{i}
471: \right) \nonumber
472: \end{math}
473: \end{center}
474: \noindent
475:
476: Now the first order change in the ground-state energy (at zero
477: temperature)
478: $E_0$ due to the displacement of $i^{th}$ bond proportional to
479: $\partial{E_0}/\partial{\delta_{i+\frac{1}{2}}}$.
480: This is zero since $E_0$ is calculated by minimizing \textit{H} in Eq.
481: (\ref{NFH}) with respect to all the bond displacements. Hence to
482: realize the contribution of bond displacements to a change in the ground-
483: state energy, we have to find the derivative of second order with respect to the
484: bond displacements.
485:
486:
487: Defining the elements of the dynamical matrix \textit{D} to be
488:
489: \begin{center}
490: \begin{math}
491: D_{{i+\frac{1}{2}},{j+\frac{1}{2}}}=\frac{\partial^2{E_0}}{\partial{\delta_{i+
492: \frac{1}{2}}}\partial{\delta_{j+\frac{1}{2}}}}
493: \end{math}
494: \end{center}
495:
496: \noindent where all the derivatives are evaluated at those values of $\delta$'s
497: which are obtained by minimizing the Hamiltonian (\ref{NFH}) \cite{6}. Using
498: this
499: definition for \textit{D} we get the lattice Hamiltonian to be
500:
501: \begin{equation}
502: H_{ph}=\frac{1}{2}M\sum_i \dot{u_i}^2+\frac{1}{2}
503: \sum_{i,j}D_{i+\frac{1}{2}i,j+\frac{1}{2}}[\lambda(u_{i+1}-u_i)+\delta_{i+\frac
504: {1}{2}}^{(0)}][\lambda(u_{j+1}-u_j)+\delta_{j+\frac{1}{2}}^{(0)}]
505: \label{LH}
506: \end{equation}
507: \noindent where $u_i$ is the displacement of the ion at the
508: lattice point $i$, $M$ its mass,
509: $\delta_{i+\frac{1}{2}}^{(0)}$ is the equilibrium value of
510: $\delta_{i+\frac{1}{2}}$ which minimizes the Hamiltonian and
511: $\lambda$ is the positive constant which is a measure of the coupling that
512: exists between the lattice points and spins.
513:
514: Now we define a transformation on displacements of lattice points as shown
515: below,
516:
517: \begin{equation}
518: \tilde{u}_i=u_i-c_i
519: \label{UT}
520: \end{equation}
521:
522: \noindent we see that with periodic boundary conditions, and with the choice
523: $c_1=0$ we
524: have
525: \begin{center}
526: \begin{math}
527: \lambda(c_{i+1}-c_i)+\delta_{i+\frac{1}{2}}^{(0)}=0
528: \end{math}
529: \end{center}
530: since $\sum_i\delta_{i+\frac{1}{2}}=0$.
531:
532: Writing the lattice Hamiltonian (\ref{LH}) in terms of $\tilde{u}_i$'s, thus
533: simplifies it to the form
534:
535: \begin{eqnarray}
536: H_{ph}&=&\frac{1}{2}M\sum_i
537: \dot{\tilde{u}_i}^2+\frac{\lambda^2}{2}\sum_{i,j}D_{i,j}
538: (\tilde{u}_{i+1}-\tilde{u}_i)(\tilde{u}_{j+1}-\tilde{u}_j) \nonumber
539: \end{eqnarray}
540:
541: \noindent Simplifying it further by defining
542:
543: \begin{equation}
544: \tilde{D}_{i,j}=D_{i-\frac{1}{2},j-\frac{1}{2}}+D_{i+\frac{1}{2},j+\frac{1}{2}}
545: -D_{i-\frac{1}{2},j+\frac{1}{2}}-D_{i+\frac{1}{2},j-\frac{1}{2}}
546: \end{equation}
547:
548: \noindent we get the lattice Hamiltonian $H_{ph}$ (\ref{LH}) to be
549:
550: \begin{eqnarray}
551: H_{ph}&=&\frac{1}{2}M\sum_i
552: \dot{\tilde{u}_i}^2+\frac{\lambda^2}{2}\sum_{i,j}\tilde{D}_{i,j}
553: \tilde{u}_i\tilde{u}_j
554: \label{MLH}
555: \end{eqnarray}
556:
557: Now the frequency, $\omega_i$, of the $i^{th}$ phonon, (proportional
558: to its energy) is given by
559:
560: \begin{equation}
561: \omega_i=\lambda\sqrt{\frac{\mu_i}{M}}
562: \label{freq}
563: \end{equation}
564:
565: \noindent where $\mu_i$ is the $i^{th}$ eigenvalue of the transformed dynamical
566: matrix,
567: $\tilde{D}$, and $M$ is the mass of each lattice point \cite{6}.
568:
569: \begin{figure}[h]
570: \centerline{\psfig{figure=phonon2.ps,height=10.5 cm}}
571: \caption{Square root of eigenvalues of the transformed dynamical matrix for
572: the system in dimerized phase.}
573: \label{phonon}
574: \end{figure}
575:
576: The transformed dynamical matrix for the dimerized phase is
577: evaluated assuming bond displacements of the form
578: $\delta_i=\delta (-1)^i$
579: and the root of the eigenvalues of the corresponding Hamiltonian are plotted in
580: an increasing order as shown in Fig. (\ref{phonon}). This curve
581: gives the phonon dispersion for the dimerized phase accurate upto a constant
582: factor given
583: by Eq. (\ref{freq}) \cite{8}.
584: While evaluating the
585: elements of the transformed dynamical matrix, various symmetries like inversion
586: symmetry about the bonds, translational symmetry of lattice points, etc., are
587: used to generate all the elements from just the first few rows or columns.
588:
589: \subsection{Numerical method}
590: The eigenfunctions of the Hamiltonian, $H$ for the case of dimerized phase
591: (zero magnetization) are
592: used to evaluate the transformed dynamical matrix using the second order
593: perturbation theory as shown below
594:
595: \begin{eqnarray}
596: && \tilde{D}(m,n) \nonumber \\
597: && =\sum_{j,k}\left[\frac{<n+1|H|n><n|H|n+1>+<m+1|H|m><m|H|m+1>}{E(j)-E(k)}
598: \right] \nonumber \\
599: &&
600: \end{eqnarray}
601:
602: \noindent where N is the system size, j represents an occupied single-particle
603: state and
604: takes the values from 1 to 50, k represents an unoccupied single-particle state
605: and ranges
606: from 51 to 100, $E(j)$ and $E(k)$ denote the single-particle energies, $m$ and
607: $n$ denote the site numbers with $|m>$ representing a state in which the
608: $m^{th}$ site is occupied.
609: While most of the matrix elements of $\tilde{D}$ obtained using the above
610: equation are found to agree very well with those determined by the analytical
611: method described in the previous section, a few off-diagonal elements are seen
612: to differ slightly.
613:
614: