1: \documentclass[12pt]{article}
2: \usepackage{amsmath}
3: \usepackage{amssymb}
4: \usepackage{mathtext}
5: \usepackage{graphicx}
6:
7: \textwidth=15cm \textheight=21cm
8:
9: \begin{document}
10: \section*{DYNAMICS OF THREE VORTICES\\ ON A PLANE AND A SPHERE --- II.\\
11: General compact case}\footnote{REGULAR AND CHAOTIC DYNAMICS, V.3, No.~2, \ 1998\\
12: {\it Received August 15, 1998}\\
13: AMS MSC 76C05}
14:
15:
16: \begin{centering}
17: A.\,V.\,BORISOV\\
18: Faculty of Mechanics and Mathematics\\
19: Department of Theoretical Mechanics
20: Moscow State University\\
21: Vorob'ievy gory, 119899 Moscow, Russia\\
22: E-mail: borisov@uni.udm.ru\\
23: V.\,G.\,LEBEDEV\\
24: Physical Faculty,\\
25: Department of Theoretical Physics
26: Udmurt State University \\
27: Universitetskaya, 1, Izhevsk, Russia, 426034 \\
28: E-mail: lvg@uni.udm.ru\\
29: \end{centering}
30:
31: \begin{abstract}
32: Integrable problem of three vortices on a plane and sphere are considered.
33: The classification of Poisson structures is carried out.
34: We accomplish the bifurcational analysis using the variables introduced in
35: previous part of the work.
36: \end{abstract}
37:
38: \section{Introduction}
39:
40: In this part of our work the integrable problem of dynamic of vortices
41: on planes and sphere are considered. We base on Hamiltonian form of reduced
42: systems, which poisson structure was given in the previous part \cite{BP}.
43:
44: The problem of motion of two vortices on plane was completely
45: investigated by G.\,Helmholz \cite{Melesh}, who had established,
46: that in general case two vortices
47: make an uniform rotary motion around of vorticity centre, with frequency
48: $$\Omega = \frac{\Gamma _ 1 + \Gamma _ 2}{2\pi r ^ 2}, $$
49: where $ \Gamma _ k $ is intensity of point vortices.
50: The vorticity centre thus is moving uniformly and rectilinearly.
51: If $ \Gamma _ 1 = -\Gamma _ 2 $
52: the vorticity centre is situated on infinity, and two vortices are moving
53: forward.
54:
55: The problem about a motion of three vortices is much more complex. Despite of
56: numerous works, first of which were the dissertation of ~Gr{\"o}bli 1877
57: \cite{Grobli} and Greenhill's research \cite{green}, the complete and evident
58: classification of a motion still does not exist.
59: From the point of view of integrability problem, this
60: problem was considered by Poincare in the treatise on the vortices theory
61: \cite{Poincare}. He wrote out explicitly a complete set of noncommutative
62: integrals.
63: In a modern period the problem of a motion of three vortices on a plane was
64: studied in works \cite{Aref,Aref2,Rott,TT,Novikov},
65: and from the point of view of the topological analysis --- in
66: \cite{Selivanova}. Unfortunately, these works have added a little to
67: achievements of classicists both
68: in presentation, and in completeness of the description of motions. Their
69: basic contents come down either to geometrical interpretation of a
70: motion and computer modeling of separate trajectories, or to some general
71: topological constractions, that is not coordinated to physical behaviour of
72: system. Partly, it is caused by the fact, that the problem of three vortices on a
73: plane does not belong to that integrable systems, which complete analysis is
74: possible in a class of rather simple (for example, elliptic) special functions
75: (the general solution has indefinitely--sheeted branching on complex plane
76: of time because of logarithmic terms in Hamiltonian). The exception consists
77: of some special cases (for example, the case of equal intensities of
78: vortices). In comparison with a problem of three vortices on a plane, a
79: motion of three vortices on sphere, that is also being integrable, is not
80: investigated at all.
81:
82: We give here new analysis, based on algebraic and geometrical research of
83: the reduced system (in variable, which in the previous part of work \cite{BP}
84: were named ``internal'', as though we are not adhere this term anymore),
85: and then we analyze absolute motion, without use of exact quadrature. Such
86: approach, basing on representation of the equations of motions on algebra
87: \cite{BP}, allows to establish some analogies between problem of three
88: vortices and Euler-Poinsot case in rigid body dynamics, and also to receive
89: more evident description of motions of system.
90:
91: \section{Algebraic classification}
92:
93: As it was shown in the previous part of work \cite{BP}, the equation of a
94: motion of three vortices on the plane can be written down as Hamiltonian system,
95: determined by Lie-Poisson brackets in form of
96: \begin{equation}
97: \label{Tpb}
98: \begin{array}{l}
99: \displaystyle \{M _ i, M _ j \} =-4a _ k\Delta,\medskip\\
100: \displaystyle \{M _ i, \Delta \} = (a _ j -a _ k) M _ i +
101: (a _ j + a _ k) (M _ j-M _ k), \\
102: \end{array}
103: \end{equation}
104: and by Hamiltonian
105: \vspace{-4mm}
106: \begin{equation}
107: \label{Tpa}
108: H = -{1\over 8\pi}(\Gamma _{2} \Gamma _{3} \log M _{1} + \Gamma _{1}
109: \Gamma _{3} \log M_{2} +
110: \Gamma _{1} \Gamma _{2} \log M _{3}).
111: \end{equation}
112: where $M_k$ are squares of pairwise distances between vortices,
113: $\Delta$ is a size of the oriented area pulled on three vortices
114: (here and further we shall assume, that the indexes $ i, j, k $ accept
115: accordingly values $ 1,2,3 $ and their cyclic rearrangements, and the
116: coefficients $ a _ 1, a _ 2, a _ 3 $ are inverse intensities $ a _ k = 1 /
117: {\Gamma _ k}). $
118:
119: Lie-Poisson brackets (\ref{Tpb}) are degenerated and have two central functions.
120: One of them (linear) is integral of the complete moment
121: \vspace{-3mm}
122: \begin{equation}
123: \label{Tpá}
124: D = \sum _ k a _ k M _ k,
125: \end{equation}
126: the other (square-law Casimir function) arises from geometrical
127: Henon ratio, that connects the area of triangle with sides
128: \begin{equation}
129: \label{Tpd}
130: F = (2\Delta) ^ 2 + M _ 1M _ 1 + M _ 2M _ 2 + M _ 3M _ 3-2 (M _ 1M _ 2 + M _ 1M
131: _ 3 + M _ 2M _ 3).
132: \end{equation}
133: For real motions $ F = 0. $
134:
135: The real type of Lie-Poisson algebra (\ref{Tpb}) depends on values of
136: intensities $ \Gamma _ 1, \Gamma _ 2, \Gamma _ 3. $
137: In fact, choosing a new basis in form $D,e_1,e_2,e_3$, where $D$ is
138: defined by~(\ref{Tpá})
139: \begin{equation}
140: \label{Tp2}
141: \begin{array}{l}
142: \displaystyle e _ 1 = \frac{\Delta}{2\sqrt {A}}, \\
143: \displaystyle e _ 2 = \frac {(a _ 2-a _ 3) M _ 1 + (a _ 3-a _ 1) M _ 2 + (a _ 1-
144: a _ 2) M _ 3}{2\sqrt {2AB}}, \\
145: \displaystyle e _ 3 = \frac
146: {(a _ 2a _ 3-a _ 1 ^ 2) M _ 1 + (a _ 1a _ 3-a _ 2 ^ 2) M _ 2 + (a
147: _ 1a _ 2-a _ 3 ^ 2) M _ 3}{2A\sqrt {2B}} \\
148: \end{array}
149: \end{equation}
150: where
151: $$A=|a_1a_2+a_2a_3+a_1a_3|,$$
152: $$B=(a_1-a_2)^2+(a_2-a_3)^2+(a_1-a_3)^2,$$
153: under the condition
154: \vspace{-4mm}
155: \begin{equation}
156: \label{Tp5}
157: (a_1a_2+a_2a_3+a_1a_3)>0,
158: \end{equation}
159: we receive that the algebra of vortices is decomposed in the direct sum
160: $l(4)\approx$ $\mathbb{R}\oplus so(3):$
161: \begin{equation}
162: \label{Tp7}
163: \{D,e_k\}=0\mbox{, }
164: \{e_1,e_2\}=e_3\mbox{, }
165: \{e_2,e_3\}=e_1\mbox{ }
166: \{e_3,e_1\}=e_2,
167: \end{equation}
168: and by condition
169: \vspace{-4mm}
170: \begin{equation}
171: \label{Tp6}
172: (a_1a_2+a_2a_3+a_1a_3)<0,
173: \end{equation}
174: in direct sum $l(4)\approx$ $\mathbb{R}\oplus so(2,1):$
175: \begin{equation}
176: \label{Tp8}
177: \{D,e_k\}=0,\quad \{e_1,e_2\}=-e_3,\quad
178: \{e_2,e_3\}=e_1,\quad
179: \{e_3,e_1\}=e_2.
180: \end{equation}
181: Though for case equal intensities the coefficient $B=0$,
182: using the passage to the limit, it is easy to show that basis
183: (\ref{Tp2}) can be correctly determined in this case as well.
184:
185:
186: Let's consider, what features of a motion can be obtained from the structure
187: of such algebraic decomposition. Using squared Casimir function for
188: algebra $ so(3) $
189: \begin{equation}
190: \label{Tp9}
191: G ^ 2 = e _ 1 ^ 2 + e _ 2 ^ 2 + e _ 3 ^ 2,
192: \end{equation}
193: and expressing in it a vector $ e _ 1 $ with the help of Heron ratio
194: (\ref{Tpd})
195: with $ F = 0, $ we discover, that relative dynamics of vortices with
196: intensities, satisfying to a condition (\ref{Tp5}) will be equivalent to a
197: motion of some "representing" point on symplectic sheet, that is a
198: two-dimensional sphere with radius, determined by value of linear Casimir
199: function
200: \begin{equation}
201: \label{Tp10}
202: e _ 1 ^ 2 + e _ 2 ^ 2 + e _ 3 ^ 2 =\frac 1 {16}
203: \left (\frac {a_1a_2a_3 D}{a _ 1a _
204: 2 + a _ 2a _ 3 + a _ 1a _ 3} \right
205: ) ^ 2.
206: \end{equation}
207: The situation for algebra $ so (2,1) $ is similar that arise
208: in a case (\ref{Tp6}). Using Casimir function of algebra~$so(2,1)$
209: \begin{equation}
210: \label{Tp17}
211: G ^ 2 = e _ 3 ^ 2-e _ 2 ^ 2-e _ 2 ^ 2,
212: \end{equation}
213: it is possible to notice, that the relative motion of vortices is reduced
214: to the motion of point on surface of two-cavities hyperboloid, that is
215: determined by condition,
216: \begin{equation}
217: \label{Tp18}
218: e _ 3 ^ 2-e _ 1 ^ 2-e _ 2 ^ 2 =\frac 1 {16}
219: \left (\frac {a_1a_2a_3D}{a _ 1a _ 2 +
220: a _ 2a _ 3 + a _ 1a _ 3} \right
221: ) ^ 2.
222: \end{equation}
223: In case of (\ref{Tp5}), motions of representing point (and therefore
224: three vortices in system of vorticity centre) is finite for every $D$,
225: therefore this case we shall futher name ``compact'',
226: and in case (\ref {Tp6}), with which obviously may be exist running up
227: trajectories by additional conditions for $\Gamma_i$ and for initial
228: vortices positions, the name ``uncompact''.
229: %????
230:
231: Motions, arising under condition
232: \begin{equation}
233: \label{Tp20}
234: ( a _ 1a _ 2 + a _ 2a _ 3 + a _ 1a _ 3) = 0
235: \end{equation}
236: require separate consideration. In basis
237: \begin{equation}
238: \label{Tp23}
239: \begin{array}{l}
240: \displaystyle D=\frac {a_1M_1+a_2M_2+a_3M_3}{(a_1+a_2)^2},\\
241: \displaystyle e_1=\frac {\Delta}{(a_1+a_2)},\\
242: \displaystyle e_2=\frac {M_3}{2(a_1+a_2)},\\
243: \displaystyle e_3=\frac {(a_1+a_2+a_3)M_3-(M_1+M_2)(a_1+a_2)}{2(a_1+a_2)^2},\\
244: \end{array}
245: \end{equation}
246:
247: Lie-Poisson brackets can be reduced to form
248: \begin{equation}
249: \label{Tp27}
250: \displaystyle \{ D,e_k\}=0,\quad \displaystyle \{ e_1,e_2\}=e_3,\quad
251: \displaystyle \{ e_3,e_1\}=-D,\quad \displaystyle \{ e_2,e_3\}=e_1.
252: \end{equation}
253: In this case algebra of vortices is not decomposed in the straight sum of
254: one-dimensional and three-dimensional algebras, and becomes four-dimensional
255: solveable algebra with maximal solveable ideal $N = \{D, e _ 1, e _ 3 \}.$
256: Algebra has squared Casimir function, following from (\ref{Tpd})
257: \begin{equation}
258: \label{Tp28}
259: G ^ 2 = e _ 1 ^ 2 + e _ 3 ^ 2-2De _ 2.
260: \end{equation}
261: Real dynamics of vortices occurs on symplectic sheet,
262: that is set by conditions $G = 0$ and $D~=~const,$ which
263: determine paraboloid, taking place through the beginning of coordinates.
264: By $ D = 0 $, being necessary collaps condition (merge of vortices),
265: paraboloid degenerates to the straight line, con\-ter\-mi\-nous with an axis
266: $ e _ 3.$
267:
268: \section{Bifurcational analysis of motion of vortices on plane}
269:
270: Let's describe most evident geometrical interpretation of motions, used in
271: \cite{TT} and presented on~Fig.~1. In space
272: $ M _ 1, M _ 2, M _ 3 $ the level of linear integral of the moment sets a
273: plane.
274: The inequalities $ M _ k> 0 $ allocate on it the area, in which motion
275: occurs. It is
276: necessary to exclude from this area unphysical values of distances, for
277: which an
278: inequality of a triangle $ (\Delta^ 2 < 0)$ does not hold. This area is shown
279: at
280: a figure by black colour. When approaching to it, in the equations of a
281: motion
282: for variable $M _ 1, M _ 2, M _ 3, $ we schould change the sign of time and
283: direction of motion. It is necessary to note, that
284: the described geometrical interpretation is not equivalent to a motion of
285: representing point on symplectic sheet, and changes of direction of motions
286: are the consequence of features of a projection from space
287: $ \Delta, M _ 1, M _ 2, M
288: _ 3 $ in space $ M _ 1, M _ 2, M _ 3. $
289:
290: Let's hold condition (\ref{Tp5}), then bracket (\ref{Tpb}) is defines algebra
291: $so(3)\oplus \mathbb{R}$. In this case symplectic sheet is compact ($\mathbb{S}^2$), and
292: motions of vortices are finite. Let's assume, at first, that signs of all
293: intensities are identical --- in this case condition (\ref{Tp5}) is obviously
294: fullfiled. The first integrals of a motion of vortices on a plane we shall
295: write down as
296:
297: \begin{equation}
298: \label{Tp11}
299: \begin{array}{l}
300: \displaystyle D = a_1 M_1 + a_2 M_2 + a_3 M_3,\medskip\\
301: \displaystyle h = M_1^{a_1} M_2^{a_2} M_3^{a_3}, \\
302: \end{array}
303: \end{equation}
304: (integral of energy, for convenience, is presented in
305: exponential form). The integrals (\ref{Tp11}) are dependent, i.e.
306: Jacobi matrix of the first integrals (\ref{Tp11}) is degenerate
307: \begin{equation}
308: \label{Tp12}
309: Rank \left (\frac {\partial (D, h)}{\partial (M _ 1, M _ 2, M _ 3)} \right) < 2
310: \end{equation}
311: only in one case: $ M _ 1 = M _ 2 = M _ 3 $
312: (in three-dimensional space $ M _ 1, M _ 2, M _ 3 $ the vortices will form
313: correct triangle). Curves, resulting with crossing
314: of levels of integrals (\ref{Tp11}) are similar to polhodes
315: in rigid body dynamics (Poinsot interpretation of Euler case)
316: and, in a compact case, are ovals (Fig.~1). The solution as correct
317: triangle occurs with a contact of a surface $ h $ and plane $ D $
318: in one point. They limit by energy the area of possible motions (APM) from
319: above and the corresponding bifurcational curve has a form
320: \vspace{-5mm}
321: \begin{equation}
322: \label{Types13}
323: H (D) = \left (\frac {D}{\sum a _ i}\right)^{(\sum a_i)}.
324: \end{equation}
325: {\bf Remark 1.} {\it The particular solutions, appropriate
326: to the given curve --- three vortices in tops of correct triangle,
327: rotating as a rigid body around of vorticity centre are refered as
328: ``Thomson's" and are steady with fulfilment of a condition (\ref{Tp5}). J.
329: J. Thomson has specified them for any number of vortices of equal
330: intensity, and has shown, that in linear approach such configurations will
331: be steady for number of vortices $ N\leqslant 6 $, and with $ N\geqslant 7
332: $ --- unstable (Thomson theorem) \cite{ThomsonJ}. However because of
333: presence of resonances in system the linear approach is not sufficient.
334: The analysis of stability in nonlinear approach with use of Birkhoff
335: normalization was carried out in \cite{Hazin}. It was found, that the
336: theorem Thomson is fair in the exact statement --- in sense of Laypunov
337: stability.}
338:
339: The other change of motion occurs with a contact of a crossing
340: line of integrals' levels with curves, determined by the equation
341: $\Delta (M_1, M_2, M_3) = 0$ $(\sqrt {M_i} + \sqrt {M_j} = \sqrt {M_k}),$
342: that arise from inequalities of triangles and limit physically
343: allowable area of values (see Fig.~1).
344:
345: Points of a contact correspond to collinear configuration of three vortices.
346: Three vortices thus settle down on one straight line and rotate as a unit
347: around of vorticity centre.
348:
349: {\bf Remark 2.} {\it Collinear and the triangular configurations in dynamics of three
350: vortices have
351: analogues in the classical selestial mechanics \cite{ArnoldKozlovNei}. There
352: are, accordingly, Euler and Lagrange particular solutions of three body
353: problem.}
354:
355: Collinear configurations are obtained from a condition of a contact
356: \begin{equation}
357: \label{Tp13}
358: \Delta = 0, \quad \dot \Delta = 0.
359: \end{equation}
360: that lead, having same sedate functional dependence,
361: which in a general case can be presented as
362: \begin{equation}
363: \label{Tp14}
364: H(D)=f(a_1, a_2, a_3)D^{(a_1+a_2+a_3)},
365: \end{equation}
366: where $f(a_1,a_2,a_3)$ some function from parameters.
367: Passing to homogeneous coordinates $ x, y $
368: \begin{equation}
369: \label{Tp15}
370: M_1 = xM_3,\quad M_ 2 = yM _ 3,\quad y = (1 + x) ^ 2, \quad x = z ^ 2,
371: \end{equation}
372: we discover, that collinear configurations correspond to positive
373: roots of the equations of the third degree
374: \begin{equation}
375: \label{Tp16}
376: \alpha_1 (1\pm z) (z\pm 2) + \alpha _2 ((1\mp z) z-\alpha _3 (1\pm 2z) z (1\pm
377: z) = 0,
378: \end{equation}
379: where
380: $$\alpha _1 ={\Gamma _2 + \Gamma _3}, \quad
381: \alpha _2 ={\Gamma _1 + \Gamma _3}, \quad
382: \alpha _3 ={\Gamma _1 + \Gamma _2}. $$
383: The lines of crossing of levels $ D $ and $ h $ can concern borders of area
384: of physically possible motion $ \Delta = 0 $ in space of variables
385: $ (M _ 1, M _ 2, M _ 3) $ by various ways (Fig.~1). Depending on it the
386: quantity of bifurcational curves also will be different. In particular, in a
387: case, when all of intensities are different $ a _1 \ne a _2\ne a _3\ne a _1 $
388: three branches corresponding to collinear motions will exist (Fig.~2a).
389: With equality of two of them, for example, $ a _ 1 = a _ 2\ne a _ 3 $
390: two appropriate bifurcational curves merge. And, at last, in case of equal
391: intensities $ a _ 1 = a _ 2 = a _ 3 $ all three curves degenerate in one.
392:
393: The calculation of angular speed of Thomson's and collinear configurations
394: concerning vorticity centre is given, for example in \cite{Melesh}. It
395: monotonously decreases with increase of the complete moment of system
396: vortices.
397:
398: As a measure of stability of stationary configurations it is possible
399: to use a square of untrivial eigenvalues $\lambda^2$ of
400: linearized system of the equations of a motion of three vortices. For
401: Thomson's configurations it is simple to receive the explicit formula
402: \begin{equation}
403: \label{Tpf}
404: \lambda ^ 2 = -3a _ 1a _ 2a _ 3\frac {(a _ 1 + a _ 2 + a _ 3) ^ 3}{D ^ 2},
405: \end{equation}
406: that shows, that in a considered case they are steady, in
407: difference from collinear, which as shown on Fig. 2a a are unstable.
408:
409: Generally speaking, the following statement is fair, that is confirmed
410: by geometrical interpretation on Fig.~1:
411: If Thomson's solution is steady, the appropriate set
412: of collinear solutions are unstable, and on the contrary.
413: Therefore condition (\ref {Tpf}) determines a type of
414: stability not only Thomson's configurations, but also collinear
415: configurations with the given values of $D.$
416:
417: Let's consider other case of a motion of three vortices,
418: with fullfiled conditions (\ref{Tp5}) (easy to see, that all other possible
419: cases, when the condition is fair, could be reduced to two considered cases).
420: Let's assume, that the intensity of one of vortices has opposite sign in
421: comparison with other two (for example, $ \Gamma _ 1 = 1/a _ 1 < 0 $). The
422: condition (\ref{Tp5}) in this case means, that $ -\Gamma _ 1 < \Gamma _ 2 +
423: \Gamma _ 3 $ (i.e. intensity of chosen vortex is greater than the intensity
424: of other two).
425:
426: Let's specify only main differences of this problem from previous. First of
427: all from the formula~(\ref{Tpf}) for Thomson's configurations it follows,
428: that they are unstable. Collinear configuration (it is not three of them
429: anymore, but only one) thus, on the contrary, is steady. Bifurcational curve
430: of appropriate unique collinear configuration (see Fig.~2b) is above a
431: curve, appropriate to Thomson's configuration (it is simple to show, that
432: under condition of~(\ref{Tp5}) complete moment is always more than
433: zero), and the area of possible motion coincides with the whole square
434: $ D> 0, \quad h> 0 $.
435:
436: To obtain more complete representation about the considered motions, let's
437: study a phase portrait of system of three vortices on a plane in symplectic
438: coordinates.
439:
440: \section{Symplectic coordinates for vortices on a plane}
441:
442: Since symplectic sheets represent two-dimensional spheres
443: (\ref{Tp11}), it is convenient to use cylindrical coordinates $ l, L $ as
444: canonical coordinates (we shall keep for them a designation of Andoyer--Deprit
445: coordinates, adopted in rigid body dynamics \cite{BorEm}) for algebra
446: $ so (3): $
447: \begin{equation}
448: \label{Tp18a}
449: \begin{array}{l}
450: \displaystyle e_1 = L, \\
451: \displaystyle e_2 = \sqrt {G^2-L^2} \sin l, \\
452: \displaystyle e_3 = \sqrt {G^2-L^2} \cos l. \\
453: \end{array}
454: \end{equation}
455: Expressions for variables $ M _ 1, M _ 2, M _ 3 $ in
456: canonical coordinates look
457: like
458: \begin{equation}
459: \label{Tp19}
460: \begin{array}{ll}
461: M_i=&\frac {\displaystyle a_j+a_k}{\displaystyle \sqrt {2}}{\displaystyle \frac
462: {D}{A}}+2F(a_j-a_k)F\sqrt {\frac {\displaystyle A}{\displaystyle B}}\sin l+\medskip\\
463: &{\displaystyle \frac{4F}{\sqrt{B}}}(a_j(a_j-a_i)+a_k(a_k-a_i))\sqrt{\frac{2}{B}} \cos l,\\
464: \end{array}
465: \end{equation}
466: where
467: $$ A = a _ 1a _ 2 + a _ 2a _ 3 + a _ 3a _ 1, $$
468: $$ B = (a _ 1-a _ 2) ^ 2 + (a _ 2-a _ 3) ^ 2 + (a _ 1-a _ 3) ^ 2, $$
469: $$ F = \sqrt {G ^ 2-L ^ 2}. $$
470:
471: In that special case of equal intensities the derived expressions
472: (\ref{Tp17}) lead to the expression for Hamiltonian mentioned in \cite{Aref,
473: Khanin}. In last work it is used the classical way of a reduction to a
474: problem with one degree of freedom for arbitrary intensities (such as
475: Jacobi's unit elimination).
476:
477: The phase portrait in canonical coordinates describes a motion
478: of representing point $ (L, l) $ on a surface of two-dimensional sphere. The
479: pulse $L$ has a value of the oriental area parallelogram, that is
480: constructed
481: on three vortices $(L = 2\Delta)$. Development of a phase portrait
482: is presented on Fig.~4a--d for four different specific cases of the
483: intensities of vortices ratio.
484:
485: The given figures illustrate the analogy between the motion of three vortices
486: and rigid body dynamics. Comparing Fig. 2a (in case of equal
487: intensities) with phase
488: portrait of a Euler-Poinsot problem (see for example \cite{BorEm}),
489: it is possible to connect a collinear configuration (situated on a straight
490: line $ L = 0 $) with unstable permanent motions of a rigid body around of an
491: middle axis of ellipsoid of inertia, the Thomson's solutions (with $ L/G = 1$)
492: --- with rotations around of a large (small) axis of ellipsoid of inertia.
493: The periodic solutions of a problem of two vortices (two of three vortices
494: always are in one point, and their intensities are added), situated on
495: straight line $ L = 0 $ it is possible to connect with steady permanent
496: rotations around of a small (large) axis of elipsoid of inertia. When system
497: pass through the collinear state ( three vortices on one straight line ---
498: $ L = 0 $), oriented area changes the sign. Deformation of a phase picture
499: with unequal intensities is shown on Fig.~4b,c. Thus Thomson's solutions
500: are displaced with straight $ L/G = 1, $ that on sphere should be imagined as
501: being pulled together in a point). Fig.~4d present the phase portrait, with
502: one negative intensity, but condition (\ref{Tp5}) is also fullfiled.
503:
504: Other motions of system of three vortices, that is not a state of equilibrium on
505: phase portraits, correspond to quasiperiodic motions of vortices on a plane.
506: Depending on the topology of a phase curve these motions have
507: various features --- some of them repeatedly pass
508: through the collinear state, others --- never get into it. Such areas
509: are separated by separatrix curve, for that curve three vortices are
510: approach to collinear (unstable) configuration infinitely long.
511: The particular trajectories of vortices are given in various works
512: \cite{Aref2, Aref, Rott, Melesh,TT}, beginning from Gr\"obli~\cite{Grobli}
513: and Greenhill \cite{green}.
514:
515: Appropriate simplectic coordinates for algebra $ so (2,1) $ could be
516: chosen as
517: \begin{equation}
518: \label{Tp20a}
519: \begin{array}{l}
520: \displaystyle e _ 1 = L, \\
521: \displaystyle e _ 2 = \sqrt {G ^ 2 + L ^ 2} \mbox {\, sh}\,l, \\
522: \displaystyle e _ 3 = \sqrt {G ^ 2 + L ^ 2} \mbox {\, ch}\,l. \\
523: \end{array}
524: \end{equation}
525: where $G,$ as well as in a case $so(3)$, parametrize various
526: simplectic sheets. The difference of the case (\ref{Tp6})
527: is displayed in noncompactness of symplectic sheet, therefore
528: the leaving of vortices to infinity is possible, that is excluded for
529: condition (\ref{Tp5}).
530:
531:
532: \section{Three vortices on sphere}
533:
534: The problem of motion of vortices on sphere is more interesting, but less
535: investigated. This problem has more important applied meaning in comparison
536: with a problem of motion of vortices on a plane, because it is directly
537: connected with physics of atmosphere. Many substantial problems (collapse
538: and evasion of vortices, existence of steady stationary (static)
539: configurations)
540: can be investigated on model of an ideal liquid. Their solutions can become a
541: basis for more complex models (connected, for example, with presence of small
542: viscosity), that describe real dynamic vortical processes (motion of
543: cyclones and etc.) in an atmosphere of the Earth.
544:
545: Dynamics of two vortices on sphere is quite similar to a flat case.
546: Here the general situation is the rotation around of an axis, that is passing
547: through centre of spheres and analogue of
548: vorticity centre (point located on chord, connecting two
549: vor\-te\-ces with radius-vector
550: $$\vec r = \frac {\Gamma _ 1\vec r _ 1 + \Gamma _ 2\vec r _ 2}{\Gamma _ 1 +
551: \Gamma _ 2}, $$
552: where $\vec r _ 1, \vec r _ 2 $ --- radius-vector of the first and second
553: vortices). As well as in a flat case distance between vortices remains
554: constant.
555: Angular speed of such rotation
556: \begin{equation}
557: \label{Sp111}
558: \Omega = \frac {1}{2\pi M} \sqrt {(\Gamma _ 1 + \Gamma _ 2) ^ 2-\frac {\Gamma
559: _ 1\Gamma _ 2M}{R ^ 2}},
560: \end{equation}
561: where $ R $ --- radius of sphere, $ M $ --- square of distance between the
562: vortices (square of the length of a chord, that connect the location of two vortices
563: on sphere). Under condition of $ \Gamma _ 1 = -\Gamma _ 2 $, two vortices are
564: moving on sphere on two identical parallels, located on different parties from
565: equator, that was noticed Gromeka \cite{Grom}.
566:
567: The problem of motion of three vortices on sphere was considered in work
568: \cite{Bog1}, in that work probably for the first time the integrability of the
569: problem was specified.
570: The equations of motion of three vortices on sphere in
571: variables of mutual distances (chords) and volumes can be
572: written down~\cite{BP} as the Hamiltonian equations with nonlinear algebra of Poisson brackets
573: \begin{equation}
574: \label{Sp1}
575: \begin{array}{l}
576: \displaystyle \{M _ i, M _ j \} =-\frac {4a _ k\Delta} {R}, \\
577: \displaystyle \{M _ i,
578: \Delta \} = R (a _ j-a _ k) M _ i + R (a _ j + a _ k) (M _ j-M _ k) + \frac {M _
579: i}{2R} (a _ kM _ k-a _ jM _ j), \\
580: \end{array}
581: \end{equation}
582: that also have two central functions, one of
583: them (linear) is the same as (\ref{Tpá}), and the
584: second (cubic) function has the form
585: \begin{equation}
586: \label{Spd}
587: \begin{array}{rl}
588: F=&(2\Delta) ^ 2 + R^2(M _ 1M _ 1 + M _ 2M _ 2 + M _ 3M _ 3)-2 R^2(M _ 1M _ 2 +\medskip\\
589: &M _ 1M _ 3 + M _ 2M _ 3)+M_1M_2M_3 = 0,
590: \end{array}
591: \end{equation}
592: and Hamiltonian (\ref{Tpa}).
593:
594: Let's notice, that linear approximation of structure (\ref{Sp1})
595: coincides with the Poisson bracket of three vortices on a plane. Besides,
596: it is simple to show, that after replacement
597: $$
598: \frac {d\tau}{dt} = \pm\frac {\Delta}{2\pi M _ 1M _ 2M _ 3}
599: $$
600: the equations of a motion of vortices turn in Lotky--Volterra system of the form
601: \begin{equation}
602: \label{Spl}
603: \frac d{d\tau} M_i=\Gamma_iM_i(M_j-M_k),
604: \end{equation}
605: similarly to the flat case \cite{BP}. This trajectory isomorphism is also
606: piecewise \cite{BP}.
607:
608: Geometrical interpretation for a plane, presented on
609: Fig.~1, can be also transferred to sphere. Curious and
610: earlier, probably, not marked fact is that phase trajectories in variable
611: $M_1,M_2,M_3$ for a case of sphere and plane,
612: with given intensities, coincide with phase trajectories
613: of the same Lotky--Volterra system (\ref{Sp1}). Thus main effects
614: in dynamics of vortices are determined by the part of phase
615: trajectories of Lotky--Volterra system gets that in area
616: $\Delta^2(M_1,M_2,M_3)>0.$ The motion of vortices occurs only
617: in this part, because with the approach of trajectories to border of area
618: (and achievement by vortices collinear state) in the equations (\ref{Spl})
619: it is necessary to change a sign of time.
620: The form of this area is a little various
621: for a plane and sphere (formula (\ref {Tpd}) and (\ref {Spd}))
622: and represented on Fig.~7.
623:
624: {\bf Remark 3.} {\it The analogy with Lotky--Volterra system is
625: very useful for study of vortices collaps problem and is related to
626: Bolin regularization in the \cite{ArnoldKozlovNei} (the
627: difference is, that in the Kepler problem is fixed a constant of energy
628: level). Problems of regularization and collapse, that arise only in noncompact
629: cases, will be discussed in the next part of work.}
630:
631: {\bf Remark 4.} {\it The issue of reduction of
632: three vortices problem on sphere to canonical Hamiltonian system
633: with one degree of freedom is complex and, probably, it could not be solved
634: constructively. If one use the initial form of record \cite{BP},
635: then one obtain the six-dimensional Hamiltonian system, that have as the
636: integrals Hamiltonian function and noncommutative set of integrals of moment,
637: each of them is nonlinear function of phase variable (in difference from a
638: case of a plane considered in \cite{Khanin}).}
639:
640: {\it The canonization of the reduced system also problematic,
641: in variable $ (\vec M, \Delta) $ the canonization is equivalent to
642: introduction of Darboux coordinates on two-dimensional symplectic sheet
643: determined by a common level of Casimir functions. It require the the
644: solution of nonlinear system of the partial differential equations.}
645:
646: {\it Really, the nonlinear algebra of Poisson brackets (\ref{Sp1}) cannot
647: be investigated in same detail, as in the flat case. Linear
648: approximation of this structure is capable to make some
649: qualitative conclusions only for a situation
650: close to simultaneous collaps (i.e. when distance between
651: vortices is small in relation to radius of curvature).
652: From it, nevertheless, it is possible to make a conclusion that
653: necessary conditions of simultaneous collaps of vortices on a plane
654: are also fair for the case of sphere, since the influence of nonlinear terms
655: is small enough near the collaps.}
656:
657: {\it The most simple form the bracket (\ref{Sp1}) has in case of three equal
658: intensities at a zero level of linear Casimir function $ (D = 0). $
659: After transition (for linearization of a bracket (\ref{Sp1})) to
660: canonical basis (\ref{Tp2}) it is possible to present bracket}
661: \begin{equation}
662: \label{Sp22}
663: \begin{array}{l}
664: \displaystyle \{e _ 1, e _ 2 \} = e _ 3 + \frac {1}{2} (e _ 3 ^ 2-e _ 2 ^ 2),
665: \\
666: \displaystyle \{e _ 2, e _ 3 \} = e _ 1, \\
667: \displaystyle \{e _ 3, e _ 1 \} = e _ 2-e _ 2e _ 3. \\
668: \end{array}
669: \end{equation}
670: {\it It is curious to note, that nonlinear terms, included in a bracket,
671: (\ref{Sp22}), are similar to the terms, originating in Kovalevsky integral
672: of the Euler--Poisson equations \cite{BP}.}
673:
674: {\it One of main sections of sheet for a case
675: $ \Gamma _ 1 = \Gamma _ 2 = \Gamma _ 3 $ are shown on Fig.~5 in
676: dependence from curvature of sphere $ k = 1/R $.
677: With $ k = 0, $ the symplectic sheet, that is determined by algebra
678: $ So (3) $, is also sphere. The change $ k $ results in a surface, whose
679: coherent component is homeomorphic to sphere. It would be interesting to
680: investigate the topology of a symplectic sheet with various values of
681: intensities and values of integral of the moment $ D $, that is the limited
682: function on sphere (on which vortices move).
683: Last remark allows to reject uncompact
684: components of symplectic sheets, since a motion on sphere is always
685: finite.}
686:
687: {\bf Remark 5.} {\it The general principle of the classification of vortices motions on sphere
688: given below is the continuation on parameter of complete moment of stationary
689: configurations that is known near $ D = 0. $ In last case the influence of
690: curvature is not significant, that corresponds already investigated
691: flat problem (it is also equivalent to consideration of linear
692: approximation of structure (\ref{Sp1})).}
693:
694: The conditions for Thomson's of configurations on sphere are similar to
695: conditions on plane
696: $$ M _ 1 = M _ 2 = M _ 3. $$
697: Conditions for existence of collinear configurations
698: $ (\Delta = \dot \Delta = 0) $ in case of sphere are reduced
699: to system of three algebraic equations.
700: Introducing constants
701: $$\alpha _ 1 = \Gamma _ 3 + \Gamma _ 2, \quad \alpha _ 2 = \Gamma _ 1 + \Gamma _
702: 3, \quad
703: \alpha _ 3 = \Gamma _ 2 + \Gamma _ 1,\quad
704: \beta _ 1 = \Gamma _ 3-\Gamma _ 2, \quad \beta _ 2 = \Gamma _ 1-\Gamma _ 3,
705: \quad
706: \beta _ 3 = \Gamma _ 2-\Gamma _ 1, $$
707: we write down system of the equations as
708: \begin{equation}
709: \label{Vr14}
710: \begin{array}{l}
711: \displaystyle 2 (M _ 1M _ 2 + M _ 1M _ 3 + M _ 2M _ 3) - (M _ 1 ^ 2 + M _ 2 ^ 2
712: + M _ 3 ^ 2) -\frac {M _ 1M _ 2M _ 3}{R ^ 2} = 0, \\[12pt]
713: \displaystyle a _ 1M _ 1 + a _ 2M _ 2 + a _ 3M _ 3-D = 0, \\[12pt]
714: \displaystyle
715: 2 (\alpha _ 1\frac {(M _ 3-M _ 2)} {M _ 1} + \alpha _ 2\frac {(M _ 1-M _ 3)} {M
716: _ 2} + \alpha _ 3\frac {(M _
717: 2-M _ 1)}{M _ 3} + \\[12pt]
718: \displaystyle (\beta _ 1M _ 1 + \beta _ 2M _ 2 + \beta _ 3M _ 3) \frac {1} {R ^
719: 2} = 0, \\[12pt]
720: \end{array}
721: \end{equation}
722: where $D$ is the value of integral of the moment. The given system can be solved
723: numerically. Results of numerical research of bifurcational diagrams
724: for cases of various values of intensities are presented on~Fig.~6.
725:
726: {\bf Remark 6.} {\it As well as in case of a plane we are interesting
727: only in positive roots of the equations~(\ref{Vr14}).
728: Besides we should notice, that range of values of functions, determined by
729: integrals of energy and moment are limited by values
730: $ D _{m} $ and $ E _ {m} $}
731: $$ D _{m} = 4R ^ 2\frac {(a _ 1a _ 2 + a _ 2a _ 3 + a _ 1a _ 3) ^ 2} {a _ 1a
732: _ 2a _ 3}, $$
733: $$
734: E _{m} = (4R ^ 2)^{\sum a_k} \prod_k \left ({\frac
735: { a _ 1 + a _ 2 + a _ 3}{a _ 1 + a _ 2 + a _ 3 + a _ 1a _ 2a _ 3/a _ k ^ 2}}
736: \right)^{a _ k},
737: $$
738: {\it In the latter case appropriate value of the complete moment is calculated
739: by the formula}
740: $$
741: d_E=(4R^2)(a_1+a_2+a_3)\sum_k {\frac {a _ k}
742: {(a_1+a_2+a_3+a_1a_2a_3/a_k^2)}}.
743: $$
744:
745: As initial approach for the solutions of the equations (\ref{Vr14})
746: with small values of the complete moment we chose the roots of the
747: equations (\ref{Tp15}). Continuation of bifurcational curves with increase
748: of the moment (mutual distances) was made by predictor-corrector method
749: with use of the Newton iterative procedure.
750:
751:
752: For each set of intensities we give the bifurcational diagrams and
753: plots of relative and absolute motions, on which we show the value of angular
754: speed, corner of an inclination of a plane of vortices under the relation
755: to an axis of rotation, square of eigenvalue, that is determining the
756: stability of linearized system.
757:
758:
759: Interesting effect of a spherical motion of vor\-te\-ces, absent in flat case, is
760: the birth of new (and in a considered case steady) collinear con\-fi\-gu\-ra\-tions
761: from a problem of two vortices by values $d_i=4R^2(a_j+a_k)$. There is a dis\-in\-te\-gra\-tion of one vortex of
762: total intensity $\Gamma_i+\Gamma_j$ to two vortices with intensities
763: $\Gamma_i$ and $ \Gamma_j$ with increase of $D.$ As shown in Fig.~6, these
764: configurations aspire to merge with collinear configurations, turning out
765: with con\-ti\-nu\-a\-tion on parameter
766: from a flat problem (with small $ D/R ^ 2 $), and then cease to exist, with
767: further increase of $ D.$
768:
769:
770: Let's consider bifurcational diagram for a case $a _ 1\ne a _ 2\ne a _ 3\ne
771: a_1$ (see Fig.~6a). The Thomson's configuration (top curve) exist
772: only by energy $ E\le E_{T}.$ With $ D = d _ T $, by reaching maximum
773: possible value $ E_{T}<E_{m}$, Tomson's configuration merges
774: (point A) with collinear configuration, determined by the most bottom
775: branch (appropriate to bottom branch on Fig.~2 bifurcational
776: diagram for plane). Thus all three vortices lay in an equatorial plane, but distance between
777: them are not equal each other. Passing through a maximum of energy $E_{m}$
778: by $D=d_E$, this configuration evolve futher with increases of $D$ to a
779: problem of two vortices (with $D=d_3$). Two other
780: collinear configurations appropriate to collinear curves on Fig.~2 for
781: the plane, merge with collinear solutions, arising from
782: problems of two vortices with values of the moment $d_1, d_2$
783: (disintegration of one vortex).
784: Such arising configurations are absent in case of a plane.
785: The occurrence could be understand from Fig.~8, that shown take off
786: physical area from borders, on that the moment achieved value $M_k=0.$
787:
788: In a case, if the ratio $a_1=a_2\ne a_3$ is true, there is a merge not only
789: Thomson's and collinear branch in a point $A$, but also collinear
790: configurations, arising by $D=d_1=d_2$ (see. Fig.~6b).
791:
792: At last for three equal intensities all possible
793: configurations merge together in a point $A$ by the maximal values
794: energy $E=E_m=E_T$ and moment $D=d_T.$
795:
796: In an absolute motion with increase of the mo\-ment (mutual distances)
797: angular speed of rotation of Thomson's configurations monotonously
798: decreases (see~Fig.~10), and corner of an inclination normal of a triangle to an
799: axis of rotation grows monotonously also from zero (as for a plane) up to value
800: $\pi/2$ by the moment of merge of Thomson's and collinear configurations
801: (Fig.~9), by exception of a case equal intensities, when an inclination of an
802: axis change not.
803:
804: {\bf Remark 7.} {\it The only possible motions of stationary
805: configurations (i.e. configurations, with distance between vortices do not
806: varies) are the rotations around of some motionless axis.
807: It is a consequence of representability of the equations of vortical dynamics
808: as the first order equations relatively to dynamic variables.}
809:
810: In contrast of Thomson's configuration, axis of rotation of collinear
811: configurations always lays in a plane of vortices and does not vary by
812: change values of the moment.
813:
814: With increasing of the moment the diagrams of angular speed
815: are slumped monotonously and disappearance with merging.
816: It is necessary to note large size of angular speeds of vortices, arising at the
817: moment of birth of a new vortex from a problem two vortices. Within the
818: framework of the accepted model this increase of angular speed concerns
819: to different trajectories, but if is present "weak" dissipation, that it's
820: possible observe slow evolution both values of energy and moment. Certainly,
821: there are the additional condition existing of such effects in a physical
822: situation more important is the stability of appropriate stationary
823: motions.
824:
825: Part of collinear configurations, occurring from similar configurations on a plane,
826: are unstable already in linear approach, as it
827: is visible from Fig.~11, because they being analogues configurations on a plane.
828: However collinear configurations, appearing from a problem two vortices,
829: are steady. The nature of this stability is well visible from geometrical
830: interpretation presented on Fig.~8. It is possible, that the phenomena of such
831: grades occurring in an atmosphere of the Earth (that have obviously
832: dissipation). That is crucial for occurrence of various catastrophic processes
833: (type hurricanes), accompanying sharp reorganization of dynamics of vortical
834: formations. Inverse process of collaps (merge) of two vortices,
835: that is impossible for model of an ideal liquid, in case of small dissipation
836: and reduction $ D $ can result in formation of atmospheric vortices with
837: large angular speed of rotation.
838:
839: The Thomson's solutions are steady up to the moment of passage through
840: static configuration. The analogue of the formula (\ref{Tpf}) for Thomson's
841: configurations on sphere is the expression
842: \begin{equation}
843: \label{Tpg}
844: \lambda ^ 2 = \frac {D-3R^2(a _ 1 + a _ 2 + a _ 3)}{9D^2} a_1a_2a_3
845: (a _ 1 + a _ 2 + a _ 3).
846: \end{equation}
847: That shows, that Tomson's configuration is unstable with value the moment
848: $$ D> 3R ^ 2 (a _ 1 + a _ 2 + a _ 3),$$
849: appropriate to the maximal value, for such configurations.
850:
851: For a case of one negative intensity
852: $ (\Gamma _ 1 < 0, \quad -\Gamma _ 1> \Gamma _ 2 + \Gamma _ 3) $ bifurcational
853: diagram is given on~Fig.~6d. In this case behaviour
854: bifurcational curves by increase $ D $ is similar to behaviour by already
855: considered situations. Existing in case of a plane Thomson's and
856: collinear configurations merge in a point $A$ (see Fig.~6d), and after this
857: disappear by merging from one of collinear branches, that birth from a problem t
858: of two vortices. Difference from a case only positive by increase $D$
859: intensities is displayed in existence of collinear solutions, not
860: limited on energy from above. These solutions occur also
861: due to stationary configurations of a problem of two vortices.
862: The change of parameters of an absolute motion of vortices is qualitative by
863: nothing differs from a case positive intensity except for that, that all of
864: collinear configuration are steady, in that time as Thomson's --- is unstable
865: (Fig.~11).
866:
867: In summary we shall explicitly allocate the basic differences of a spherical
868: case from flat, arising with increase of the complete moment $ D: $
869: \begin{enumerate}
870: \item arising of Thomson's and collinear configurations on sphere;
871: \vspace{-3mm}
872: \item arising of steady collinear configurations from a problem of two
873: vortices rotating with large (infinite) angular speed in the moment of
874: occurrence;
875: \vspace{-3mm}
876: \item an inclination and evolution of a plane Thomson's configurations;
877: \vspace{-3mm}
878: \item existence of static configurations on sphere.
879: \end{enumerate}
880: \vspace{-2mm}
881: Researches of other opportunities of a motion of three vortices on a plane and
882: to sphere resulting to collaps and which is running up to trajectories will be
883: are given in the following part of work.
884:
885: The authors thank I.S.Mamaev and N.N.Simakov for useful discussions and help in work.
886: The work is carried out under the support of Russian Fond of Fundamental Research
887: (96--01--00747) and Federal programm "States Support of Integration High Education
888: and Fundamental Science" (pro\-ject~No.~294).
889:
890:
891:
892: \begin{thebibliography}{99}
893:
894: \bibitem{ThomsonJ}{\it J.\,J.\,Thomson.}{ A treatise on the motion of
895: Vortex rings.} London: Macmillan, 1883, 124~p.
896:
897: \bibitem{Grobli}{\it W.\,Gr\"obli} Specialle Probleme \"uber die Bewegung
898: Geredliniger paralleler Wirbelf\"aden". Vierteljahrsch. D. Naturforsch.
899: Geselsch. Z\"urich. V.~22, 1877, P.~37--81, P.~129--165.
900:
901: \bibitem{Poincare} {\it H.\,Poincar\'e.} { Th\'eorie des tourbillous.} Paris,
902: 1893, 205~p.
903:
904: \bibitem{Grom} {\it I.\,S.\,Gromeka.} { On vortex motions of liquid on a
905: Sphere.} Collected papers. Moscow, AN USSR, 1952, P.~296 (in Russia).
906:
907: \bibitem {green} A.\,G.\,Greenhill. { Plane vortex motion.} Quat. J. of
908: Pure appl. Math., 1878, V.~15, P.~10--29.
909:
910: \bibitem{Hazin} {\it L.\,G.\,Hazin.} { Regular poligons of point vortices and
911: Resonance instability of stationary states.} DAN USSR,
912: V.~230, 1976, No. ~4, P.799--802 (in Russian).
913:
914: \bibitem{Barut} {\it A.\,Barut, R.\,Raczka} {Theory of Group
915: Representations and Applications.}
916: PWN. Polish Scientific Publishers, 1977.
917:
918: \bibitem{Bog2} {\it V.\,A.\,Bogomolov.} { Dynamics of vorticity on a sphere.}
919: Mech. of liquid and gas, 1977, No. ~6, P.~57--65 (in Russian).
920:
921: \bibitem{Bog1} {\it V.\,A.\,Bogomolov.} { Two-dimensional fluid dynamics
922: on a sphere}. Izv. Acad. Sci. USSR Atmos. Oceanic Phys., V.~15, 1979,
923: No. 1, P.~29--35 (in Russian). % others in article P.~18-22
924:
925: \bibitem{Aref2} {\it H.\,Aref.} { Integrable, chaotic and turbulent vortex motion
926: in two-dimensional flows.} Ann. Rev. Fluid Mech. V.~15, 1983, P.~345--389.
927:
928: \bibitem{ArnoldKozlovNei} {\it V.\,I.\,Arnold, V.\,V.\,Kozlov, A.\,I.\,Neishtadt}
929: { Mathematical aspects of the classical and selestial mechanics.}
930: Moskow: VINITI. Results of a science and engineering. Contemparary
931: problems of mathematics. Fundamental directions,
932: V.~3, 1985, 304~p.
933:
934: \bibitem{Aref} {\it H.\,Aref.} { Motion of three vortices revisited.} Phys.
935: Fluids. V.~31, 1988, No. ~6, P.~1392--1409.
936:
937: \bibitem{Rott} {\it N.\,Rott, H.\,Aref.} { Three-vortex motion with zero total
938: circulation.} J. Of Appl. Math. And Phys. (ZAMP), V.~40, 1989, P.~473--500.
939:
940: \bibitem{Melesh} {\it V.\,V.\,Meleshko, M.\,Yu.\,Konstantinov} {Dynamics of
941: vorticity structure.} Kiev, Naukova dumka, 1993, 277~p.
942:
943: \bibitem{Selivanova} {\it E.\,N.\,Selivanova.} {Topology of problem
944: of three dot vortices.} Works Math. Inst. RAN. V.~205, 1994, P.~141--149.
945:
946: \bibitem{BorEm} {\it A.\,V.\,Borisov, K.\,V\,Emelyanov} { Nonintegrability
947: and stochacticity of rigid body dynamics.} Izhevsk, Isd-vo Udm. univ., 1995,
948: 57~p.
949:
950: \bibitem{Bagrets} {\it A.\,A.\,Bagrets, D.\,A.\,Bagrets.} { Nonintegrability of two
951: problems in vortex dynamics.} Chaos, V.~7, 1997, Issue 3, P.~368--375.
952:
953: \bibitem{BP} {\it A.\,V.\,Borisov, A.\,E.\,Pavlov.} {Dynamics and statics
954: of vortices on a plane and a sphere.} Reg. \& Chaot. Dyn., 1998, V.~3, No. ~1,
955: P.~28--38.
956:
957: \bibitem{TT} {\it J.\,Tavantzis, L.\,Ting.} { The dynamics of three
958: vortices revised.} Phys. Fluids, 1988, V.~31, No. ~6, P.1392--1409.
959:
960: \bibitem{Novikov} {\it E.\,A.\,Novikov, Yu.\,B.\,Sedov.} {
961: Collaps of vortices.} Zh.Eksp.Teor.Fiz., V.~77, 1979,
962: No. ~2(8), P.~588--597.
963:
964: \bibitem{Khanin} {\it K.\,M.\,Khanin.} { Quasi-periodic motions of vortex
965: systems.} Physica 4D., V.~3, 1982, P.~261-269.
966: \end{thebibliography}
967:
968: \begin{figure}[ht!]
969: $$
970: \includegraphics{meta.eps}
971: \includegraphics{metb.eps}
972: $$
973: \caption{Dark part corresponds to nonphysical area
974: on a plane, that is set by linear integral $ D = const $ and limited
975: by conditions $ M _ k> 0, $ for a compact case:
976: a) All of intensities are positive;
977: b) One of intensities is negative.
978: The point $A$ corresponds to the Tomson's solution; $F$ ---
979: collilear solutions; $B, C, D$ --- points, in with which two vortices are
980: merged in one.}
981: \end{figure}
982:
983: \begin{figure}[ht!]
984: $$
985: \includegraphics{bifpa.eps}
986: \includegraphics{bifpb.eps}
987: $$
988: \caption{The view of bifurcation diagram for three vortices on a plane
989: in compact case:
990: a) With positive intensities;
991: b) One negative intensity. The dot-dashed line corresponds to the Thomson{'}s
992: solution, continuous lines --- to collinear configurations.}
993: \end{figure}
994:
995: \begin{figure}[ht!]
996: $$
997: \includegraphics{bifupa.eps}
998: \includegraphics{bifupb.eps}
999: $$
1000: \caption{Dependence of a squares of eigenvalues,
1001: determining stability in linear approach from integral of complete
1002: moment for stationary configurations in a compact case:
1003: a) With positive intensities;
1004: b) One negative intensity.
1005: The dot-dashed line corresponds to the Thomson's solutions,
1006: continuous lines --- to collinear configurations.}
1007: \end{figure}
1008:
1009: \begin{figure}[ht!]
1010: \begin{center}
1011: \begin{tabular}{cc}
1012: \includegraphics{phap.eps} & \includegraphics{phbp.eps} \medskip\\
1013: \includegraphics{phcp.eps} & \includegraphics{phdp.eps}\\
1014: \end{tabular}
1015: \end{center}
1016: \caption{Phase plane for cases:
1017: a) Positive equal intensities;
1018: b) Case $ a _ 1 = a _ 2 \ne a _ 3 $
1019: c) Three positive various intensities;
1020: d) One negative intensity.}
1021: \end{figure}
1022:
1023: \begin{figure}[ht!]
1024: $$
1025: \includegraphics{sealg.eps}
1026: $$
1027: \caption{Deformation of section of symplectic sheet
1028: for linearized algebra of vortices on sphere depending on change
1029: of sphere's radius of curvature.}
1030: \end{figure}
1031:
1032: \begin{figure}[ht!]
1033: \begin{center}
1034: \begin{tabular}{cc}
1035: \includegraphics{bifa.eps} & \includegraphics{bifb.eps} \medskip\\
1036: \includegraphics{bifc.eps} & \includegraphics{bifd.eps}\\
1037: \end{tabular}
1038: \end{center}
1039: \caption{The bifurcational diagram for cases:
1040: a) Various positive intensities;
1041: b) Positive intensities (two coincide);
1042: c) Equal positive intensities;
1043: d) One negative intensity.
1044: The designations of curves coincide with Fig.~2.}
1045: \end{figure}
1046:
1047: \begin{figure}[ht!]
1048: \begin{tabular}{c}
1049: \includegraphics{lva.eps}\\
1050: \includegraphics{lvb.eps}\\
1051: \end{tabular}
1052: \caption{Geometrical picture is isomorfic with a Lotky-Volterra
1053: problem. A part of a phase plane of Lotky-Volterra, that is
1054: limited by a contour, corresponds to physical area of a problem of three
1055: vortices: a) on a plane; b) on sphere with a set of identical intensities.}
1056: \end{figure}
1057:
1058: \begin{figure}[ht!]
1059: $$
1060: \includegraphics{meaa.eps}
1061: \includegraphics{mebb.eps}
1062: \includegraphics{mecc.eps}
1063: $$
1064: \caption{Geometrical picture of birth of collinear solutions
1065: from a problem of two vortices. Consecutive figures a), b), c) show change
1066: of points of a contact of borders of physically of allowable area
1067: with increase of the complete moment.
1068: The designations correspond Fig.~1.}
1069: \end{figure}
1070:
1071: \begin{figure}[ht!]
1072: \begin{center}
1073: \begin{tabular}{cc}
1074: \includegraphics{omc.eps} & \includegraphics{omb.eps} \medskip\\
1075: \includegraphics{oma.eps} & \includegraphics{omd.eps}\\
1076: \end{tabular}
1077: \end{center}
1078: \caption{Angular speeds for cases:
1079: a) Various positive intensities;
1080: b) Positive intensities (two coincide);
1081: c) Equal positive intensities;
1082: b) One negative intensity.}
1083: \end{figure}
1084:
1085: \begin{figure}[ht!]
1086: $$
1087: \includegraphics{anga.eps}
1088: \includegraphics{angd.eps}
1089: $$
1090: \caption{Corners of an inclination for cases:
1091: a) Various positive intensities;
1092: b) One negative intensity.}
1093: \end{figure}
1094:
1095: \begin{figure}[ht!]
1096: \begin{center}
1097: \begin{tabular}{cc}
1098: \includegraphics{usta.eps} & \includegraphics{ustb.eps} \medskip\\
1099: \includegraphics{ustc.eps} & \includegraphics{ustd.eps}\\
1100: \end{tabular}
1101: \end{center}
1102: \caption{Factor of stability $\lambda^2$ for cases:
1103: a) Various positive intensities;
1104: b) Positive intensities (two coincide);
1105: c) Equal positive intensities;
1106: d) One negative intensity.}
1107: \end{figure}
1108:
1109:
1110:
1111:
1112:
1113: \end{document}
1114: