A new solvable many-body problem of goldfish type

Transcription

1 Oksana Bihun Department of Mathematics, University of Colorado, Colorado Springs, 40 Austin Bluffs Pkwy, Colorado Springs, CO 8098, USA Francesco Calogero Physics Department, University of Rome La Sapienza p. Aldo Moro, I-0085 Roma, Italy Istituto azionale di Fisica ucleare, Sezione di Roma, Italy A new solvable many-body problem of goldfish type is introduced and the behavior of its solutions is tersely discussed. Keywords: Solvable -body problems; many-body problems; isochronous systems; completely periodic solutions; goldfish type systems. 000 Mathematics Subject Classification: 70F0, 70K4.. Introduction otation.. Hereafter is (unless otherwise indicated) an arbitrary integer,, the (generally complex) numbers z n z n (t) are the dependent variables, t ( time ) is the independent variable, superimposed dots denote time-differentiations, and indices such as n, m, l run over the integers from to unless otherwise indicated (see for instance below in (.a) the limitation l n on the values of l). Below we often omit to indicate explicitly the t-dependence of various quantities, when this can be done without causing misunderstandings. Hereafter matrices are denoted by upper-case boldface letters (so that, for instance, the matrix C has the elements C nm ). Lowercase boldface letters stand for -vectors (so that, for instance, the -vector z has the components z n ); and the imaginary unit is denoted by i (so that i =, and i is of course not a -vector!). We occasionally use the Kronecker symbol, with its standard definition: δ mn = for m = n, δ mn = 0 for m n. And let us mention the standard convention according to which an empty sum vanishes and an empty product equals unity, i. e. K j=j ( f j ) = 0, K j=j ( f j ) = if K < J. A prototypical goldfish many-body model [ 4] is characterized by the translation-invariant equations of motion z n = i ω ż n + l=, l n ( ) żn ż l. (.a) z n z l

2 Oksana Bihun and Francesco Calogero A Hamiltonian producing these equations of motion reads as follows: [ H (ζ ;z) = n= i ω z n + exp(ζ n ) ] (z n z l ), (.b) l=, l n where of course the coordinates ζ n ζ n (t) are the canonical momenta corresponding to the canonical particle coordinates z n z n (t). The solution of the corresponding initial-values problem is provided by the roots z n z n (t) of the following, rather neat, algebraic equation in the variable z: l=, l n [żl (0) + i ω z l (0) z z l (0) ] = i ω exp(i ω t). (.c) (ote that this is actually a polynomial equation of degree in z, as seen by multiplying it by m= [z z m (0)]). Hence this model is isochronous (whenever the parameter ω is positive, as we generally assume hereafter; the special case ω = 0 is the prototypical, nonisochronous, case...): all its solutions are completely periodic, with the period T = π/ω or, possibly, due to an exchange of the particle positions through the motion, with a period that is a (generally small with respect to its possible maximal value!, see [5]) integer multiple of T. Several solvable generalizations of the goldfish model, characterized by ewtonian equations of motion featuring additional forces besides those appearing in the right-hand side of (.a), are known: see for instance [ 4] and references therein. Remark.. Above and hereafter we call a many-body model solvable if its initial-values problem can be solved by algebraic operations, such as finding the zeros of a known t-dependent polynomial of degree (of course such an algebraic equation can be explicitly solved only for 4). Recently a simple technique has been introduced [6], which allows to identify and investigate additional solvable models of goldfish type; and a few examples of such models yielded by this new approach have been identified and tersely discussed [6]. The model treated in this paper is another such new model, which is perhaps itself interesting (as all solvable models tend to be), and moreover allows as reported in a separate paper [7] to obtain remarkable Diophantine results for the zeros of (monic) polynomials of degree the coefficients of which are the zeros of Hermite polynomials (see, for instance, [8]) of degree.. The Model and its Solutions The ewtonian equations of motion of the new many-body problem of goldfish type read as follows: with ( żn ż l z n = l=, l n z n z l [ {(z n ) m m= c m = ( ) m ) [ ω c m + (z n z l )] l=, l n ]} (c m c l ), (.a) l=, l m [ m ] (z sr ). (.b) s <s...<s m r=

3 Here and hereafter the symbol s <s...<s m signifies the sum from to over the m (integer) indices s j with j =,...,m and the restriction s < s <... < s m ; of course this sum vanishes if m >, consistently with otation.. The solutions z n z n (t) of this -body problem are provided consistently with the expressions (.b) by the zeros of the following t-dependent (monic) polynomial ψ (z;t) of degree in z: ψ (z;t) = z + m= [ cm (t) z m], (.) where the coefficients c m (t) are themselves the solutions of the system of ODEs c m = ω c m + (c m c l ). (.) l=, l m Because this is a well-known solvable model, the time-dependence of these quantities c m (t) can be obtained by solving an algebraic (in fact polynomial) problem, indeed the solution of the initialvalue problem of this dynamical system, (.), is provided by the following prescription (see for instance section 4.. in [4] or [, 9]): the quantities c m c m (t) are the eigenvalues of the (t-dependent) matrix with C(t) = C(0) cos(ω t) + Ċ(0) C(0) = diag[c m (0)], sin(ω t) ω, (.4a) (.4b) Ċ(0) = diag[ċ m (0)] + i [M(0), C(0)], (.4c) where of course (see (.b)) c m (0) = ( ) m ċ m (0) = ( ) m and in the right-hand side of (.4c) { m } [z sr (0)], (.4d) s <s...<s m r= s <s...<s m { m q= ż sq (0) m r=, r q [z sr (0)] }, (.4e) [M(0), C(0)] M(0) C(0) C(0) M(0) (.4f) with the matrix M(0) defined componentwise in terms of the initial data z n (0) as follows: M nm (0) = [z n (0) z m (0)], n m, M nn (0) = l=, l n M nl (0) = l=, l n [z n (0) z l (0)]. (.4g) ote that these formulas provide an explicit definition of the time-dependent coefficients c m (t) in terms of the initial data z n (0), ż n (0) of the -body problem of goldfish type characterized

4 Oksana Bihun and Francesco Calogero by the ewtonian equations of motion (.), via algebraic operations, amounting essentially to the solution of polynomial equations of degree ; and that the values z n (t) at time t of the particle coordinates z n are then provided by the zeros of the polynomial ψ (z;t), explicitly known (see (.)) in terms of its coefficients c m (t). It is thereby demonstrated that the -body problem of goldfish type characterized by the ewtonian equations of motion (.) is solvable (see Remark.). Remark.. Let us call attention to a (well known) tricky point associated with the solution as described above of the -body problem of goldfish type characterized by the ewtonian equations of motion (.). The identification of the eigenvalues of a given matrix is only unique up to permutations, and likewise the identification of the zeros of a polynomial is only unique up to permutations. Therefore the coordinates z n = z n (t) yielded by the solution detailed above are only identified up to permutations of their labels n. The (only) way to identify a specific coordinate say, the coordinate z (t) that corresponds to the initial data z (0), ż (0) is by following the (continuous) time evolution of the coordinate z (t) from its (assigned) initial value z (0) to its value z (t) at time t. In this manner one arrives at the uniquely defined value of the coordinate z (t) corresponding to the initial value z (0), which coincides of course with that uniquely yielded by the time evolution of the -body problem (.). An analogous phenomenology is also relevant to the solution of model (.a). Because of the way system (.) is constructed, its equilibria can be obtained by finding the zeros of the polynomials whose coefficients are equilibria of system (.). On the other hand, it is known that the zeros of the -th degree Hermite polynomial are equilibria of system (.) (see for instance []). This relationship allows to prove Diophantine properties of the polynomials whose coefficients are the zeros of Hermite polynomials. These properties, together with the study of the behavior of system (.) in the immediate vicinity of its equilibria, are discussed in a separate paper [7]. The findings reported above imply the possibility to display in completely explicit form the solution of the -body problem of goldfish type characterized by the ewtonian equations of motion (.) for =,, 4; but we doubt this would be very illuminating, and we therefore leave this task to the eager reader. We rather like to emphasize that these findings imply that, for arbitrary and arbitrary positive ω this -body problem is isochronous, all its solutions satisfying the periodicity property z n (t + T ) = z n (t), (.5) with T = π/ω (or, possibly, the period T might be replaced by an integer multiple of π/ω, generally small with respect to its possible maximal value!, which is of course a quite large number if is large: see [5]). We end this paper by displaying a few examples of this phenomenology. Example.. For = and ω =, taking into account that c = z z and c = z z, we see that system (.) reduces to z = ż ż z z z = ż ż z z + [ (z z ) (z z ) z [ z ] (z ) (z + z + z z ) (z ) (z + z + z z ), ]. (.6)

5 ( ).0 ( ) 0.5 ( ) ( ) ( ) ( ) -0.5 ( ) -.0 ( ) Fig.. Equilibria of system (.6) in the complex z-plane. Each equilibrium is represented by the two points ẑ ( j) and ẑ ( j) labeled by the index j =,,,4, see the display (.6 Equilibria). The assignment ω = is motivated by the fact that this quantity, ω, can be eliminated from system (.a) by a constant rescaling of the dependent variables z n (t) and of the independent variable t. System (.6) has 4 equilibrium configurations (ẑ ( j) j),ẑ(, whose approximate numerical values are given below: with ẑ ( j) (ẑ (),ẑ() (ẑ (),ẑ() (ẑ (),ẑ() ) = ( i, i), ) = ( i, i), ) = ( i, i), (ẑ (4),ẑ(4) ) = (.6575, ). ), j =,,,4, up to the exchange of ẑ( j) (.6 Equilibria) These equilibria can be obtained either by the substitution z n (t) = ẑ n, n =, into system (.6) and the subsequent solution of the resulting system of algebraic equations for ẑ,ẑ, or by finding the zeros of the polynomials whose coefficients are the equilibria of system (.) for =. We note that, in this case where = and ω =, system (.) has the four equilibria (, ), (, ), ( i, i ), and ( i, i ). Two of them are the zeros ± of the Hermite polynomial H (c) = 4c, which is consistent with the known fact that the zeros of the -th order Hermite polynomial are the equilibria of system (.). In Fig. each equilibrium of system (.6) is represented by the two points ẑ ( j) and ẑ ( j), labeled by ( j), where j =,,,4. Below we provide graphs of the real and imaginary parts of the components of the solution (z (t),z (t)) of system (.6) as functions of time, and some trajectories of the particles z and z in the complex z-plane. These graphs have been obtained by numerical integration of system (.6), using Mathematica 0. We employed the command DSolve with the automatic choice of the method and the accuracy. The program script is provided in the Appendix.

6 Oksana Bihun and Francesco Calogero Fig.. System (.6), initial conditions (.6a). Graphs of the real (upper curve) and imaginary (lower curve) parts of the coordinate z (t); period π Fig.. System (.6), initial conditions (.6a). Trajectory, in the complex z-plane, of z (t). The square respectively the dot indicate the positions of ẑ () respectively z (0), i.e. of a nearby equilibrium point respectively the initial value of z (t); period π Fig. 4. System (.6), initial conditions (.6a). Graphs of the real (lower curve) and the imaginary (upper curve) parts of the coordinate z (t); period π. (.6a) Initial conditions z (0) = ẑ () + 0.0, z (0) = ẑ () + 0.0, ż (0) = 0.0, ż (0) = 0.0. See Fig.,, 4, 5. (.6b) Initial conditions z (0) = ẑ () i, z (0) = ẑ () i, ż (0) = 0.0, ż (0) = 0.0. See Fig. 6, 7. (.6c) Initial conditions z (0) = ẑ () i, z (0) = ẑ () i, ż (0) = 0.0, ż (0) = 0.0. See Fig. 8, 9. (.6d) Initial conditions z (0) = ẑ () i, z (0) = ẑ () i, ż (0) = 0.005, ż (0) = See Fig. 0 and. (.6e) Initial conditions z (0) = ẑ (4) i, z (0) = ẑ (4) i, ż (0) = 0.0, ż (0) = 0.0. See Fig. 0 and.

7 Fig. 5. System (.6), initial conditions (.6a). Trajectory, in the complex z-plane, of z (t). The square respectively the dot indicate the positions of ẑ () respectively z (0), i.e of a nearby equilibrium point respectively the initial value of z (t); period π Fig. 6. System (.6), initial conditions (.6b). Trajectory, in the complex z-plane, of z (t).the square respectively the dot indicate the positions of ẑ () respectively z (0), i.e. of a nearby equilibrium point respectively the initial value of z (t); period π. Example.. For = and ω =, taking into account that c = z z z, c = z z + z z + z z, and c = z z z, we see that system (.) reduces to z = ż ż + ż ż z z z z [ z (z z )(z z ) F (z,z,z ) + z F (z,z,z ) + F (z,z,z )) ], z = ż ż + ż ż z z z z [ + z (z z )(z z ) F (z,z,z ) + z F (z,z,z ) + F (z,z,z )) ], z = ż ż ż ż z z z z [ z (z z )(z z ) F (z,z,z ) + z F (z,z,z ) + F (z,z,z )) ], (.8a)

8 Oksana Bihun and Francesco Calogero Fig. 7. System (.6), initial conditions (.6b). Trajectory, in the complex z-plane, of z (t).the square respectively the dot indicate the positions of ẑ () respectively z (0), i.e. of a nearby equilibrium point respectively the initial value of z (t); period π Fig. 8. System (.6), initial conditions (.6c). Trajectory, in the complex z-plane, of z (t). The square respectively the dot indicate the positions of ẑ () respectively z (0), i.e. of a nearby equilibrium point respectively the initial value of z (t); period π. where F (z,z,z ) = z + z + z (z + z + z z z z ) (z + z + z + z z + z z + z z ), F (z,z,z ) = z z z z z z + (z + z + z + z z + z z + z z ) + (z z + z z + z z + z z z ), F (z,z,z ) = z z z + (z + z + z z z z ) (z z + z z + z z + z z z ). (.8b)

9 Fig. 9. System (.6), initial conditions (.6c). Trajectory, in the complex z-plane, of z (t). The square respectively the dot indicate the positions of ẑ () respectively z (0), i.e. of a nearby equilibrium point respectively the initial value of z (t); period π Fig. 0. System (.6), initial conditions (.6d). Trajectory, in the complex z-plane, of z (t). The square respectively the dot indicate the positions of ẑ () respectively z (0), i.e. of a nearby equilibrium point respectively the initial value of z (t); period π Fig.. System (.6), initial conditions (.6d). Trajectory, in the complex z-plane, of z (t). The square respectively the dot indicate the positions of ẑ () respectively z (0) of a nearby equilibrium point respecttively the initial value of z (t); period π. We obtained equilibria of system (.8) as follows. First, we found the equilibria of system (.) for = ; they are given by (0,, ) and (0,i, i ), up to the permutations of the three

10 Oksana Bihun and Francesco Calogero Fig.. System (.6), initial conditions (.6e). Trajectory, in the complex z-plane, of z (t). The square respectively the dot indicate the positions of ẑ (4) respectively z (0), i.e. of a nearby equilibrium point respectively the initial value of z (t); period π Fig.. System (.6), initial conditions (.6e). Trajectory, in the complex z-plane, of z (t). The square respectively the dot indicate the positions of ẑ (4) respectively z (0), i.e. of a nearby equilibrium point respectively the initial value of z (t); period π. coordinates. Second, we found the zeros of the monic polynomials whose coefficients are the equilibria of system (.) for =. These zeros are equilibriumum solutions of system (.) because of how this system is constructed. Therefore, system (.8) has at least equilibrium configurations (ẑ ( j) j) j) j) j),ẑ(,ẑ( ), j =,,...,, up to the permutations of ẑ(, ẑ( and ẑ ( j), whose approximate

11 numerical values are given below: (ẑ (),ẑ(),ẑ() ) = ( i, i,.44048),,ẑ() ) = ( i, i, i),,ẑ() ) = ( i, i, i),,ẑ(4) ) = (0.709, i, i),,ẑ(5) ) = ( 0.785, i, i),,ẑ(6) ) = (0, i, i),,ẑ(7) ) = ( i, i,.666 i), (ẑ (8),ẑ(8),ẑ(8) ) = (0, i, i), (ẑ (9),ẑ(9),ẑ(9) ) = ( i,.666 i, i), (ẑ (0),ẑ (0),ẑ (0) ) = (0, i, i), (ẑ (),ẑ() (ẑ (),ẑ() (ẑ (4),ẑ(4) (ẑ (5),ẑ(5) (ẑ (6),ẑ(6) (ẑ (7),ẑ(7) (ẑ (),ẑ (),ẑ () ) = ( i, i, 0.785), (ẑ (),ẑ (),ẑ () ) = (0,.8778, ). (.8 Equilibria) It is possible that the system of algebraic equations characterizing the equilibria of (.8) has additional solutions besides those listed above. A direct attempt to solve this system of algebraic equations using Mathematica 0 was unsuccessful, and we did not deem the matter surficiently relevant to justify further investigations. In Fig. 4 each equilibrium of system (.8) is represented by the three points ẑ ( j) j), ẑ( and ẑ ( j), labeled by the index j, where j =,,...,. Below we provide some trajectories, in the complex plane, of the particles z, z and z whose evolution is described by system (.8) with the initial conditions z (0) = ẑ () + 0.,ż (0) = 0. z (0) = ẑ () + 0.,ż (0) = 0., z (0) = ẑ () + 0.,ż (0) = 0., (.8c) see Fig. 5, 6 and 7. These graphs have been obtained by numerical integration of system (.8) with the initial conditions (.8c), using Mathematica 0. We employed the command DSolve with the automatic choice of the method and the accuracy. The program script is provided in the Appendix.. Outlook The interest of the many-body model introduced and discussed above is demonstrated by its solvable and isochronous character as well as by the remarkable trajectories it features already in the simple = and = cases, as shown by the graphs reported above. It is also amusing to observe as the cognoscienti will have noted that this solvable model has been obtained by appropriately combining the two solvable equations see (.a) and (.) which are the prototypes of two, quite different, basic families of solvable many-body problems of ewtonian type. Moreover as reported in [7] the findings reported above provide the point of departure to obtain Diophantine properties of the zeros of each of the! (monic) polynomials the coefficients of which are the zeros of the Hermite polynomial H (z) of degree (these! polynomials correspond of course to

12 Oksana Bihun and Francesco Calogero ,8,0, Fig. 4. Equilibria of system (.8) in the complex z-plane. Each equilibrium is represented by the three points ẑ ( j) j), ẑ( and ẑ ( j) indexed and labeled by j =,,...,, see display (.8 Equilibria) Fig. 5. System (.8), initial conditions (.8c). Trajectory, in the complex z-plane, of z (t). The square respectively the dot indicate the positions of ẑ () respectively z (0), i.e. of a nearby equilibrium point respectively the initial value of z (t); period π. the! permutations of the zeros of H (z)). And a rather ample vista of further developments is provided by these findings. 4. Acknowledgements One of the authors (OB) would like to thank for the hospitality of the University of Rome La Sapienza during her visit in the summer 05.

13 Fig. 6. System (.8), initial conditions (.8c). Trajectory, in the complex z-plane, of z (t). The square respectively the dot indicate the positions of ẑ () respectively z (0), i.e. of a nearby equilibrium point respectively the initial value of z (t); period π Fig. 7. System (.8), initial conditions (.8c). Trajectory, in the complex z-plane, of z (t). The square respectively the dot indicate the positions of ẑ () respectively z (0), i.e. of a nearby equilibrium point respectively the initial value of z (t); period π. 5. Appendix. Scripts to Plot Approximate Solutions of Systems (.6) and (.8) 5.. A Mathematica Script to Plot Approximate Solutions of System (.6) (* This is a script to plot approximate solutions of system (.6) with initial conditions (.6a) through (.6e). Please feel free to modify and to try other initial conditions *) (* To find the equilibria of (.6), we first find the equilibria of system (.) with omega= *) omega=; LHSEqnEquilibriaC[cc_,cc_]=Simplify[-omega^ cc+/(cc-cc)^]; LHSEqnEquilibriaC[cc_,cc_]=Simplify[-omega^ cc+/(cc-cc)^]; EqilibriaCoatedC=Simplify[Solve[{LHSEqnEquilibriaC[zz,zz]==0, LHSEqnEquilibriaC[zz,zz]==0},{zz,zz}]];

14 Oksana Bihun and Francesco Calogero EquilibriaRepeatedC=Simplify[{zz,zz}/.EqilibriaCoatedC] CC=EquilibriaRepeatedC; (* P[[n,]] and P[[n,]] are the zeros of the polynomial z^+cc[[n,]] z +CC[[n,]], so (P[[n,]], P[[n,]]) are equilibria of (.6) *) P=Table[0,{n,,4}, {m,,}]; Do[{SolCoated=Solve[zz^+CC[[n,]] zz +CC[[n,]]==0,zz]; Sol=Simplify[zz/.SolCoated]; P[[n,]]=Simplify[Sol[[]]]; P[[n,]]=Simplify[Sol[[]]];},{n,,4}] P=[P] (* Equilibria of (.6), compare with (.7) in the paper *) MatrixForm[P] (* List of the initial conditions (.6a) through (.6e) (.6a) EquilibriumZ=P[[,]]; EquilibriumZ=P[[,]]; z0=equilibriumz+0.0;zp0=0.0; z0=equilibriumz+0.0;zp0=-0.0; (.6b) EquilibriumZ=P[[,]]; EquilibriumZ=P[[,]]; z0=equilibriumz i;zp0=-0.0; z0=equilibriumz I;zp0=0.0; (.6c) EquilibriumZ=P[[,]]; EquilibriumZ=P[[,]]; z0=equilibriumz i;zp0=-0.0; z0=equilibriumz I;zp0=0.0; (.6d) EquilibriumZ=P[[,]]; EquilibriumZ=P[[,]]; z0=equilibriumz i;zp0=-0.005; z0=equilibriumz I;zp0=0.005; (.6e) EquilibriumZ=P[[4,]]; EquilibriumZ=P[[4,]]; z0=equilibriumz i;zp0=-0.0; z0=equilibriumz I;zp0=0.0; *) (* Choose the initial conditions for system (.6), either from the list above or your own*) EquilibriumZ=P[[4,]]; EquilibriumZ=P[[4,]]; z0=equilibriumz I;zp0=-0.0; z0=equilibriumz+0.0;zp0=0.0; (* Solve system (.6) with the last initial conditions, numerically*) s=dsolve[{z [t]==( z [t] z [t])/(z[t]-z[t])-/(z[t]-z[t]) (omega^ z[t]^-((z[t]-))/(z[t]+z[t]+z[t] z[t])^), z [t]==-(( z [t] z [t])/(z[t]-z[t]))+/(z[t]-z[t]) (omega^ z[t]^-((z[t]-))/(z[t]+z[t]+z[t] z[t])^),z[0]==z0, z [0]==zp0,z[0]==z0,z [0]==zp0},{z,z},{t,40}]; (* Plot the real and the imaginary parts of z[t], the first component of the solution of system (.6) with the assigned initial conditions *) Plot[{{Re[Evaluate[{z[t]}/.s]][[]][[]],PlotStyle->{Black,Thick}}, Im[Evaluate[{z[t]}/.s]][[]][[]]},{t,0,0},PlotRange->All, Axes->True, AxesStyle->Directive[Bold,], PlotStyle->{Black,Thick}] (* From the last plot, guess the period of z[t], verify if the guess is

15 correct by plotting the real and the imaginary parts of z[t+period Guess]-z[t] *) PeriodGuess=[ Pi];Plot[{{Re[Evaluate[{z[t+PeriodGuess]}/.s]][[]][[]] -Re[Evaluate[{z[t]}/.s]][[]][[]],PlotStyle->{Black,Thick}}, Im[Evaluate[{z[t+PeriodGuess]}/.s]][[]][[]] -Im[Evaluate[{z[t]}/.s]][[]][[]]},{t,0,0}, PlotRange->All, Axes->True, AxesStyle->Directive[Bold,], PlotStyle->{Black,Thick}] (* Plot the real and the imaginary parts of z[t], the first component of the solution of system (.6) with the assigned initial conditions *) Plot[{{Re[Evaluate[{z[t]}/.s]][[]][[]], PlotStyle->{Black,Thick}}, {Im[Evaluate[{z[t]}/.s]][[]][[]], PlotStyle->{Black,Thick}}},{t,0,0}, PlotRange->All, Axes->True, AxesStyle->Directive[Bold,],PlotStyle->{Black,Thick}] (* From the last plot, guess the period of z[t], verify if the guess is correct by plotting the real and the imaginary parts of z[t+period Guess]-z[t] *) PeriodGuess=[ Pi]; Plot[{{Re[Evaluate[{z[t+PeriodGuess]}/.s]][[]][[]] -Re[Evaluate[{z[t]}/.s]][[]][[]],PlotStyle->{Black,Thick}}, Im[Evaluate[{z[t+PeriodGuess]}/.s]][[]][[]] -Im[Evaluate[{z[t]}/.s]][[]][[]]}, {t,0,0},plotrange->all, Axes->True, AxesStyle->Directive[Bold,], PlotStyle->{Black,Thick}] (* Plot the trajectory, in the complex z-plane, of the solution z[t] of system (.6) with the assigned initial conditions *) Graph=ParametricPlot[{Re[Evaluate[{z[t]}/.s]][[]][[]], Im[Evaluate[{z[t]}/.s]][[]][[]]},{t,0,0},PlotRange->All, Axes->True, AxesStyle->Directive[Bold,], PlotStyle->{Black,Thick}, AspectRatio->/.]; Show[Graph,Graphics[ {PointSize[Large], Point[{Re[EquilibriumZ],Im[EquilibriumZ]}], Point[{Re[z0],Im[z0]}] }]] (* Plot the trajectory, in the complex z-plane, of the solution z[t] of system (.6) with the assigned initial conditions *) Graph=ParametricPlot[{Re[Evaluate[{z[t]}/.s]][[]][[]], Im[Evaluate[{z[t]}/.s]][[]][[]]},{t,0,0},PlotRange->All, Axes->True, AxesStyle->Directive[Bold,],PlotStyle->{Black,Thick}, AspectRatio->/.5]; Show[Graph,Graphics[

16 Oksana Bihun and Francesco Calogero {PointSize[Large],Point[{Re[EquilibriumZ],Im[EquilibriumZ]}], Point[{Re[z0],Im[z0]}] }]] 5.. A Mathematica Script to Plot Approximate Solutions of System (.8) (* This is a script to plot approximate solutions of system (.8) with initial conditions (.0). Please feel free to modify and to try other initial conditions *) c=(-z-z-z); c=z z+z z+z z; c=-z z z; (* To find (some of the) equilibria of system (.8), we first find the equilibria of system (.) for = and omega= and then find the zeros of the monic polynomials whose coefficiants are equilbria of (.)*) FF[z_,z_,z_]=Simplify[-c+/(c-c)^+/(c-c)^]; FF[z_,z_,z_]=Simplify[-c+/(c-c)^+/(c-c)^]; FF[z_,z_,z_]=Simplify[-c+/(c-c)^+/(c-c)^]; omega=; LHSEqnEquilibriaC[cc_,cc_,cc_]=Simplify[-cc+/(cc-cc)^+/(cc-cc)^]; LHSEqnEquilibriaC[cc_,cc_,cc_]=Simplify[-cc+/(cc-cc)^+/(cc-cc)^]; LHSEqnEquilibriaC[cc_,cc_,cc_]=Simplify[-cc+/(cc-cc)^+/(cc-cc)^]; EqilibriaCoatedC=Simplify[Solve[{LHSEqnEquilibriaC[zz,zz, zz]==0, LHSEqnEquilibriaC[zz,zz,zz]==0, LHSEqnEquilibriaC[zz,zz,zz]==0},{zz,zz,zz}]]; EquilibriaRepeatedC=Simplify[{zz,zz,zz}/.EqilibriaCoatedC] CC=EquilibriaRepeatedC; (* P[[n,]], P[[n,]] and P[[n,]] are the zeros of the polynomial z^+cc[[n,]] z^ +CC[[n,]] z+cc[[n,]], so (P[[n,]], P[[n,]], P[[n,]] ) are equilibria of (.8) *) P=Table[0,{n,,}, {m,,}]; Do[{SolCoated=Solve[zz^+CC[[n,]] zz^ +CC[[n,]]zz+CC[[n,]]==0,zz]; Sol=Simplify[zz/.SolCoated]; P[[n,]]=Simplify[Sol[[]]]; P[[n,]]=Simplify[Sol[[]]]; P[[n,]]=Simplify[Sol[[]]];},{n,,}] (* (Some) equilibria of (.8), compare with the list (.9) in this paper *) P=[P]; MatrixForm[P] (* Initial Conditions (.0)*) EquilibriumZ=P[[,]];

17 EquilibriumZ=P[[,]]; EquilibriumZ=P[[,]]; z0=equilibriumz+0.; zp0=0.; z0=equilibriumz+0.;zp0=0.; z0=equilibriumz+0.;zp0=0.; (* Solve system (.8) with the last initial conditions, numerically*) LHSEquilibriaEqn[z_,z_,z_]=/((z-z)(z-z)) Simplify[ (z^ff[z,z,z]+z FF[z,z,z]+FF[z,z,z])]; LHSEquilibriaEqn[z_,z_,z_]=/((z-z)(z-z)) Simplify[ (z^ff[z,z,z]+z FF[z,z,z]+FF[z,z,z])]; LHSEquilibriaEqn[z_,z_,z_]=/((z-z)(z-z)) Simplify[ (z^ff[z,z,z]+z FF[z,z,z]+FF[z,z,z])]; s=dsolve[{z [t]==( z [t] z [t])/(z[t]-z[t]) +( z [t] z [t])/(z[t]-z[t]) - LHSEquilibriaEqn[z[t],z[t],z[t]],z [t] ==( z [t] z [t])/(z[t]-z[t]) +( z [t] z [t])/(z[t]-z[t])- LHSEquilibriaEqn[z[t],z[t],z[t]], z [t]==( z [t] z [t])/(z[t]-z[t])+( z [t] z [t])/(z[t]-z[t]) - LHSEquilibriaEqn[z[t],z[t],z[t]],z[0]==z0, z [0]==zp0,z[0]==z0,z [0]==zp0,z[0]==z0, z [0]==zp0},{z,z,z},{t,0},AccuracyGoal->Automatic, Method->Automatic]; (* Plot the real and the imaginary parts of z[t], the first component of the solution of system (.6) with the assigned initial conditions *) Plot[{{Re[Evaluate[{z[t]}/.s]][[]][[]],PlotStyle->{Black,Thick}}, Im[Evaluate[{z[t]}/.s]][[]][[]]},{t,0,0},PlotRange->All, Axes->True, AxesStyle->Directive[Bold,], PlotStyle->{Black,Thick}] (* From the last plot, guess the period of z[t], verify if the guess is correct by plotting the real and the imaginary parts of z[t+period Guess]-z[t] *) PeriodGuess=[ Pi]; Plot[{{Re[Evaluate[{z[t+PeriodGuess]}/.s]][[]][[]] -Re[Evaluate[{z[t]}/.s]][[]][[]],PlotStyle->{Black,Thick}}, Im[Evaluate[{z[t+PeriodGuess]}/.s]][[]][[]] -Im[Evaluate[{z[t]}/.s]][[]][[]]},{t,0,0},PlotRange->All, Axes->True, AxesStyle->Directive[Bold,], PlotStyle->{Black,Thick}] (* Plot the real and the imaginary parts of z[t], the second component of the solution of system (.6) with the assigned i nitial conditions *) Plot[{{Re[Evaluate[{z[t]}/.s]][[]][[]],PlotStyle->{Black,Thick}}, {Im[Evaluate[{z[t]}/.s]][[]][[]],PlotStyle->{Black,Thick}}},{t,0,0}, PlotRange->All, Axes->True, AxesStyle->Directive[Bold,], PlotStyle->{Black,Thick}] (* From the last plot, guess the period of z[t], verify if the

18 Oksana Bihun and Francesco Calogero guess is correct by plotting the real and the imaginary parts of z[t+period Guess]-z[t] *) PeriodGuess=[ Pi]; Plot[{{Re[Evaluate[{z[t+PeriodGuess]}/.s]][[]][[]] -Re[Evaluate[{z[t]}/.s]][[]][[]],PlotStyle->{Black,Thick}}, Im[Evaluate[{z[t+PeriodGuess]}/.s]][[]][[]] -Im[Evaluate[{z[t]}/.s]][[]][[]]}, {t,0,0},plotrange->all, Axes->True, AxesStyle->Directive[Bold,], PlotStyle->{Black,Thick}] (* Plot the real and the imaginary parts of z[t], the third component of the solution of system (.6) with the assigned initial conditions *) Plot[{{Re[Evaluate[{z[t]}/.s]][[]][[]],PlotStyle->{Black,Thick}}, {Im[Evaluate[{z[t]}/.s]][[]][[]],PlotStyle->{Black,Thick}}},{t,0,0}, PlotRange->All, Axes->True, AxesStyle->Directive[Bold,], PlotStyle->{Black,Thick}] (* From the last plot, guess the period of z[t], verify if the guess is correct by plotting the real and the imaginary parts of z[t+period Guess]-z[t] *) PeriodGuess=[ Pi]; Plot[{{Re[Evaluate[{z[t+PeriodGuess]}/.s]][[]][[]] -Re[Evaluate[{z[t]}/.s]][[]][[]],PlotStyle->{Black,Thick}}, Im[Evaluate[{z[t+PeriodGuess]}/.s]][[]][[]] -Im[Evaluate[{z[t]}/.s]][[]][[]]}, {t,0,0},plotrange->all, Axes->True, AxesStyle->Directive[Bold,], PlotStyle->{Black,Thick}] (* Plot the trajectory, in the complex z-plane, of the solution z[t] of system (.8) with the assigned initial conditions *) Graph=ParametricPlot[{Re[Evaluate[{z[t]}/.s]][[]][[]], Im[Evaluate[{z[t]}/.s]][[]][[]]},{t,0,0},PlotRange->All, Axes->True, AxesStyle->Directive[Bold,],PlotStyle->{Black,Thick}, AspectRatio->/.]; Show[Graph,Graphics[{PointSize[Large], Point[{Re[EquilibriumZ],Im[EquilibriumZ]}], Point[{Re[z0],Im[z0]}] }]] (* Plot the trajectory, in the complex z-plane, of the solution z[t] of system (.8) with the assigned initial conditions *) Graph=ParametricPlot[{Re[Evaluate[{z[t]}/.s]][[]][[]], Im[Evaluate[{z[t]}/.s]][[]][[]]},{t,0,0},PlotRange->All, Axes->True, AxesStyle->Directive[Bold,], PlotStyle->{Black,Thick},AspectRatio->/.5]; Show[Graph,Graphics[{PointSize[Large], Point[{Re[EquilibriumZ],Im[EquilibriumZ]}], Point[{Re[z0],Im[z0]}] }]] (* Plot the trajectory, in the complex z-plane, of the solution z[t] of system (.8) with the assigned initial conditions *)

19 Graph4=ParametricPlot[{Re[Evaluate[{z[t]}/.s]][[]][[]], Im[Evaluate[{z[t]}/.s]][[]][[]]},{t,0,0},PlotRange->All, Axes->True, AxesStyle->Directive[Bold,], PlotStyle->{Black,Thick},AspectRatio->/.5]; Show[Graph4,Graphics[{PointSize[Large], Point[{Re[EquilibriumZ],Im[EquilibriumZ]}], Point[{Re[z0],Im[z0]}] }]] References [] F. Calogero, The neatest many-body problem amenable to exact treatments (a goldfish?), Physica D 5-5 (00) [] F. Calogero, Motion of poles and zeros of special solutions of nonlinear and linear partial differential equations, and related solvable many body problems, uovo Cimento 4B (978) [] F. Calogero, Classical many-body problems amenable to exact treatments (Lecture otes in Physics Monographs m66, Springer, Heidelberg, 00). [4] F. Calogero, Isochronous systems (Oxford University Press, Oxford, 008; marginally updated paperback edition 0). [5] D. Gómez-Ullate and M. Sommacal, Periods of the goldfish many-body problem, J. onlinear Math. Phys., Suppl. (005) 5 6. [6] F. Calogero, ew solvable variants of the goldfish many-body problem, Studies Appl. Math., in press. [7] O. Bihun and F. Calogero, Diophantine properties of the zeros of (monic) polynomials the coefficients of which are the zeros of Hermite polynomials, arxiv: [math-ph], (submitted for publication). [8] A. Erdélyi (editor), Higher transcendental functions, vol. II (McGraw-Hill, ew York, 95). [9] M. A. Olshanetsky and A. M. Perelomov, Classical integrable finite-dimensional systems related to Lie algebras, Phys. Rep. 7 (98) 400.