Vortical and Self-similar Flows of 2D Compressible Euler Equations Manwai Yuen∗ Department of Mathematics and Information Technology, 3 The Hong Kong Institute of Education, 1 0 10 Po Ling Road, Tai Po, New Territories, Hong Kong 2 n Revised 19-Jan-2013 a J 8 1 Abstract ] h Thispaperpresentsthevorticalandself-similar solutions for2D compressibleEulerequa- p - tions using the separation method. These solutions complement Makino’s solutions in radial h t symmetry without rotation. The rotational solutions provide new information that furthers a m our understanding of ocean vortices and reference examples for numerical methods. In ad- [ dition, the corresponding blowup, time-periodic or global existence conditions are classified 1 through an analysis of the new Emden equation. A conjecture regarding rotational solutions v 9 in 3D is also made. 7 4 4 MSC: 76U05, 35C05, 35C06, 35B10, 35R35 . 1 0 3 KeyWords: Vortex,CompressibleEulerEquations,VorticalSolution,Self-similarSolution, 1 : Time-periodic Solution v i X r 1 Introduction a In fluid mechanics, the N-dimensional isentropic compressible Euler and Navier-Stokes equations are expressed as follows: ρ +∇·(ρ~u)=0 t (1) ρ[~ut+(~u·∇)~u]+K∇ργ =0, where ρ = ρ(t,~x) denotes the density of the fluid, ~u = ~u(t,~x) = (u1,u2,··· ,uN) ∈ RN is the velocity, ~x=(x ,x ,··· ,x )∈RN is the space variable, and K >0, γ >1 are constants. 1 2 N ∗E-mailaddress: [email protected] 1 2 M.W.Yuen The Euler equations have been studied in great detail by numerous scholars because of their significance in a variety of physical fields, such as fluids, plasmas, condensed matter, astrophysics, oceanography, and atmospheric dynamics. These equations are also important in physics and are widely used in different areas of study. For instance, the Euler system is the basic model for shallow water flows [3]. It also provides a good model of the superfluids produced by the Bose-Einstein condensates in the dilute gases of alkali metal atoms, in which identical gases do not interact at very low temperatures [4]. At the microscopic level, fluids or gases are formed by many tiny discrete molecules or particles that collide with one another. As the cost of directly calculating the particle-to-particle or molecule-to-molecule evolution of fluids on a large scale is expensive, approximation methods are needed to considerably simplify the process. Therefore, at the macroscopic scale, the continuum assumption, which considers fluids as continuous, is applied tothemodeling. Here,theEulerequationsprovideagooddescriptionofthefluidsatthestatistical limit of a large number of small ideal molecules or particles by ignoring the less influential effects, such as the self-gravitational forces and relativistic effect [1]. For a mathematical introduction of the Euler equations, readers are referred to [8] and [2]. The constructionof analyticalor exact solutions is animportant area in mathematical physics andapplied mathematics,as it canfurther classify their nonlinearphenomena. Fornon-rotational flows,MakinofirstobtainedtheradialsymmetrysolutionsfortheEulerequations(1)inRN in1993 [9]. A number ofspecial solutionsfor these equations[6], [7], [12], [13], and [14]were subsequently obtained. For rotational flows, Zhang and Zheng [15] constructed explicitly rotational solutions for the Euler equations with γ =2 in 1997: r2 1 1 ρ= , u = (x+y), u = (x−y), (2) 8Kt2 1 2t 2 2t where x=rcosθ and y =rsinθ. Very recently, Kwong and Yuen [5] constructed a family of rotational solutions for the Euler- Poisson equations ρ +∇·(ρ~u)=0 t ρ(~u +(~u·∇)~u)+K∇ρ=−ρ∇Φ (3) t ∆Φ(t,~x)=2πρ, with N =2 and γ =1: · ρ(t,~x)= 1 ef(r/a(t)), ~u(t,~x)=a(t)(x,y)+ ξ (−y,x) a(t)2 a(t) a(t)2 a¨(t)= −λ + ξ2 , a(0)=a >0, a˙(0)=a (4) f··(s)+1f·a((st))+2πa(etf)3(s)=2λ, f(00)=α, f·(0)=10, s K K with arbitrary constants ξ 6=0, a0, a1 and α. The rotational case complements Yuen’s solutions without rotation (ξ = 0) [11]. In this paper, Vortical Flows of 2D Euler Equations 3 basedon the foregoingdevelopment, we provide the correspondingvorticalflows for 2D compress- ible Euler equations (1) with γ >1 in the following result. Theorem 1 For γ > 1, there exists a family of vortical flows in radial symmetry for the com- pressible Euler equations (1) in 2D, max (−λ(γ−1)s+α)γ−11, 0 ρ(t,~x)= (cid:18) 2Kγ (cid:19) a(t)2 ~u(t,~x)=aa·((tt))(x,y)+a(ξt)2(−y,x) (5) a¨(t)− ξ2 − λ =0, a(0)=a >0, a˙(0)=a , a(t)3 a(t)2γ−1 0 1 with the self-similar variable s= x2+y2 and arbitrary constants ξ 6=0, a , a , and α. a(t)2 0 1 Remark 2 This result complements Makino’s solutions in radial symmetry without rotation (ξ = 0). The vortical solutions (5) provide new information that may further our understanding of oceans vortices and reference examples for numerical methods in computational physics. Remark 3 Zhang and Zheng’s solution (1) for γ =2 is a special case in our solutions (5). 2 2D Vortical and Self-similar Flows Here, we provide a lemma for the vortical flows in 2D for the mass equation (1) . This lemma 1 originates in Kwong and Yuen’s paper [5], which constructs periodic and spiral solutions for 2D Euler-Poissonequations (3) with γ =1. Lemma 4 ([5]) For the equation of the conservation of mass, ρ +∇·(ρ~u)=0, (6) t there exist the following solutions: f( r ) a˙(t) G(t,r) a(t) ρ(t,~x)=ρ(t,r)= , ~u(t,~x)= (x,y)+ (−y,x) (7) a(t)2 a(t) r with the radial r = x2+y2 and arbitrary functions f ≥0; G and a(t)>0∈C1. Proof. The functiopnal structure F(t,r) G(t,r) ρ(t,~x)=ρ(t,r), ~u(t,~x)= (x,y)+ (−y,x) (8) r r with an arbitrary C1 function F(t,r), can be plugged intothe mass equation (6) toverify the result: ρ +∇·(ρ~u) (9) t ∂ Fx Gy ∂ Fy Gx =ρ + ρ −ρ + ρ +ρ (10) t ∂x r r ∂y r r (cid:20) (cid:21) (cid:20) (cid:21) 4 M.W.Yuen ∂ Fx ∂ Fx ∂ Gy ∂ Gy =ρ + ρ +ρ −( ρ) −ρ( ) t ∂x r ∂x r ∂x r ∂x r (cid:18) (cid:19) ∂ Fy ∂ Fy ∂ Gx ∂ Gx + ρ +ρ +( ρ) +ρ( ) (11) ∂y r ∂y r ∂y r ∂y r (cid:18) (cid:19) xFx x x F x =ρ +ρ +ρ F +ρ −ρFx t rr r rr r r r3 xGy xy(cid:16) (cid:17) x yFy y y −ρ −ρG +ρGy +ρ +ρ F rr r rr r r3 rr r rr r F y yGx y x (cid:16) y (cid:17) +ρ −ρFy +ρ +ρ G −ρGx (12) r r3 rr r rr r r3 (cid:16) (cid:17) xFx x x F x =ρ +ρ +ρ F +ρ −ρFx t rr r rr r r r3 yFy y(cid:16) y (cid:17) F y +ρ +ρ F +ρ −ρFy (13) rr r rr r r r3 (cid:16) (cid:17) 1 =ρ +ρ F +ρF +ρF . (14) t r r r Then, the self-similar structure is taken for the density function, f( r ) a(t) ρ(t,~x)=ρ(t,r)= (15) a(t)2 and F(t,r)= a˙(t)r for velocity ~u to balance equation (14) [11]: a(t) ∂ f( r ) ∂ f( r )a˙(t)r f( r )a˙(t) f( r )a˙(t) a(t) a(t) a(t) a(t) = + + + (16) ∂t a(t)2 ∂r a(t)2 a(t) a(t)2 a(t) a(t)2 a(t) · · · −2a(t)f(r/a(t)) a(t)rf(r/a(t)) = − a(t)3 a(t)4 · · · · f(r/a(t))a(t)r f(r/a(t))a(t) f(r/a(t))a(t) + + + (17) a(t)3 a(t) a(t)2 a(t) a(t)2 a(t) =0. (18) The proof is completed. The computational proof for Theorem 1 is as follows. Proof of Theorem 1. The procedure for the proof for vortical fluids is similar to that for non-vorticalfluids [9], [14]. It is clear that the following function · f(s) a(t) ξ ρ(t,~x)= , ~u(t,~x)= (x,y)+ (−y,x), (19) a(t)2 a(t) a(t)2 with an arbitrary C1 function f ≥ 0 of the self-similar variable s = x2+y2 = r2 and a unde- a(t)2 a(t)2 termined C2 time-function a(t) > 0, satisfies Lemma 4 for the mass equation (1) . For the first 1 momentum equation (1) , we obtain: 21 ∂ ∂u ∂u ∂ =ρ u +u 1 +u 1 +Kγργ−1 ρ (20) 1 1 2 ∂t ∂x ∂y ∂x (cid:18) (cid:19) Vortical Flows of 2D Euler Equations 5 ∂ ∂u ∂u f(s)γ−1 1 ∂ 1 1 =ρ u +u +u +Kγ f(s) (21) ∂t 1 1 ∂x 2 ∂y a(t)2(γ−1)a(t)2∂x (cid:18) (cid:19) ∂ ∂u ∂u f(s)γ−2 2x =ρ u +u 1 +u 1 +Kγ f˙(s) (22) ∂t 1 1 ∂x 2 ∂y a(t)2(γ−1)a(t)2 (cid:20) (cid:21) ∂ a˙(t)x− ξ y + a˙(t)x− ξ y ∂ a˙(t)x− ξ y ∂t a(t) a(t)2 a(t) a(t)2 ∂x a(t) a(t)2 =ρ (23) + (cid:16) ξ x+ a˙(t)y (cid:17)∂ (cid:16)a˙(t)x− ξ y(cid:17)+K(cid:16)γf(s)γ−22xf˙(s(cid:17)) a(t)2 a(t) ∂y a(t) a(t)2 a(t)2γ (cid:16) (cid:17) (cid:16) (cid:17) a¨(t) − a˙(t)2 x+ 2ξa˙(t)y+ a˙(t)x− ξ y a˙(t) a(t) a(t)2 a(t)3 a(t) a(t)2 a(t) =ρ (24) (cid:16)+ ξ x+(cid:17)a˙(t)y − ξ (cid:16)+Kγf(s)γ−22x(cid:17)f˙(s) a(t)2 a(t) a(t)2 a(t)2γ (cid:16) a¨(t) ξ2(cid:17)(cid:16) (cid:17) f(s)γ−2 =ρ − x+Kγ 2xf˙(s) (25) a(t) a(t)4 a(t)2γ (cid:20)(cid:18) (cid:19) (cid:21) ρx = λ+2Kγf(s)γ−2f˙(s) , (26) a(t)2γ h i with the Emden equation a¨(t)− ξ2 = λ a(t)3 a(t)2γ−1 (27) a(0)=a >0, a˙(0)=a , 0 1 with arbitrary constants ξ and λ>1. To ensure that equation (26) is zero, the following ordinary differential equation is required: λ+2Kγf(s)γ−2f˙(s)=0 (28) f(0)=α>0, f˙(0)=0. The function f(s) in the foregoing ordinary differential equation can be solved exactly: 1 λ(γ−1) γ−1 f(s)= − s+α . (29) 2Kγ (cid:18) (cid:19) For the second momentum equation (1) , we also have: 22 ∂ ∂u ∂u ∂ =ρ u +u 2 +u 2 +Kγργ−1 ρ (30) 2 1 2 ∂t ∂x ∂y ∂y (cid:18) (cid:19) ∂ ∂u ∂u f(s)γ−2 2y =ρ u +u 2 +u 2 +Kγ f˙(s) (31) ∂t 2 1 ∂x 2 ∂y a(t)2(γ−1)a(t)2 (cid:20) (cid:21) ∂ ξ x+ a˙(t)y + a˙(t)x− ξ y ∂ ξ x+ a˙(t)y ∂t a(t)2 a(t) a(t) a(t)2 ∂x a(t)2 a(t) =ρ (32) + (cid:16) ξ x+ a˙(t)y(cid:17)∂ (cid:16) ξ x+ a˙(t)y(cid:17)+K(cid:16)γf(s)γ−22yf˙(s(cid:17)) a(t)2 a(t) ∂y a(t)2 a(t) a(t)2γ (cid:16) (cid:17) (cid:16) (cid:17) −2ξa˙(t)x+ a¨(t) − a˙(t)2 y+ a˙(t)x− ξ y ξ a(t)3 a(t) a(t)2 a(t) a(t)2 a(t)2 =ρ (33) + ξ (cid:16)x+ a˙(t)y a˙(cid:17)(t) +K(cid:16) γf(s)γ−22yf˙((cid:17)s) a(t)2 a(t) a(t) a(t)2γ (cid:16)a¨(t) ξ2 (cid:17) f(s)γ−2 =ρ − y+Kγ 2yf˙(s) (34) a(t) a(t)4 a(t)2γ (cid:20)(cid:18) (cid:19) (cid:21) ρy = λ+2Kγf(s)γ−2f˙(s) (35) a(t)2γ h i 6 M.W.Yuen =0. (36) with the Emden equation (27) and function (29). Therefore, for ξ 6= 0, we have the vortical and self-similar flows for the compressible Euler equations (1) in 2D [9]. To ensure the non-negative density function, we require that 1 max −λ(γ−1)s+α γ−1 , 0 2Kγ ρ(t,~x)= (cid:18)(cid:16) (cid:17) (cid:19). (37) a(t)2 This completes the proof. Lemma 5 For the Emden equation, a¨(t)− ξ2 − λ =0 a(t)3 a(t)2γ−1 (38) a(0)=a >0,a˙(0)=a , 0 1 with arbitrary constants ξ 6=0, λ and γ >1, the total energy is a˙(t)2 ξ2 λ E(t):= + + . (39) 2 2a(t)2 (2γ−2)a(t)2γ−2 We have the following. (1) For 1<γ <2, if E(0)<0, the solution is time-periodic; otherwise, the solution is global. (2) For γ =2, (2a) with ξ2 ≥−λ, the solution is global; (2b) with ξ2 <−λ, the solution blows up at a finite time if −λ−ξ2 a < ; (40) 1 a p 0 otherwise, the solution is global. (3) For γ >2, (3a) with λ≥0, the solution is global; 1 (3b) with λ<0 and a constant a = −λ 2γ−4, Max ξ2 (3bI) and a ≥a , (cid:16) (cid:17) 0 Max if E(0) ≤ F (a ) or E(0) > F (a ) with a ≥ 0, the solution is global; otherwise, the pot Max pot Max 1 solution blows up at a finite time. (3bII) and a <a , 0 Max if E(0)≥F (a ) with a >0, thesolution is global; otherwise, the solution blows upat a finite pot Max 1 time. Proof. For equation (38), we multiply a˙(t) and then integrate it, as follows: a˙(t)2 ξ2 λ + + =E(t), (41) 2 2a(t)2 (2γ−2)a(t)2γ−2 Vortical Flows of 2D Euler Equations 7 with a constant E(0)= a21 + ξ2 + λ . 2 2a20 (2γ−2)a20γ−2 We define the kinetic energy as: a˙(t)2 F := , (42) kin 2 and the potential energy as: ξ2 λ F = + . (43) pot 2a(t)2 (2γ−2)a(t)2γ−2 The total energy is conserved thusly: dE(t) d = (F +F )=0. (44) kin pot dt dt The classical energy method for second-order autonomous ordinary differential equations (which readerscanrefertopages793–798in[10]),canbeappliedtoanalyzethecorrespondingqualitative properties of the Emden equation (38). (1)For1<γ <2,thereexistsauniqueglobalminimumforthepotentialfunction(43)fora(t)>0, and lim F (a(t))=+∞ and lim F (a(t))=0. If pot pot a(t)→0+ a(t)→+∞ a2 ξ2 λ E(0)= 1 + + <0, (45) 2 2a2 (2γ−2)a2γ−2 0 0 the solution is time-periodic; otherwise, it is global. (2)Forγ =2,theordinarydifferentialequation(38)degeneratesintotheclassicalEmdenequation ξ2+λ a¨(t)= . (46) a(t)3 (2a) For ξ2 ≥−λ, the solution is global. (2b) For ξ2 < −λ, the potential function (43) is strictly increasing with lim F (a(t)) = 0. pot a(t)→+∞ The solution blows up at a finite time, if −λ−ξ2 a < ; (47) 1 a p 0 otherwise, the solution is global. (3) For γ >2, (3a)withλ≥0,thepotentialfunction(43)isadecreasingfunctionofa(t)>0. Thus,thesolution is global. (3b) With λ<0,the potentialfunction (43) achievesa unique globalmaximum, F (a ), with pot Max 1 a = −λ 2γ−4 for a(t)>0. Max ξ2 (3bI) An(cid:16)d a (cid:17)≥a , 0 Max if E(0) ≤ F (a ) or E(0) > F (a ) with a ≥ 0, the solution is global; otherwise, the pot Max pot Max 1 solution blows up at a finite time. (3bII) and a <a , 0 Max 8 M.W.Yuen ifE(0)≥F (a )witha >0,thesolutionisglobal;otherwisethesolutionblowsupatafinite pot Max 1 time. The proof is now complete. The foregoing lemma makes it easy to determine the blowup or global existence of the corre- sponding solutions (5) for the compressible Euler equations (1) in 2D. Corollary 6 The total energy is defined a˙(t)2 ξ2 λ E(t):= + + , (48) 2 2a(t)2 (2γ−2)a(t)2γ−2 for the Emden equation (5) . 3 (1) For 1<γ <2, if E(0)<0, solution (5) is time-periodic; otherwise, solution (5) is global. (2) For γ =2, (2aI) with ξ2 >−λ or ξ2 =−λ and a ≥0, solution (5) is global; 1 (2aII) with ξ2 =−λ and a <0, solution (5) blows up at T =−a /a . 1 1 0 (2b) with ξ2 <−λ, solution (5) blows up at a finite time if −λ−ξ2 a < ; (49) 1 a p 0 otherwise, solution (5) is global. (3) For γ >2, (3a) with λ≥0, solution (5) is global; 1 (3b) with λ<0 and a constant a = −λ 2γ−4, Max ξ2 (3bI) and a ≥a , (cid:16) (cid:17) 0 Max if E(0)≤F (a ) or E(0)>F (a ) with a ≥0, solution (5) is global; otherwise, solution pot Max pot Max 1 (5) blows up at a finite time. (3bII) and a <a , 0 Max if E(0)≥F (a ) with a >0, solution (5) is global; otherwise, solution (5) blows up at a finite pot Max 1 time. 3 Conclusion and Discussion This paper provides a class of self-similar vortical flows for the 2D compressible Euler equations. The resultpresentedhereincomplements Makino’ssolutions in radialsymmetry without rotation. In addition, the corresponding blowup or global existence conditions are classified by analyzing the new Emden equation (5) . Solution (5) is also the solution of the compressible Navier-Stokes 3 equations in 2D: ρ +∇·(ρ~u) =0 t (50) ρ[~ut+(~u·∇)~u]+K∇ργ =µ∆~u, Vortical Flows of 2D Euler Equations 9 with a positive constant µ. Based on the existence of the rotational and self-similar solutions (5) in 2D, it is natural to conjecture that there exists a class of rotationalsolutions for the compressible Euler equations (1) in 3D, that complements those in radial symmetry, [9], [14]: ρ= f(s) 3 kΠ=1ak (51) ui = aa˙ii (xi−d∗i)+d˙∗i, for i=1,2,3, where 1 ξ(γ−1) γ−1 f(s)=max − s+α , 0 , (52) 2Kγ (cid:18) (cid:19) ! with 3 x −d∗ d∗ =d∗ +td∗ , s= ( k k)2. (53) i i0 i1 a k k=1 X Here, ξ,d∗ ,d∗ , α≥0 are constants, and functions a =a (t): i0 i1 i i a¨ = ξ for i=1,2,3 i 3 γ−1 ai(cid:18)kΠ=1ak(cid:19) (54) ai(0)=ai0 >0, a˙i(0)=ai1, with arbitrary constants a and a . i0 i1 Further researchwill be carried out to investigate the possibility of these vorticalsolutions in 3D. References [1] CercignaniC.,IllnerR.,andPulvirentiM.(1994),TheMathematicalTheoryofDiluteGases, Applied Mathematical Sciences 106, Springer-Verlag, New York. [2] Chen G.Q. and Wang D.H. (2002), The Cauchy Problem for the Euler Equations for Com- pressible Fluids, Handbook of Mathematical Fluid Dynamics, I, 421–543, North-Holland, Amsterdam. [3] ConstantinA.(2006),BreakingWaterWaves,EncyclopediaofMathematicalPhysics,383–386, Elsevier. [4] Einzel D. (2006), Superfluids, Encyclopedia of Mathematical Physics, 115–121,Elsevier. [5] KwongM.K.andYuenM.W.,PeriodicalSelf-similarSolutionstotheIsothermalEuler-Poisson Equations for Spiral Galaxies, Pre-print. [6] Li T.H. (2005), Some Special Solutions of the Multidimensional Euler Equations in RN, Comm. Pure Appl. Anal. 4, 757–762. 10 M.W.Yuen [7] Li T.H. and Wang D.H. (2006), Blowup Phenomena of Solutions to the Euler equations for Compressible Fluid Flow, J. Differential Equations 221, 91–101. [8] LionsP.L.(1996,1998),MathematicalTopics inFluidMechanics, 1, 2, Oxford Lecture Series in Mathematics and its Applications, 3, 10, Oxford Science Publications. The Clarendon Press, Oxford University Press, New York. [9] Makino T. (1993), Exact Solutions for the Compressible Eluer Equation, Journal of Osaka Sangyo University Natural Sciences 95, 21–35. [10] Nagle R.K., Saff E.B.and Snider A.D. (2008), Fundamentals of Differential Equations and Boundary Value Problems, 5th edition, Addison-Wesley, Reading. [11] Yuen M.W. (2008), Analytical Blowup Solutions to the 2-dimensional Isothermal Euler- PoissonEquations of Gaseous Stars, J. Math. Anal. Appl. 341, 445–456. [12] YuenM.W.(2011),Exact,Rotational,InfiniteEnergy,BlowupSolutionstothe3-dimensional Euler Equations, Phys. Lett. A 375, 3107–3113. [13] YuenM.W.(2011),PerturbationalBlowupSolutionstotheCompressible1-dimensionalEuler Equations, Phys. Lett. A 375, 3821–3825. [14] Yuen M.W. (2012), Self-similar Solutions with Elliptic Symmetry for the Compressible Euler andNavier–StokesEquationsinRN, Commun. Nonlinear Sci. Numer. Simul.17,4524–4528. An, H.L. and Yuen, M.W. (2013), Supplement to ”Self-similar Solutions with Elliptic Symmetry for the Compressible Euler and Navier-Stokes Equations in RN” [Commun Nonlinear Sci Numer Simu. 17 (2012) 4524-4528], Commun. Nonlinear Sci. Numer. Simul. 18, 1558–1561. [15] ZhangT.andZhengY.X.(1997),ExactSpiralSolutionsoftheTwo-dimensionalEulerEqua- tions, Discrete Contin. Dynam. Systems 3, 117–133.