International Journal of Non-Linear Mechanics 47 (2012) 413417

Contents lists available at ScienceDirect

International Journal of Non-Linear Mechanics

journal homepage: www.elsevier.com/locate/nlm

On order reduction of non-linear equations of mechanics and mathematical physics, new integrable equations and exact solutions
A.D. Polyanin a,, A.I. Zhurov a,b,
a b

Institute for Problems in Mechanics, Russian Academy of Sciences, 101 Vernadsky Avenue, Bldg 1, 119526 Moscow, Russia Cardiff University, School of Dentistry, Heath Park, Cardiff CF14 4XY, UK

a r t i c l e i n f o
Article history: Received 16 August 2010 Received in revised form 16 April 2011 Accepted 18 April 2011 Available online 2 June 2011 Keywords: Hydrodynamic-type transformations Backlund transformations Integrable equations Order reduction Exact solutions NavierStokes equations Euler equations Boundary-layer equations Crocco transformation Generalized Calogero equation Non-linear equations of mathematical physics

abstract
Some classes of non-linear equations of mechanics and mathematical physics are described that admit order reduction through the use of a hydrodynamic-type transformation, where a rst-order partial derivative is taken as a new independent variable and a second-order partial derivative is taken as the new dependent variable. The results obtained are used for order reduction of hydrodynamic equations (NavierStokes, Euler, and boundary layer) and deriving exact solutions to these equations. Associated B acklund transformations are constructed for evolution equations of general form (special cases include Burgers, Korteweg-de Vries, and many other non-linear equations of mathematical physics). A number of new integrable non-linear equations, inclusive of the generalized Calogero equation, are considered. & 2011 Published by Elsevier Ltd.

1. Preliminary remarks In uid mechanics, two special (hydrodynamic-type) transformations are used for reducing the order of boundary-layer equations [13]. One of them is the Crocco transformation, in which a rst-order partial derivative is taken as a new independent variable and a second-order partial derivative is taken as the new dependent variable (this applies to the equation for the stream function). So far, using the Crocco transformation has been limited solely to the theory of boundary layer. The present paper reveals that the domain of application of the Crocco transformation is much broader. It can be successfully used for reducing the order of wide classes of non-linear equa tions with mixed derivatives and constructing Backlund transformations for evolution equations off arbitrary order and quite general form, special cases of which include Burgers and Korteweg-de Vries type equations as well as many other non-linear equations of mathematical physics. The Backlund transformations

obtained with the Crocco transformation may, in turn, be used for constructing new integrable non-linear equations. Examples of the generalized Calogero equation and a number of other integrable non-linear second-, third-, and fourth-order equations are considered. A generalization of the Crocco transformation to the case of three independent variables is given. Various Backlund transformations and their applications to specic equations of mathematical physics can be found, for example, in [314]. In the present paper, the term integrable equation applies to non-linear partial differential equations that admit solution in terms of quadratures or solutions to linear differential or linear integral equations.

2. Non-linear equations that admit order reduction with the Crocco transformation Consider the nth-order non-linear equation with a mixed derivative   @2 u @2 u @u @2 u @3 u @n u at u bt x 2 F t , , 2 , 3 ,..., n : 1 @t @x @x @x @x @x @x 1. Regardless of the specic form of the right-hand side, this ~ t , x family of equations has the following general property: if u

Principal corresponding author. Corresponding author at: Institute for Problems in Mechanics, Russian

Academy of Sciences, 101 Vernadsky Avenue, Bldg 1, 119526 Moscow, Russia. E-mail addresses: polyanin@ipmnet.ru (A.D. Polyanin), zhurovai@cadiff.ac.uk (A.I. Zhurov). 0020-7462/$ - see front matter & 2011 Published by Elsevier Ltd. doi:10.1016/j.ijnonlinmec.2011.04.032

414

A.D. Polyanin, A.I. Zhurov / International Journal of Non-Linear Mechanics 47 (2012) 413417

is a solution to Eq. (1), then the function 1 ~ t , x jt bt jt j0t t , uu at at c0, 2

where jt is an arbitrary function, is also a solution to Eq. (1). ~ t , x jt is another solution to (1). If at  0, then u u 2. Denote

passing to the Crocco variables (5), (3) leads to the rst-order equation   at Z bt @ 1 @ g t , Z f t , Z , @t F @Z F F which becomes linear with the substitution F 1=C. In the special case of at 1, bt 0, f t , ux 0, and g t , ux g ux , Eq. (10) reduces to the Calogero equation, which was considered in [15,16] (see also [3, pp. 433434]). Example 2 (Equation arising in gravitation theory). The non-linear third-order equation utxx kuuxxx , which is cross-disciplinary between projective geometry and gravitation theory [16,17], can be reduced, by integrating with respect to x, to the form utx kuuxx 1 2kwx ct ,
2

@u , @x

@2 u : @x2

Dividing (1) by uxx F, differentiating with respect to x, and taking into account (3), we obtain
n2 Ft utx Fx @ F t , Z, F, Fx , . . . , Fx at Z bt : 2 @x F F F

Let us pass in (4) from the old variables to the Crocco variables: t , x, u ut , x )t , Z,

F Ft , Z,

11

where Z and F are dened by (3). The derivatives are transformed as follows: @ @Z @ @ @ uxx F , @x @Z @Z @x @Z @ @ @Z @ @ @ utx : @t @t @t @Z @t @Z

where ct is an arbitrary function. Eq. (11) is a special case of Eq. (10), and hence can be reduced to a linear rst-order equation. Example 3 (NavierStokes and Euler equations). An unsteady three-dimensional ow of a viscous incompressible uid may be described by the NavierStokes and continuity equations  2  @Vn @Vn @Vn @Vn 1 @ Vn @2 Vn @2 Vn V1 V2 V3 rn P n , @t @x @y @z r @x2 @y2 @z2 n 1, 2, 3, @V1 @V2 @V3 0, @x @y @z 12

As a result, Eq. (4), and hence the original equation (1), is reduced to the (n 1)st-order equation at Z bt    @ 1 @ 1 @F @n2 F F t , Z , F, F , . . . , n 2 : @t F @Z F @Z @x 6

The higher derivatives are calculated by the formulas @k u @k2 F @ @k3 F F , @Z @xk3 @xk2 @xk @ @ F , @x @Z k 3, . . . , n:

Given a solution to the original equation (1), formulas (3) dene a solution to Eq. (6) in parametric form. Let F Ft , Z be a solution to Eq. (6). Then, in view of (3), the function u(t,x) satises the equation uxx Ft , ux , 7

where x, y, and z are Cartesian coordinates, t is time, V1, V2, and V3 are the uid velocity components, P is the pressure, r is the uid density, and n is the kinematic viscosity; also r1 P @P =@x, r2 P @P=@y, and r3 P @P=@z. Eqs. (12) are obtained under the assumption that the bulk forces are potential and included into pressure. In the degenerate case of n 0, Eqs. (12) become the Euler equations for an ideal (inviscid) uid. The equations of motion of a viscous incompressible uid, (12), admit exact three-dimensional solutions of the form V1 u, 1 @u V3 z , 2 @x Z P 1 1 @u @u dx, pt y2 z2 st u2 n 2 @x @t r 4 1 @u V2 y , 2 @x

which can be treated as an ordinary differential equation in x with parameter t. The general solution to Eq. (7) may be written in parametric form as x Z Z ds jt , Z0 Ft , s u Z Z s ds ct , Z0 Ft , s 8

where jt and ct are arbitrary functions and Z0 is an arbitrary constant. Since the derivation of (6) is based on differentiating (1), one of the arbitrary functions in solution (8) is redundant. In order to remove this redundancy, it sufces to substitute (8) into (1). However, it is more convenient to take advantage of the solution property (2) and note that solution (8) must also possess this property. In view of this, the general solution to the original equation (1) can be rewritten in the parametric form x Z Z ds jt , Z0 Ft , s u Z Z s ds 1 bt jt j0t t , at Z0 Ft , s 9

where p(t) and s(t) are arbitrary functions of time t, and u ut , x satises the non-linear third-order equation   @2 u @2 u 1 @u 2 @3 u u 2 n 3 pt , 13 @t @x 2 @x @x @x which is a special case of Eq. (1) with a(t) 1, b(t) 0, and 2 F nuxxx 1 2ux pt . The Crocco transformation (5) brings (13) to the non-linear second-order equation   @F 1 @F @2 F Z2 pt nF2 2 , 14 2 @t @Z @Z which can be rewritten in the form of a non-linear equation of convective heat conduction with a parabolic, Poiseuille-type velocity prole:     @C 1 @C @ 1 @C 1 Z 2 pt n , C : 2 @Z F2 @Z @t @Z F It should be noted that in the special case of inviscid uid (n 0), the original non-linear equation (13) is reducible to the

where jt is an arbitrary function. Example 1 (Generalized Calogero equation). With F f t , ux uxx g t , ux , which corresponds to the non-linear second-order equation utx f t , ux at ubt xuxx g t , ux , 10

A.D. Polyanin, A.I. Zhurov / International Journal of Non-Linear Mechanics 47 (2012) 413417

415

linear rst-order partial differential equation (14), which can be solved by the method of characteristics. Example 4 (System of hydrodynamic-type equations). Consider the system of equations  2 @2 u @2 u @u @3 u @u u 2 pt , 15 n 3 qt @t @x @x @x @x @x @v @v @u @2 v u v n 2, @t @x @x @x 16

p s2 a2 t ,

q qt is an arbitrary function,

where C1, C2, and s are arbitrary constants. By setting qt ns2 j0t t with periodic jt , one obtains a periodic solution in both x and t. More complicated solutions to Eq. (15) can be found in [20]. Remark. With regards to Examples 3 and 4, which consider exact solutions to hydrodynamic equations, the interesting recent paper [21] should be mentioned, which constructs a rst integral to the NavierStokes equations and discusses its applications. 3. Some generalizations Consider the non-linear nth-order equation
n ct utx at u bt xuxx dt ux utx ut uxx F t , ux , uxx , . . . , ux ,

which describes several classes of exact solutions to the Navier Stokes equations in two and three dimensions [3,1820]. The non-linear equation (15) is independent of v and can be treated separately. Although linear in v, Eq. (16) involves the function u, which is governed by Eq. (15). The Crocco transformation (5) brings system (15) and (16) to the form @F @F @2 F Z2 qZ p Z qF nF2 2 , @t @Z @Z @v @v @2 v Z2 qZ p Zv nF2 2 : @t @Z @Z 17

23 which becomes (1) for c(t) 1 and d(t) 0. 1. Regardless of the specic form of the right-hand side, this family of equations possesses the following general prop~ t , x is a solution to Eq. (23), then the function erty: if u ~ t , x jt ct , uu where j jt and c ct are related by dt c0t at c ct j0t bt j (either function can be chosen arbitrarily), is also a solution to (23). 2. Let us divide (23) by uxx, differentiate the resulting equation with respect to x, and then pass to the Crocco variables (5), (3) to obtain the (n 1)st-order equation
atZ bt

18

Here and henceforth, the arguments of p(t) and q(t) are omitted for brevity. Eq. (18) has been obtained by isolating the mixed derivative utx in (15) and using the resulting expression. Eq. (18) has exact solutions of the form v AZ BF C , 19

where A A(t), B B(t), and C C(t) are unknown functions determined from an appropriate system of ordinary differential equations. This fact can be proved by substituting (19) into (18) and taking into account (17). Formula (19) allows one to arrive at the following important result with regard to solutions of the original equation (16). Let u u(t,x) be a solution to Eq. (15). Then Eq. (16) admits the solution v A0t Aq A @u @2 u B 2 , @x @x 20

dtZ ct

   @ 1 @ 1 @F @n2 F : F t, Z, F, F , . . . , n2 @Z @t F @Z F @x

where A A(t) and B B(t) satisfy the ordinary differential

0 0 A00 tt qAt p qt A 0,

21 22 R

Example 5. In the special case of n 3, at bt ct 0, dt 1, and F f uxx x , Eq. (23) is a general boundary-layer equation for a non-Newtonian uid [3], with u being the stream function. By the Crocco transformation (5), this equation can be reduced to the second-order equation ZFt F2 f FZZ , which can be linearized by the substitution C 1=F if f F 1=F. Remark. Eq. (23) can be generalized by adding the arguments m Jx , . . . , Jx , with J uxx utxx utx uxxx , to the function F. 4. Using the Crocco transformation for constructing RF-pairs and B acklund transformations Consider a fairly general nth-order evolution equation

B0t qB 0:

Thee general solution to (22) is B C1 exp q dt, where C1 is an arbitrary constant. Listed below are some exact solutions to Eq. (15) representable in terms of elementary functions and suitable for nding exact solutions to Eq. (16) using formulas (20). 1. Generalized separable solution rational in x: u a0t t bt x at 6n , x at q 4b, p b0t 3b ,
2

n : ut at u bt xux F t , ux , uxx , uxxx , . . . , ux

24

where a at and b bt are arbitrary functions. 2. Generalized separable solution exponential in x:

~ t , x is a solution to Eq. (24), then the General property: if u function ~ t , x ct C , uu where C is an arbitrary constant and c ct satises the linear ordinary differential equation c0t bt c Cat 0 is also a solution to (24). Differentiating (24) with respect to x yields an (n 1)st-order equation with a mixed derivative of the form (1): utx at u bt xuxx at u2 x bt ux @ n F t , ux , uxx , uxxx , . . . , u x : @x 25

a0 u at e bt , p 0, q t sbs2 n, a where a at and b bt are arbitrary functions and s is an arbitrary constant. By choosing periodic functions as at and
sx

bt, one obtains time-periodic solutions. 3. Multiplicative separable solution periodic in x:   Z qt dt , u at sinsx C1 , at C2 exp ns2 t

416

A.D. Polyanin, A.I. Zhurov / International Journal of Non-Linear Mechanics 47 (2012) 413417

By passing in (25) from t, x, u to the Crocco variables (5), we one arrives at the nth-order equation
3at Z 2bt

ux vy 0,

34

  @ 1 @ 1 @2 @F @n2 F at Z2 bt Z F t, Z, F, F , . . . , n2 : @Z @t F @Z F @Z2 @x

26 Eqs. (24) and (26) are linked by the Backlund transformation

n2 ut at u bt xZ F t , Z, F, Fx , . . . , Fx ,

which describes a at unsteady boundary layer with pressure gradient (u and v are the uid velocity components) [13]. Eqs. (34) can be reduced, by introducing a stream function w such that u wy and v wx , to a single third-order equation [1,3]: wty wy wxy wx wyy nwyyy f t , x: 35

ux Z,

uxx F:

27

Remark. Sometimes, it is convenient to rewrite (26) in the form

Ct atZ2 btZCZ 3at Z 2btC

@2 F, @Z2

This equation is a special case of (33) (up to the obvious renaming u-w. Dividing (35) by wyy followed by differentiating with respect to y and passing from t, x, y, w to the Crocco variables t, x, Z wy , F wyy , one arrives at the second-order equation @F @F @F @2 F Z f t , x nF2 2 , @t @x @Z @Z 36

Example 6. The non-linear second-order equation ut auux kuxx

m

28

is a special case of (24). The Backlund transformation (27) reduces (28) to the equation 3aZ

@ 1 @ 1 @2 aZ2 k 2 Fm : @t F @Z F @Z

which is reduced, with the substitution F 1=C, to the nonlinear heat equation   @C @C @C @ 1 @C Z f t , x n : 37 2 @Z C @Z @t @x @Z Remark. In the steady-state case with @=@t 0 and f t , x 0, Eq. (36) reduces to the one considered in [1,3]. 1. In the special case f t , x f t , Eq. (37) admits an exact solution of the special form Z 1 C Z x, t, x xZt tf t dt, t t3 : 3 Hence we arrive at the integrable equation   @Z @ 1 @Z n , @t @x Z 2 @x

29

In the case m 1, the original equation (28) is the unnormalized Burgers equation; it generates the new integrable equation (29), which can be rewritten as a non-linear convective heat transfer equation with a source:

Ct aZ2 CZ kC2 CZ Z 3aZC, C 1=F:

In the case m 1, the resulting equation (29) becomes linear after substituting F 1=C. Hence, the original equation (28) also admits linearization for m 1. Example 7. The linear third-order equation ut auxxx buxx 30

38

is a special case of (24) with F auxxx buxx aFFZ bF, at  0, and bt  0. By applying the Backlund transformation (27) to Eq. (30) and then substituting F 1=C, one arrives at the nonlinear equation

which can be reduced to the linear heat equation [3,22]. 2. In the more general case f t , x f t x g t , we have solutions of the special form Z C Z x, t, x jtx ctZ yt , t c2 t dt, where j jt , c ct , and y yt are determined by the linear system of ordinary differential

Ct aC3 CZ ZZ bC2 CZ Z :

31

j0t f c 0, c0t j 0, y0t g c 0:

As a result, we arrive at an integrable equation (38). 6. Conclusion

The special cases of (31) with a 0, b a 0 and b 0, a a 0 were considered in [22,3], respectively. 5. Extension of the Crocco transformation to the case of three independent variables. Application to unsteady boundarylayer equations Transformation (5) can be extended to the cases of more independent variables. In particular, it can be shown that the Crocco transformation t , x, y, u ut , x, y )t , x, Z, where Z uy and

F Ft, x, Z,
32

F uyy ,

reduces the order of the nth-order equation at , xu bt , xyuyy c1 t , xuty c2 t , xuxy d1 t , xuy uty ut uyy
n d2 t , xuy uxy ux uyy F t , x, uy , uyy , . . . , u y :

33

Example 8. Consider the Prandtl system ut uux vuy nuyy f t , x,

We have presented some classes of non-linear equations of mechanics and mathematical physics that admit order reduction through the use of the Crocco transformation, in which a rstorder partial derivative is taken as a new independent variable and a second-order partial derivative is taken as the new dependent variable. Associated Backlund transformations have been constructed for some evolution equations of general form (special cases of which include Burgers and Korteweg-de Vries type equations as well as many other non-linear equations of mathematical physics). The results obtained have been used for order reduction of hydrodynamic equations (NavierStokes, Euler, and boundary layer) and constructing exact solutions to these equations. A number of new integrable non-linear equations, inclusive of the generalized Calogero equation, have been considered. The Crocco and Backlund transformations described in the paper can be used to construct other new integrable non-linear equations off mechanics and mathematical physics.

A.D. Polyanin, A.I. Zhurov / International Journal of Non-Linear Mechanics 47 (2012) 413417

417

Acknowledgments The work was carried out under partial nancial support of the Russian Foundation for Basic Research (Grants Nos. 08-01-00553, 08-08-00530 and 09-01-00343). References
[1] L.G. Loitsyanskiy, Mechanics of Liquids and Gases, Begell House, New York, 1995. [2] H. Schlichting, M. Gersten, Boundary Layer Theory, eighth ed., Springer, Heidelberg, 2000. [3] A.D. Polyanin, V.F. Zaitsev, Handbook of Nonlinear Partial Differential Equations, Chapman & Hall/CRC Press, Boca Raton, 2004. [4] N.H. Ibragimov, CRC Handbook of Lie Group Analysis of Differential Equations, vol. 1, CRC Press, Boca Raton, 1994; N.H. Ibragimov, CRC Handbook of Lie Group Analysis of Differential Equations, vol. 2, CRC Press, Boca Raton, 1995. [5] R.M. Miura (Ed.), Backlund Transformations, Springer-Verlag, Berlin, 1975. [6] C. Rogers, W.F. Shadwick, Backlund Transformations and their Applications, Academic Press, New York, 1982. [7] N.A. Kudryashov, Analytical Theory of Nonlinear Differential Equations, Institut Kompjuternyh Issledovanii, MoscowIzhevsk, 2004 (in Russian). [8] J.J.C. Nimmo, D.J. Crighton, Backlund transformations for nonlinear parabolic equations: the general results, Proc. R. Soc. London A 384 (1982) 381401. [9] J.G. Kingston, C. Rogers, D. Woodall, Reciprocal auto-Backlund transformations, J. Phys. A: Math. Gen. 17 (1984) L35L38. property, J. Math. Phys. [10] J. Weiss, Backlund transformations and the Painleve 27 (1986) 12931305.

[11] M.C. Nucci, Nonclassical symmetries and Backlund transformations, J. Math. Anal. Appl. 178 (1993) 294300. [12] A. Karasu, S.Yu. Sakovich, Backlund transformation and special solutions for the DrinfeldSokolovSatsumaHirota system of coupled equations, J. Phys. A: Math. Gen. 34 (2001) 73557358. [13] M. Wang, Y. Wang, Y. Zhou, An auto-Backlund transformation and exact solutions to a generalized KdV equation with variable coefcients and their applications, Phys. Lett. A 303 (2002) 4551. [14] X. Yonga, Z. Zhangb, Y. Chen, Backlund transformation, nonlinear superposition formula and solutions of the Calogero equation, Phys. Lett. A 372 (2008) 62736279. [15] F. Calogero, A solvable nonlinear wave equation, Stud. Appl. Math. 70 (1984) 189199. [16] M.V. Pavlov, The Calogero equation and Liouville-type equations, Theor. Math. Phys. 128 (2001) 927932. [17] V.S. Dryuma, On initial value problems in theory of second order ODEs, in: M. Boiti, L. Martina et al.(Eds.), Proc. Workshop on Nonlinearity, Integrability, and All That: Twenty Years after NEEDS79 (Lecce, Italy), World Scientic, Singapore, 2000, pp. 109116. [18] A.D. Polyanin, Exact solutions to the NavierStokes equations with generalized separation of variables, Dokl. Phys. 46 (10) (2001) 726731. [19] V.V. Pukhnachev, Symmetries in the NavierStokes equations, Usp. Mekh. 4 (1) (2006) 676 (in Russian). [20] S.N. Aristov, A.D. Polyanin, Exact solutions of unsteady three-dimensional NavierStokes equations, Dokl. Phys. 54 (7) (2009) 316321. [21] M. Scholle, A. Haas, P.H. Gaskell, A rst integral of NavierStokes equations and its applications, Proc. R. Soc. A-Math. Phys. Eng. Sci. 467 (2125) (2011) 127143. [22] G.W. Bluman, S. Kumei, On the remarkable nonlinear diffusion equation au b2 ux x ut 0, J. Math. Phys. 21 (1980) 10191023.