Proc. Nat. Acad. Sci. USA Vol. 73, No. 3, pp.

687-691, March 1976 Physics

A relativistic spherical vortex

(relativistic hydrodynamics/pressure-density relation in a relativistic "incompressible" fluid/moving relativistic spherical vortex)

C. L. PEKERIS
Department of Applied Mathematics, The Weizmann Institute, Rehovot, Israel

Contributed by C. L. Pekeris, January 2, 1976

This investigation is concerned with stationABSTRACT ary relativistic flows of an inviscid and incompressible fluid. In choosing a density-pressure relation to represent relativistic "incompressibility," it is found that a fluid in which the velocity of sound equals the velocity of light is to be preferred for reasons of mathematical simplicity. In the case of axially symmetric flows, the velocity field can be derived from a stream function obeying a partial differential equation which is nonlinear. A transformation of variables is found which makes the relativistic differential equation linear. An exact solution is obtained for the case of a vortex confined to a stationary sphere. One can make all three of the components of velocity vanish on the surface of the sphere, as in the nonrelativistic Hicks spherical vortex. In the case of an isolated vortex on whose surface the pressure is made to vanish, it is found that the pressure at the center of the sphere becomes negative, as in the nonrelativistic case. A solution is also obtained for a relativistic vortex advancing in a fluid. The sphere is distorted into an oblate spheroid. The maximum possible velocity of advance of the vortex is

(2/3) c.

The purpose of this investigation is to seek solutions of the relativistic hydrodynamic equations that represent steady vortex motions analogous to Hill's spherical vortex (1) or Hicks's spherical vortex (2) in classical hydrodynamics. We shall assume the fluid to be inviscid and incompressible. The concept of "incompressibility" in relativity requires that we specify the value of the velocity of sound u. in the fluid, since the infinite value of us characteristic of the classical incompressible fluid of constant density is unacceptable relativistically. We shall assume that the velocity of sound is independent of pressure. This assumption implies a linear dependence of the density ji on the pressure p: [II p = Ao + nc-1p, where c denotes the velocity of light, and n is a constant. The values of n equal to 0, 1, and 3 correspond, respectively, to the three cases of an "incompressible" fluid discussed by Synge (3). The velocity of sound is given by us = C/ V-n, [2] so that the fluid of uniform density, with n = 0, suffers from the undesirable feature of having a velocity of sound exceeding c. The case n = 3, with Ao equal to zero, corresponds to a radiation gas. The choice of the value of n in Eq. 1 would have to be made on physical grounds. One of the results of this investigation is that the value of n = 1, corresponding to a fluid in which the velocity of sound exactly equals the velocity of light, is to be preferred for reasons of mathematical simplicity, just as the assumption of constant density in the nonrelativistic case is made for mathematical convenience. Another question that we wish to elucidate is whether, in the case of an isolated relativistic vortex on whose surface the pressure vanishes, the pressure in the interior becomes
687

negative. Such a negative pressure is found at the center of an isolated Hicks nonrelativistic spherical vortex (2). We show that in the case of axially symmetric relativistic flows, the velocity field can be derived from a stream function, which obeys a nonlinear partial differential equation. We found a function of the stream function for which the governing differential equation can be made linear in the relativistic case, and an exact solution was obtained. This function becomes identical with the stream function itself when the amplitude of the velocity field tends to zero. As in the nonrelativistic solution, it is possible to make all three of the components of velocity vanish on the surface of a sphere. Again, the pressure at the center becomes negative if the pressure on the surface of the sphere is made to vanish. The evidence from this solution is therefore that relativistic effects by themselves do not make for cohesiveness. If cavitation at the center is to be avoided, a uniform containing pressure must be applied at the surface, this pressure increasing with increase in the intensity of the flow in the vortex. Relativistic axially symmetric flows With u denoting the 3-velocity vector and
e
=

[1

_(U2/C2)]"1/2,

[3]

the relativistic hydrodynamic equation for a perfect fluid is

(3)
/2(p+c-2p)d ud -grad p-c-lu
I

[4]

while the equation of continuity takes on the form

a'y [5] + div (yu)j = 0. Yd/ + (p + C-2p) We seek a solution of the hydrodynamic Eqs. 4 and 5 that is stationary in a reference system in which the center of the

sphere is at rest.
Put
11
=
P

(C2A + p-)

dp 1c2Ip)lnp -+ (n+) C2 + (n + r)l ,)po

=Tn=+
w=

'

[6]
[7]

curl u,
(
=

cv,

Cr,

[81

then, with (a/at) vanishing, Eq. 4 becomes y2(r X v + '/2 grad v2) =-grad H. [9] Here po denotes an arbitrary reference value of the pressure,
and we have made use of the relation du =uau u+ g -=-+ u

+'A~grad u2.

[10]

688

Physics:

Pekeris

Proc. Nat. Acad. Sci. USA 73 (1976)

Eq. 9 can be written as TXv=-(1-v2) grad Q,

with

some mathematical simplification. Indeed, the divergence of the velocity then vanishes.

n + in -y. We now restrict ourselves to solutions that are indepen( of the longitude 4. In a spherical system of coordin (r,6,4), the vanishing of the X-component of [11] yields v- grad (r sin Ovp) = 0, while [5] becomes
Q
=

-ynv grad n + div (,yv) = 0. [14] use II can be eliminated from [14] by makingrrus The function

A transformation that leads to a linear differential equation for the relativistic "stream-function," in the case n = 1 In the following we restrict ourselves to the case n = 1, implying that we assume the sound velocity in the fluid to be equal to c. Even then the differential Eq. 24 for the stream-function I takes on the nonlinear form L(+) + ft = r2 sin2 g(1 v-2), [25] where L is the linear differential operator

of the relation
v grad Q= 0, which follows from [11]. The result is nv grad y = div (yv), from which it follows that d(r~sinvr -d a t H +E( r sin v s =0. t x -1 ' yn-1 ar,ny Hence, there exists a stream-function * such that n-1 aT r Vr= r 2 sin 0 ao r 6 r' r sin which, by [13], in turn leads to
=

a',sin / L()=a2*+r 6 a@sn dI' @ LNO= ~~ + r2

[26]

[15]
[16] [17]

The nonlinearity of [25] can stem from the arbitrary functions f(I) and g(I), but more seriously from the factor (1v2) which, by [18] and [19], is quadratic in A, or worse. It is possible to linearize [25] by taking

#(*+21+7*') f Op) =(1 + '14)

[27]

2n-1a

[18]

V'O

[(4') .
sin

'

[19]

[i'. f being an arbitrary function of The r and 6 components of [11] yield

IdQ _

9(*) = go +In (1 l*n)l [28] and choosing as a dependent variable the function F(I), given by I = (1/il)[(1 + 271F)12 -1_1 [29] F(') = * + 12X7742, where # and v are constants. This transformation yields the linear differential equation for F L(F) +f2F = ir2 sin2 6. [30] The velocity components are given by
Or = r2sn
1

,y2
I

dQ y2 ao
with

A a'W r2 sin2 6 ar ' A a4 r2 sin'2 6 d@'

[20]

+ 277F -/dtF

[31]

[21]

VH=~ r sin ft (1+2F-1

r

rdF

,n-1

'

aa+ -i 6 +71-1 sin 0 a yn-I d+\ + li r2 ao ,in t} ao) f

Q = g(I),

[22]

HenLce
A=
r

increases as the velocities become relativistic and as lul approaches c. The dependent variable F(I) can be considered a stream-function like I, in the sense of the vanishing of the component of velocity v,, normal to the surfaces F = constant.

[32] The parameter X is a measure of the intensity of the flow; it

vO =

3FA

sin

O'& (1 + 277F)-1/2.

sin

[23]

The relativistic spherical vortex, for the case n = 1

We now solve [30] for the case of a relativistic vortex confined to the interior of a stationary sphere of radius a. Let
F=
with

g being an arbitrary function of I. Summarizing, the steady and axially symmetric solution of the hydrodynamic Eqs. 4 and 5, for the case of a perfect fluid whose density-pressure relation is given by [1], is expressed through a stream-function ' by Eqs. 18 and 19, with I satisfying the differential equation

,?S(r) sin2 6,

[33]

S(r) =

Tn-1 d(n- 1a-)

+

j, W
=

(ai/f2,)I(r2/a2) -[jj(9r)/j(ga)]l, = (sin x/x)-cos x = V'Airx Ja(x). 2

[34]
[35]

- o)+r2ff a (ynl ~~ y'Ynsin 6 06+ ff yi r2

S satisfies the differential equation

.

[24]

The pressure is given by [12], with Q equal to g(I). It is clear from Eqs. 18 and 24 that the case n = 1 offers

S -(2/r2)S+,832S = r2 [36] of which [34] represents a solution satisfying the condition S(a) = 0, [37]

Physics: Pekeris
on the surface of the sphere. This condition assures, by [31], that on the surface of the sphere the radial component of the velocity vA(a) vanishes. By [32], the longitudinal component of velocity v,>(a) also vanishes then on the surface. Since, by [31] and [37],
[38] vo(a) = -[S(a)/a] sin 0, we can make the third component of velocity ve(a) vanish as well on the surface by choosing the arbitrary constant (3 so that S(a) vanishes. This can be achieved by choosing (3 to be a root of [39] Jk(fla) = 0. 2

Proc. Nat. Acad. Sci. USA 73 (1976)

689

denoting the 4-velocity, the energy tensor T,, of the fluid is (3) Trs = (, + c-2p)ArXAs + C-2P6rs, [46] Hence, with X42 = _Y2, we have e =-C2T4 = 'Y(c2 + P)-P = p(a )(2y2 -1) + C2jO[1 + OF + '/2v2(l + 2nF)]. [47] In the case of an isolated vortex, with p(a) = 0, we find that the total energy E, as given by the volume integral of E,
is

E=

fedV =

c2m0(1 + 1.6 v2),

[48]

where

We have thus obtained a solution of the relativistic hydrodynamic equations that represents a vortex confined to the interior of a stationary sphere. All three of the components of velocity vanish on the surface. Outside the sphere of radius a there is no manifestation of the existence of the vortex as long as the intensity of the vortex does not exceed a critical value specified below, above which the pressure at the center becomes negative and cavitation ensues. The pressure can be derived by substituting g(I) for Q in [12] and using [28]: [40] A, + 2p = (c20o + 2po)(1 + 27F)/y2. At the surface of the sphere, where F and v2 vanish, y becomes 1, and we get

[41] p(a) = po. For the interior we obtain from [40] [42] p = p(a)+c2po[?lF- '2v2(1 + 2i7F)] At the center of the sphere, S(r) vanishes like Kr2, v4g like r, and v2 approaches the value of 4t12K2 where
[43] 3-_(f[a)] Since v2 cannot exceed the value of 1, it follows that the intensity parameter q of the velocity field has an upper limit of v < 1/2K. The pressure at the center p(O) is > p(a)- Y20~po. [44]
K
=

'io = (4r/3)iwa3, v = (,Qa2/X), X = fla. [49] v is a nondimensional parameter representing the intensity of the velocity field. The integral m of q over the volume is m = fadV = m0(1 -0.8 v2), [50] the negative sign resulting from the effect of negative pressure. Nonrelativistic solution: The Hicks spherical vortex The nonrelativistic limit of our relativistic spherical vortex is the Hicks spherical vortex (2). It can be derived from our relativistic solution by considering I to be of the order of the small parameter ?, and going to the limit of the lowest i powers of q. v is of the order of uc. Thus, [27] and [28] become [51] F(' ) = ,
g (*)= go+ q*

[52]

a2

The last term in [52] is of order X2, but we need it because it is the only surviving term in k('), which alone enters Eq. 24. Eq. 29 yields [53] FA F) = T,

while [31] and [32] become

= r;smn Vr Vr=Hsi.6sineo~dr' Vo =sin O [[54] r sin 7rX in e Ov It follows that v2 is of order n2 and therefore that y is 1 + 0(72). Hence, Y can be taken as unity in [24], leading to [55] L(T) + #2F = qi2 sin2 6 differential Eq. 30 for F, as was which is identical with the to be expected from [53]. While [30] was derived for the case n = 1 only, [55] holds for all n, simply because the second term in [1] is of order q2 relative to the first term, and is therefore neglected in our nonrelativistic limit. The solution of [55] for the Hicks spherical vortex is I = qS(r) sin2 6, [56] which is formally identical with the relativistic solution for F given in [33]. The difference is, of course, that the velocities in the Hicks vortex depend linearly on I, as given in [54], while the velocities in the relativistic vortex depend nonlinearly on F, as given by [31] and [32]. The vanishing of S(a) and of A(a) through condition [39] again assures that all three components of velocity vanish on the surface of the Hicks spherical vortex. Similarly, relation [44] holds also in the nonrelativistic solution, showing that the pressure at the center of the Hicks vortex is negative if the vortex is isolated. A comparison of the relativistic solution with the nonrela-

a*

aI

Eq. [44] sets another upper limit on the amplitude factor q of the flow in the vortex, above which the pressure at the center of the sphere becomes negative and cavitation sets in. In particular, our relativistic vortex cannot be isolated, because then p(a) vanishes, and the pressure at the center becomes negative regardless of how small the amplitude factor n is. The density distribution ,u is not uniform inside the sphere because of the variable pressure. Using [1] with n = 1 and [42], we get [45] p = c- p(a)+ po[1 + iF-l2v2(1 + 27F)]. ,g(O) is positive even in the case when p(a) vanishes. The trajectories of the fluid particles are spirals on toruslike surfaces, which, in the case of the first mode, all thread a circle, centered at the origin, lying in the equatorial plane and of a radius r = 0.61 a. The energy density E remains positive throughout, as long as the radical (1 + 2qF)1/2 is real. Negative values of (1 + 2iqF) are excluded by [31] and [32]. It is found that the bound v2(0) < 1 ensures the positivity of (1 + 2nF). With X,

690

Physics:

Pekeris
Relativistic
vr
=

Proc. Nat. Acad. Sci. USA 73 (1976)

Outside the sphere the pressure can be derived from [12] by putting Q equal to the constant go. We get exp (2go) = (c2'o + 2p.)/j(c2'0 + 2po) * (1- V2)], [A8]
c ,i + 2p
=

tivistic solution is given below.

Nonrelativistic V,. = 2rq cos OS(r) r
VH
U,,

2 cos

OvnS(r)

r='[1 + 271S (r) sin2 0]1/2

=

(C2Ao+2p.)(1 V2)/(1 V2).

-

[A9]

= -t

S(r
>3
=

sin

6
.

sin 0 -r[l + 2i7S(r) sin2 0]1/2 4ii +

prS(r)

It follows from [A9] that the pressure p(a) on the surface of the sphere is given by

r' 0 rsin
=

On sin AS(r)
r

V,> =

On sin OS(r)

2tiS(r) sin2 9]1/2

p(a)= (I (/'1

- /4(1 42 sin2 0 + - V2)

.

2V' sin' oY [A10] / 4

iS(r)sin'0
Oq go)+ n

F = S(r) sin'0
f*

f('l)
g(+)

= =

=f(* + Y/2 71*2)

=
-

g(')
F(*)

1+ 77 g+ ln (1 + i')

As long as V <(2/3), p can be taken sufficiently large so as to make p(a) positive at the equator where it is the lowest. Indeed, the pressure can be made positive everywhere if p. is equal to, or exceeds, the value of

/2c2p4v2(O) - V2]/[1- V2(o)],

where v2(0) is the value of v2 at the center.

S =J r' Ifl S(r) )=a [ r2 _ jIfla)J fl2 La2 l(#!r)]

#(1+ /21ADD

APPENDIX A The moving relativistic spherical vortex, in the case n

=
1

Inside the sphere we have by [12], [23], [28], and [A8] c2j. + 2p = (c2po + 2p_)(1 -v2X + 26F)/(1 - VI), which matches [AlO] on the surface.

[All]

We wish to obtain a solution of the relativistic hydrodynamic equations for the case of a "spherical" vortex moving through a perfect fluid in the positive z-direction with a constant velocity W =cV. In a coordinate system in which the center of the vortex is at rest, the boundary condition at infinity is that the fluid move with a constant velocity in the z-direction of magnitude -W. We shall assume that in this coordinate system, moving with the vortex, the motion inside a sphere of radius a centered at the origin of coordinates is rotational, while outside this sphere the motion is irrotational. The assumption of the retention of the spherical shape of the vortex is not in contradiction with the requirement of the Fitzgerald contraction, because in the coordinate system adopted the sphere is not moving.

Outside the sphere, the longitudinal component of velocity v4, vanishes, and f(I) and k(') are zero. The stream-function I0 is, by [25], a solution of the differential equation [Al] L('I) = 0, which is the condition for irrotational motion. We shall impose the condition that the radial component of velocity at the surface vanishes:

We note from [A7] that the velocity reaches a maximum value of (3/2)V on the equator of surface of the sphere. The velocity parameter V is therefore bounded by the limit V < (2/3). [A12] By virtue of [A12] the velocity in the interior of the vortex is also less than c. We now transform our solution to a coordinate system (r',',4/) which moves with velocity -W in the z-direction relative to the original system (r,0,4)). In the primed system of coordinates the velocity of the fluid at infinity is zero, and the center of the vortex moves with velocity W in the positive z'-direction. The sphere is deformed into a Heaviside spheroid with the z'-diameter shrunk by the factor (1 - V2)1/2. With [A13] 7C = 1/ 1-V2, we have, in cylindrical coordinates (z,,4p), Z = z/'y0 = r'cos 0', [A14] while

= vr0(a)= ; [A2] so that the fluid particles originally inside the sphere do not leave it.
The appropriate solution of [Al] is q i - Y2Vr' sin' 0 + 1/2V(a3/r) sin2 0, yielding Vr =-V cos 0 + V(a'/r') cos 0,
=

r' sin O' = r sin 0 = i. This leads to the following transformation: r/ = r(1 - V2 cos2 0)1/2,

a'

[A15]
[A16]
[A17]

cos

0'

cos
=

0/[yo(l V
-

cos2

0)1/2],

[A3]
[A4]
=

sin 0'

sin 0/(1 - VI CoS2

0)2,

[AIS]

v,+ V
1+

V sin 0 + 'AV(a3/r3) sin 0, v vOO= 0. [A5] Inside the sphere the velocity field is given by [31] and [32], with
F = --2 aV sin'

VY

V(cOS2 0-'1/2sin2 0) [(r3/a')(1- V2) + V2(COS2 0-V2 sin2 0)]

[Al9]

(1 + VV)
3

01(r2/a2) [j,(flr)/j,(X)]$/j2 [Xjl(X)/j,(X)]j.

-

[A6]

F is a solution of [30] for which v7(a) vanishes and

vO(a)
=

V sin 0

[A7]

thus matching the exterior solution veo(a) given by [AS]. The parameter is no longer a root of [39], but is a function of V.

V sin 0 cos [A20] Yoly(r3/a')(1- V2) + V'(co'2 0-Y/2 sin2 0)] From [A19] and [A20] it follows that '=2V sin 0'(1-3V' cos' 0) -t(r3/a)(1- V2) + V2(cos' 0 - sin2 )f [A21] On the equator of the surface of the sphere (r a, 0 = 0' = wr/2), ve in [A21] reaches the value of unity when V = (2/3). Hence, both in the original system and in the primed system of coordinates the ve=

Physics: Pekeris
locity on the equator of the vortex reaches the value of c as V (2/8). The bound given in [A12] therefore applies to both systems, as was to be expected. This bound implies that the relativistic vor-

Proc. Nat. Acad. Sci. USA 73 (1976)

691

1. Hill, M. J. M. (1894) Phil. Trans. R. Soc. London Ser. A 185,

tex cannot advance with a velocity W greater than (2/3)c. At that

limiting velocity of advance the polar radius has shrunk by the Fitzgerald contraction to (1/3)/a = 0.745a.

213-245. 2. Hicks, W. M. (1899) Phil. Trans. R. Soc. London Ser. A 192, 33-100. 3. Synge, J. L. (1965) Relativity: The Special Theory (North-Holland, Amsterdam), 2nd ed., pp. 302-308.

Corrections

Proc. Natl. Acad. Sci. USA 73 (1976)

2527

Addendum. To the paper "Purification of folate binding factor in normal umbilical cord serum" by Barton A. Kamen and J. Douglas Caston, which appeared in the November 1975 issue of Proc. Natl. Acad. Sci. USA 72,4261-4264, the following note is added. The method for dissociation of endogenous bound folate from the natural folate-binder complex was essentially that used initially in the isolation and characterization of folate binder from hog kidney (1, 2) and subsequently employed in the discovery of a folate-binder complex in serum (3). It is a modification of an approach used to dissociate the synthetically produced folate-binder complex with the binder found in bovine milk, which was shown by Ford et al. (4) to have a strong pH dependency for association-dissociation equilibrium.
1. Kamen, B. A. & Caston, J. D. (1974) J. Lab. CGn. Med. 83, 164-171. 2. Kamen, B. A. & Caston, J. D. (1975) J. Biol. Chem. 250, 22032205. 3. Colman, N. & Herbert, V. (1974) CGun. Res. 22, 700A. 4. Ford, J. E., Salter, D. N. & Scott, K. J. (1969) J. Dairy Res. 36, 435-466.

Correction. In the article "A relativistic spherical vortex" by C. L. Pekeris, which appeared in the March 1976 issue of the Proc. Natl. Acad. Sci. USA 73, 687-691, the author has requested the following changes. On page 690, at the top of the left-hand column, the expressions [1 + 2nS(r) sin2 9]1/2 in the relativistic solutions should be replaced by [1 + 2772S(r) sin2 O]1/2, and the last term should read F(I) = I + (1/2)17I2. In Eq. 44, an editorial error was made. The correct equation
is:

p(O) = p(a) - 2c2,uon2K2

> p(a) -

(1/2)c2go

[44]

In Eqs. 45 and A6, printer's errors were made. The correct equations are:
F
=

y = c-2p(a) + yu0[1 + qF - (1/2)v2(1 + 2,7F)] -(3/2)a2V sin2 Of(r2/a2)

[45]

-[jj(#r/jI(X)]1/j2 [-A1j(A)/1j(A)]L.
-

[A6]

Correction. In the article "Affinity of myosin S-1 for F-actin, measured by time-resolved fluorescence anisotropy" by Stefan Highsmith, Robert A&Mendelson, and Manuel F. Morales published in the January issue of Proc. Natl. Acad. Sci. USA 73, 133-137, the authors have requested the following changes. On page 136 in Table 2 the association constant and inverted standard error obtained by S. Marston and A. Weber (ref. 3) was incorrectly quoted as (1.4 4 6) X 107 M-1 at 0.12 M KC1, and should be (1.4 + 0.12) X 107M-' at 0.14 M KC1.

Correction. In the article "Intramolecular ciosslinking of tropomyosin via disulfide bond formation: Evidence for chain register" by Sherwin S. Lehrer, which appeared in the September 1975 issue of the Proc. Natl. Acad. Sci. USA 72, 3377-3381, the author has requested the following changes. On page 3380, lines 10 and 11 in the right-hand column should read, "A mixture only of aa anda# chains would, .. On the same page, lines 17 and 18 in the right-hand column should read, "For the aa, af3 model the ratios would be (0.5, 0.5) and (0.6, 0.4) for a/fl = 3 and 4, respectively." Correction. In the article "Determination of the number of superhelical turns in simian virus 40 DNA by gel electrophoresis" by W. Keller, which appeared in the December 1975 issue of Proc. Natl. Acad. Sct. USA 72, 4876-4880, the author has requested the following change. On page 4879, in the sentence beginning on the tenth line of the right column, the two minus signs should be deleted. The corrected sentence is "For SV40 DNA this amounts to a reduction of r by 0.62.5200/360 = 9 turns."

Correction. In the article "Antigen stimulation of prostaglandin synthesis and control of immune responIses" by D. R. Webb and P. L. Osheroff, which appeared in the April 1976 issue of Proc. Natl. Acad. Sci. USA 73, 1300-1304, the authors have requested the following change. On p. 1301, the first line of the second column should read .... Ro 20-5720, an irreversible inhibitor, ...."