Volume 135, number 2 PHYSICS LETTERS A 13 February 1989

PARTICULAR SOLUTIONS OF A NON-LINEAR SCHRöDINGER EQUATION

CARRYING PARTICLE-LIKE SINGULARITIES REPRESENT POSSIBLE MODELS
OF DE BROGLIE’S DOUBLE SOLUTION THEORY

Jean-Pierre VIGIER
Inslitut Henri Poincaré, Laboratoire de Physique Théorique, 11, rue P. et M, Curie, 75231 Paris Cedex 05, France

Received 18 July 1988; revised manuscript received 25 October 1988; accepted for publication 5 December 1988
Communicated by A.R. Bishop

The stochastic derivation of Schrodinger’s equation as describing collective motions on top of a superfluid covariant non-
dissipative chaotic background (aether) leads to a particular non-linear Schrodinger equation if one considers possible vacuum
dissipative effects. In this Letter the existence of specific associated non-dispersive soliton-like (particle-like) solutions piloted
by the surrounding field is established. These solutions can be consdered as the first known possible representations of de Broglie’s
double solution theory.

The linear character of the Schrodinger equation is the amplitude of a regular solution of the linear
has been associated since the first years of the for- equation and S, S’ represent their respective phases.
mulation of quantum theory with the undulatory De Brogue was then able to show that under the con-
character of the possible solutions, the probabilistic ditions that: (a) the amplitude ofthe singularity var-
interpretation ofquantum mechanics in the frame of ies like 1 /r, (b) the displacement of the singular-
—

the Copenhagen school and the wave—particle dual- ity occurs in a uniform non-dispersive way, i.e., that
ity in a mutually exclusive form. On the other hand every point of the singularity moves with a uniform
any realistic approach which has attempted to reveal velocity v then velocity of the singularity v=VS/rn
simultaneous wave and particle aspects in the quan- and the velocity associated with the phase function
turn laws of nature has led to the presence of a cer- S’ of the regular wave Ø’(i.e. v’ = VS’ / rn) are iden-
tam form of non-linearity in a modified version of tical if the phase functions coincide, i.e. if S= S’. This
the Schrodinger theory [1,2]. is the “guiding principle” of de Broglie [1], which
De Broglie first formulated some non-linear as- implies that the particle beats in phase with its sur-
pects of the theory motivated by the need to enclose rounding wave and that a singularity of the 1 /r type
the possibility of particle motion description in the follows the mean lines of flow calculated from the
theory. This is the theory of the “double solution” linear solution, v’ = VS/rn, if the two solutions (the
[1] where the particle is represented by a non-dis- linear and the singular one) have the same phase. It
persive singularity in the wave function, active only is physically justified by the fact that real oscillating
in a region of the classical dimensions ofthe particle, systems only exchange energy if they beat in phase
i.e. ~ lO_13 cm and zero almost everywhere. A total with each other.
quantum solution therefore should read As one knows de Brogue never gave an explicit non-
linear form ofthe Schrodinger equation correspond-
U(x,t)=U0(x,t)exp[iS(x 1)/h]
ing to the theory of the double solution (and to the
+ U’ (x, t) exp[iS’ (x, 1)/h] guiding principle) restricting himself to the exami-
— ~ + v,,~, ‘ nation of the
described physical
above. properties
His ideas of the solution
were revived as
in a set of
where U0(x, t) is the amplitude ofthe singularity, U’ publications of Mackinnon [21 who examines pos-

0375-9601/89/s 03.50 © Elsevier Science Publishers B.V. 99

(North-Holland Physics Publishing Division)
Volume 135, number 2 PHYSICS LETTERS A 13 February 1989

sible forms of non-dispersive de Broglie wave pack-

ets and proposes soliton wave functions of the form jh = —
2ço+Vço (5)
sin Kr ~ ~—V
= Kr exp [i (WI K —

0x)] , (1) if one neglects all the non-linear terms multiplying

2/h, K the friction coefficient y. At the laboratory scale, of 2
where Krn0c/h, w=rn0c 1/2 0=rn0v/h and course, one cana neglect
(which implies theloss
negligible term iyhln(~/ç~,*)l~
of energy of the i,u
((x_vt)2 wave to the vacuum) and concentrate on a study of
r= \ 1 — v 2/c2 + Y + Z2) specific solutions of the non-linear equation

which represent a particle ofmass rn travelling in the = — h~ 2 v2 ( w*w) 1/2

+x direction with a velocity v relative to the ob- at ~V~+ VW+ ~ (çy*~)l/2 li’, (6)
server and obeying the equation
discussed by Smolin [6] and Guerra and Pusterla
000=0. (2) [5]51

Before we do this we make two remarks.

Guéret and Vigier [3] then succeeded in showing
that this soliton form of a wave function is a solution (a) All non-linear wave equations raise a funda-
mental interpretation problem in quantum mechan-
of a non-linear wave equation which is given e.g. for
ics since they violate the superposition principle (ev-
the Klein—Gordon case by the following relation idently absolutely necessary if one wants to interpret
m ~c2 [0 (U* U) ]1/2 the w wave as probability waves) which follows nat-
o U— h2 U= (U* U)”2 U, (3) urally from the linearity of the quantum mechanical
formalism.
where the non-linearity appears to have the form of In a non-linear equation the sum of two solutions
a quantum potential term Qcx: DI UI / I UI. The re- w~and W2 is never in general a solution of the wave
lation of these solutions to the linear ones is exam- equation itself unless certain constraints are im-
med in the frame of de Broglie’s double solution in posed between them.
ref. [4]. As a consequence the ~i,wave solutions of non-lin-
Parallel to this line of research (which follows de ear equations are not connected in general with
Broglie’s original reasoning) where the non-linear probability waves. One should thus explicitly ex-
equation results as a consequence of promoting a plain their connection with probability theory if one
certain form of singular solution, a completely in- introduces them within the quantum mechanical
dependent approach of Guerra and Pusterla [5] and formalism.
Smolin [6] arrived at the same result. They discov- (b) The constraints necessary to turn a sum of two
ered non-linear Schrodinger and Klein—Gordon solutions into a solution ofa non-linear equation have
equations with quantum potential non-linearity by led to one of Einstein’s most striking discoveries.
allowina formally the diffusion coefficient of the sto- With Infeld and Hofmann he has indeed shown that
chastic process in a stochastic quantization approach within the frame of the non-linear gravitational
[71 to differ h/2rn and thus justified relation (3). equations one can only superpose a Schwarzschild
In a preceding letter [8] we have established that
the introduction into the Stochastic Interpretationof 1*1 One interesting question that naturally arises as a consequence
Quantum Mechanics of vacuum dissipative effects of the above considerations is the following. Why should we
implies the substitution of the non-linear equation keep the non-lineartenn in the r.h.s. of eq. (3) since it is mu!-
tiplied by the factor y whose magnitude is practically negligi-
= — ~ ~2 2 V2 ( yj*~pr)1/2 ble. The answer is related to the properties of the solutions of
ôt 2rn W+ Vt~i+2rn (ç~/I~~)1/2 this non-linear equation, which, as shown by Hasse [9], ad-
mits soliton-like solutions. If such a solution is singular then
— iyh ln ( ~ /2~, (4) we have no right to neglect the non-linear term at least at the
site ofthe singularity, and this regardless ofthe magnitude of
to the Schradinger equation the factory.

100
Volume 135, number 2 PHYSICS LETTERS A 13 February 1989

singularity to an external gravitational field if it fol- with practically negligible energy dissipation at our
lows a geodetic of this external field. As a conse- scale) so that relation (7) reduces to the Schrödin-
ger equation (5) if k2 h2 and can be deduced from
—*
quence the laws of motion of the corresponding test
particle result from the concept of continuity of the the non-linear Lagrangian
field derivatives along time-like boundaries of geo- k2 h2
detic hypertubes. This astonishing result explains for .5~=iky’~y,— —

the first time in history the meaning of the laws of 8m

motion (in this particular case) which reflect in the k2+h2
unity offields and particles: henceforwarddescribed + 4rn [(Vw) (V~)]+ w’~V~. (8)
as field singularities in a continuous background.
It is thus tempting to extend this Einsteinian con- (2) The solution w(x, 1) of (7) can now be shown
cept to w waves considered as describing real phys- to be in a one-to-one correspondence with the so-
ical quantum fields surrounding soliton-like parti- lutions ~(x, 1) of the linear theory defined by rela-
des. In such a model micro-objects are thus described tion (5) where one substitutes k for h, i.e. with so-
by the superposition of particle-like soliton-like sin- lutions of the equation
gularities guided (piloted) along the de Broglie—
Bohm trajectories by the non-linear character of the i~ô,v= — i— ~ ~+ V~ti, (9)
wave equations. However this idea (first proposed
by de Broglie under the name of double solution the- which can be written in the form
ory [1]) implies a justification of the success of the
utilization of probabilistic ~ waves (satisfying linear w(x, t) = R exp (iS/k)
equations and the superposition principle) to de- one obtains in splitting (7) into real and imaginary
scribe observed statistical results. parts
An answer to the problems raised by these remarks
has been first proposed by Guerra and Pusterla [5]. ~ s= (VS)2+ V—
—~---

It lies in the observation that it is possible for certain ‘ 2rn 2rn R

types of non-linear equations (such as (6)) to de- a 2= div(R2VS/m)
— , (10)
fine a one-to-one correspondence between some of 1R
their continuous solutions (denoted w(x, I)) and the which are equivalent, if one assumes y= 0, to the
solutions (denoted ~(x, t) of linear equations (such usual Schrödinger theory for the wave function
as (5)) which satisfy the correspondence principle w= R exp(iS/h). At this point we must remark that
and define the same probabilistic density and current. because k~h eq. (7) cannot predict exactly the same
In the hydrodynamic description of wave fields properties as relation (5). Indeed, for example, the
[11] (characterized by a Hamilton—Jacobi equation correct energy levels ofthe Schrodinger equation (5)
and a continuity equation) this implies that ~‘ and are obtained by making ~ single valued (following to
ç~define the same Madelung—de Broglie (pilot) fluid the de Broglie—Bohr—Sommerfeld conditions on
from which one deduces, as one knows [2], all ob- closed orbits in the classical limit) so that the energy
served quantum mechanical predictions. The argu- levels deduced from w= R exp (iS/k) and
ment can be split as follows: q~’=R exp (iS/h) must be slightly different if ~t’is as-
(1) The non-linear equation (6) can be written in sumed (as one should) to be a single valued field.
the form Such differences are however not a drawback (if y is
a = —
k2
~—V2yi+V~’+y
rn
~
~
~ 1/2
W)
~ yi, (7)
small enough) since one knows such levels are only
approximate and do not account for real observed
effects like the Lamb shift.
if we replace h by a new constant k corresponding to One remarks immediately that this non-linear
the reduced diffusion coefficient v = (h/2rn )e~as- Schrodinger equation [10] is no longer tied to a pos-
sociated with a dissipative process [7,8]. Of course sible inequality of rnquantum and rnjnenia, as advanced
y = k2 h2 must be very small (since we are dealing
— by Smolin [6], nor to a physically unclear intro-

101
Volume 135, number 2 PHYSICS LETTERS A 13 February 1989

duction of two different Planck constants h’, h” as we see that one can calculate solutions i~vof the non-
proposed by Guerra and Pusterla [5]. It clearly re- linear equation (6)
sults from a slight modification of v = h 2/2rn in a — by calculating the amplitude R and phase S/h (with
background metric with random fluctuating corn- k= 11) of a solution ~ ofthe Schrodinger equation (5).
ponents. It is also clear that we can use the same ar- — by multiplying the said phase SL by a factor h/~so
gument, as Smolin [61 does, for the inertial and that the corresponding solution w of (4) becomes
quantum mass equivalence in order to estimate the
order of magnitude of this non-linearity. We there- w=R exp (iS/n) with a = h/k. (14)
fore use the experimental uncertainty in the Lamb The connection between cii and ~, thus implies the in-
shift determination of the hydrogen energy levels and troduction of a new type of superposition principle
obtain for the w fields since it remains valid for the cone-
(~
p )2 sponding ~ solutions of the linear equation (5).
~4xl0—’3, (11) Moreoverasaresultofeqs. (9)weseethattheprob-
ability densityp=R2 and the current densityj 2/
which represents an upper limit for the non-linear rn)V 1= (R
1S deduced from w correspond exactly with those
effects in the Schrödinger theory. In order to obtain of the corresponding ~, so that the two wavefunctions
a rough estimation of the scale where this pertur- ~ and w (which only differ by a multiplying factor
bation ofthe metric (and of the v = h/2rn value) ~ a =1i~/hin their phases) have in common the Ma-
comes significant we can use the results of Blokhin- delung fluid corresponding to the hydrodynamic
tsev [13] where this scale is identified with the representation of the Schrödinger equation (5) so
Planck length: that all statistical predictions of ~‘ are reproduced by
2 i~v.The only possible observable difference between
L0 = (~“) 10~32cm, (12) ~ and w evidently results from the constant multi-
\ C / 2 and x is the newtonian gravita- plying factors
equivalent to a in their
very phases.
small Suchofathe
variation difference
numericalis
tional x= 8ity/c However small this scale might seem
where constant. value of h in distributions of operators such as
to be it has to be noted that it is much larger than
, A = ihô,
—

2 which would appear for example as a very

the gravitational radii ofthe particles L5 =xrn, which small observable deviation (—~ h) from the en-
~—

is the scale where usual gravitational effects are ergy levels (in the Lamb shift for example) in re-
thought to become significant. At this scale our sim- lativistic quantum mechanics: a problem which will
plified approach breaks down [12,13] because the be discussed in a subsequent paper since it implies
basic assumption that the subquantum fluctuations a generalization of our model to relativistic wave
are quasi-independent of the metric fluctuations (the equations.
lattice being implicitly considered by a slight mod- From the preceding analysis one1L =immediately de-
RR exp (iSL/~)
ification
the metricoffluctuation
v) is no more
willvalid [12—14].In
couple that case
with the quantum duces that if one adds a regular
to a singular solution of eq. (5) its non-linearity im-
ci
random process [13,141 and the latter will no more plies that their motions are no longer independent so
be conservative; as assumed throughout this that the center of the particle-like singularity follows
approach. a specific trajectory with respect to the linear lines
Any solution of eq. (7) can be calculated from a of flow: exactly like Schwarzschild singularities fol-
solution p= RL exp (iSL/h) of eq. (5). Indeed if we low the geodetics of an external ~ field.
take as solution of eq. (4) the wave field It is thus very Interesting (taking into account the
preceding mapping ctrt9,) to analyze the physical be-
~= R exp (iSL/~) (13) haviour of some general singular solution w of (5)
related to ~ by the relations with respect to the motions of an associated regular
solution cvR in order to see if any particular solution
~=~t/~t’, ~ln(~,/~,*)= ln(çv/~ft), follows (or not) the de Broglie—Bohm trajectories
[15], which are tangent at each point to P=VSR.

102
Volume 135, number 2 PHYSICS LETTERS A 13 February 1989

As we shall now see if we add a regular solution

2rn5v2 RL
form ~ exp(iS~/k)with a singular solution of the
WR=RR

5=R5exp(iS5/~)where (I) SR=SS, (2) r

denotes the minimum distance taken on any surface as can be seen by making use of the H—J type equa-
SR = const between any point of this surface and the tions of the linear and the non-linear SE. Can we ne-
interaction of a particular drift line P= VSR with glect such a term? The answer is affirmative because
SR = const. it involves a quantum potential term of the linear
Then the center of this singularity follows the de SE, which is regular and bounded everywhere and
Broglie—Bohm trajectories [2]. This solution thus can be neglected due to the scaling by the factor
evidently provides the first known explicit example (5 ~ )2 It is therefore, to a very good approximation,
of de Broglie’s double solution model with particle- justified to consider the combination w= WL+ WN as
like singularity, a solution of the linear and non-linear SE. This jus-
The argument can be split as follows: tifies the de Broglie reasoning that the true solution,
(1) Suppose that ~ (x, t) and çv2(x, t) are two so- combining the wave and the particle aspect of mat-
lutions of the non-linear SE of the form ter, should be a common phase solution to the non-
linear SE.
~ =R, exp(iS/k), W2 =R2 exp(iS/k), (3) Suppose now that W=WL+ YiN is a solution to
the non-linear SE and cvN has a singular structure (e.g.
i.e. possessing a common phase 5, then a superpo-
sition principle holds for this choice of wave func- a 1 /r singularity) around a point r0 at time to. Then
tions ~i, and W2, and cv1 + W2 is again a solution of the everywhere except the immediate vicinity {V,0 } of
non-linear SE. The restriction to equal phase S re- r0, the wave function and the quantum potential are
sults from bounded. Since the non-linearity is scaled
2 factor we can set thebynonlin-
the al-
seesthe
thatexistence of the non-linear
readily without the phase equality term.
5, = S One most vanishing (St’)
2=
the superposition principle does not apply anymore. earity almost equal to zero in P3 {V,,3 } and the so-
—
lution of the non-linear SE is then to a good ap-
(2 )Assume that WL is a solution of the linear SE
and YiN a corresponding solution of the non-linear proximation the regular solution v= WL. Cone-
SE, and furthermore
functions, i.e. assume equal phases for both is valid,
spondingly, in1Nthe butofsince
vicinity r0 i.e. is singular
YiN {V,0 in
} the full
rsolution WL + ci
0, we can neglect WL with respect to YiN and put
Y/L =RL exp(iS/k), ci’N =RN exp(iS/k). “= YiN. We can therefore approximate a solution to
the nonlinear SE in the following way:
then one can show that the following properties hold:
(a) the superposition y/=WL+WN=(RL+RN) YiYiN, in{V,,},
x exp (iS/f) satisfies the continuity equation = WL, 1fl P3 {V,~
— },
2+ak[(RL+RN)2(akS/m)]=o. and map WL on a solution q~of (5).
a,(RL+RN) (4) Consider a simple example of a non-linear so-
The proof can be given by using the continuity lution whose amplitude is spherically symmetric and
equations of the linear and non-linear SE separately. square normalizable. The latter condition rules out
(b) The Hamilton—Jacobi type equation of the the solution advanced in refs. [1,3,4] since they ev-
Bohm decomposition [151 of the non-linear SE is idently diverge. A possible candidate is then
approximately valid, i.e.

as (VS)2 Si’2) I~i(RN+R) WN= (

2+ RN+RL withr=[(x—vt)2+y2+z2]”2.
21)l,2rexP(r/l)exP[i(Kx wt)] (16)
2m —2rn(ii (15)
Here SI YiN 12 d3x= 1, 1 is the length scale at which the
The symbol arises due to a violation of the strict 1 /r singularity becomes significant, and the con-
equality by a term of the magnitude stants are related as follows:

103
Volume 135, number 2 PHYSICS LETTERS A 13 February 1989

2—(1+A)/12], with2=5v2/v2.
w= 2rn [K
-~- at +ak(R~
4) rn
(SN)

This singularity Is non-dispersive since any point for re {V.~} . (19)

moves with the constant speed v along the x-direc- This is the equation d (RN) /dt= 0 for an undis-
J
tion.
YiN Furthermore it has
12F(r)47tr2dr= ~ the following property(17)
(~1)kF~(0) . turbed
kaksN/,transport of (RN)
which
2 along the lines of flow
coincide with the flow lines of
the linear solution provided the two phases SN and
SL are identical, i.e. SN = 5L’ Given this property [9]
Since I is a very small quantity the right hand side a non-dispersive transport of the non-linearity is
of the above equation can be approximated by the guaranteed along the lines of flow calculated from
first term in the series and yIelds the linear SE (5).

J Yi~I 2F( r)4~tr2dr~F(0). (6) Note

solutions ofthat the quantum
the type potential
of eq. (19) can beof non-linear
shown to be
constant, i.e.
We see that I YiN 12 behaves approximately as a ö(r)
function. This has some important physical conse- ~ (er~t/r) —1
(20)
quences. First of all, let us assume that the distri- e~’~/r —

bution ofsingularities in a particle or time ensemble Following our preceding considerations we can
is proportional to the I YiL I 2density Then due to the represent the total QP as follows,
Bohm—Vigier H-theorem [16], we know that this
distribution is stable, since any perturbation thereon ~ (RN +RL) — RN L~RN RL ~RL
always decays to the equilibrium OM state. Hence it RN +RL = RN +RL R+R +RL RL
preserves the quantum mechanical statistical predic-
tions. But in addition to this property I YiN 12 weights It can be easily seen that outside the singular re-
as ô-function the continuous local field variables ~ (RN RL) this expression reduces to the linear
‘~

A (x) associatedto the quantum field Yi via the Bohm QP while inside the singular region (RN>> RL) it can
prescription [15] be identified with the non-linear QP. Therefore, the
effective QP that lies at the origin of quantum cor-
ç relations is essentially throughout space equal to the
A(x) =Re , (18) QP calculated from the linear solution except at the
site of the singularity where it takes the value 1/12.
where A is the operator corresponding to A (x). Therefore no modification of the quantum correla-
Therefore one can speculate if by making the ö-type tions or statistics [16,171 is expected as a conse-
structure ofthe singularities responsible for the mea- quence of this non-linear model,
surement interaction one could account for the dif- Concluding we wish to stress that on the basis of
ference between sharp measured values and quan- this non-linear Schrodinger theory with quantum
tum expectation values. The former should then be potential non-linearity deduced from stochastic
due to measurement-like interactions of the singu- principles, de Broglie’s model on the theory of the
larities while the latter would reflect the distribution “double solution” or “pilot wave” interpretation can
of measurement events. Work in progress on this be derived and justified. The extension of this ap-
subject will be reported elsewhere. proach to relativistic quantum mechanics of spinless
(5) The transport property of the singularity can and spinning particles as well as a field theoretical
be considered on the basis of its continuity equation analysis of the non-linear equation based on a La-
(6a) [16]. Given the above form of eq. (6a) for grangianformulation are left for future publications.
YiN=RN exp(iS/k) and using the property
R~/I VR~I —‘0 for r—~0 it can be rewritten in the The author wants to thank Professor Guerra, Dr.
form Hasse and Dr. Kyprianidis for helpful discussions.

104
Volume 135, number 2 PHYSICS LETTERSA 13 February 1989

Dr. Kyprianidis’ help in particular has been very val- [8] P.N. Kaloyerou and J.P. Vigier, Phys. Lett. A 130 (1988)
uable to complete this research. 260.
[9] R.W. Hasse, Z. Phys. B 37 (1980) 83; Phys. Rev. A 25
(1982) 583.

References [10] E. Nelson,mechanics,

quantum Connection between Brownian
Contribution motion
to the Berlin and
Einstein
Symposium (1979).
[1] L. deBrogue, Une tentative d’interprétation causale et non-
linéaire de Ia mécanique ondulatoire (Gauthier-Villars, [11] J.P. Vigier, Lctt. Nuovo Cimento 29 (1980) 467.
Paris, 1956); [12]D.DhornandF.Guerra,Phys.Rev.D3l (1985) 2521;
N. Cufaro-Petroni et al., Phys. Rev. D 32 (1985) 1375; F. Guerra and R. Marra, Phys. Rev. D 28 (1983) 1916.
J.P. Vigier, Astron. Nachr. 303 (1982) 1. [13] D.I. Blokhintsev, Nuovo Cimento 16 (1960)1406; Soy. J.
[2]L. Mackinnón, Found. Phys. 8 (1978) 157; 9 (1979) 787; Part. NucI. 5 (1975) 242;
Lett.NuovoCimento3l (1981)37; 32 (1981)311. C.Frederick,Phys.Rev.Dl3(l976) 3183;
[3]Ph. Guéret and J.P. Vigier, Lett. Nuovo Cimento 35 (1982) Wh. Namsrai, Soy. J. Part. Nucl. 12 (1981) 449, and
256; 38 (1983) 125. references therein.
[4] Ph. Gu&et and J.P. Vigier, Found. Phys. 12 (1982)1087. [14] J.V. Narlikar and T. Padmanabhan, Gravity, gauge theories
[5] F. Guerra and M. Pusterla, Lett. Nuovo Cimento 34 (1982) and quantum cosmology (Reidel, Dordrecht, 1986).
351. [15]D. Bohm, Phys. Rev. 85 (1952) 166, 180.
[6] L. Smolin, Phys. Lett. A 113 (1986) 408. [16] D. Bohm and J.P. Vigier, Phys. Rev. 96 (1954) 208.
[7] E. Nelson, Phys. Rev. 150 (1966)1079; [17] F. Halbwachs, Théorie relativiste des fluides a spin
F. Guerra, Phys. Rep. 77 (1981) 263. (Gauthier-Villars, Paris, 1960) ch. 1.

105