Factorization method and new potentials with the oscillator spectrum

Bogdan Mielnik

Citation: J. Math. Phys. 25, 3387 (1984); doi: 10.1063/1.526108

View online: http://dx.doi.org/10.1063/1.526108
View Table of Contents: http://jmp.aip.org/resource/1/JMAPAQ/v25/i12
Published by the AIP Publishing LLC.

Additional information on J. Math. Phys.

Journal Homepage: http://jmp.aip.org/
Journal Information: http://jmp.aip.org/about/about_the_journal
Top downloads: http://jmp.aip.org/features/most_downloaded
Information for Authors: http://jmp.aip.org/authors

Downloaded 01 Aug 2013 to 132.174.255.3. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jmp.aip.org/about/rights_and_permissions
 Factorization method and new potentials with the oscillator spectrum
Bogdan Mielnika )
Departamento de Flsica, Centro de Investigaci6n y de Estudios Avanzados dell P. N., 07000 Mexico D. F.,
Mexico
(Received 2 March 1984; accepted for publication 15 June 1984)
A one-parameter family of potentials in one dimension is constructed with the energy spectrum
coinciding with that of the harmonic oscillator. This is a new derivation of a class of potentials
previously obtained by Abraham and Moses with the help of the Gelfand-Levitan formalism.
PACS numbers: 02.30.Em, 03.65. - w

I. INTRODUCTION These relations allow the construction of the eigenvec-

tors and eigenvalues of H. If 1/1 is an eigenvector of H
Contrasted with general relativity, where new solutions (H1/I = ,1,1/1) the functions a*1/I and a1/l [provided that they are
are found every now and then, the class of exactly soluble nonzero and belong to L 2(R )] are new eigenvectors corre-
problems of quantum mechanics (QM) has not greatly ex- sponding to the eigenvalues A + 1 and A - 1, respectively:
panded. The two most commonly used exact methods to
determine the spectra in QM are the method of the orthogo- H(a*1/I) = a*(H + 1)1/1 = (A + l)a*1/I, (2.7)
nal polynomials and the algebraic method of "factoriza- H(a1/l) = a(H - 1)1/1 = (A - l)a1/l. (2.8)
tion." The potentials for which exact solutions exist form a Since the operator H is positively definite, one immedi-
rather narrow family, including the elastic and Coulomb po-
ately finds the lowest energy eigenstate 1/10 as the one for
tentials (modified by the 1/r 2 terms), the Morse potential, which
the square potential wells, and a few others, and the common
opinion is that this is everything exactly soluble in Schro- .1.
a'f/O = "----' d
V-r'-e
x'/2.1. "----'.1. ( )
'f/o=V-r''f/OX = Coe - x'/2
, (2.9)
dinger's quantum mechanics. Hence, it might be of interest dx
to notice that in some occasions the "factorization" method and one checks that the corresponding eigenvalue is ,1,0 = ~.
seems not yet completely explored. In particular, it allows Using now the operator a*, one subsequently constructs the
the construction of a class of potentials in one dimension, ladder of other eigenvectors 1/In corresponding to the next
which have the oscillator spectrum, but which are different eigenvalues An = n + !:
from the potential of the harmonic oscillator. This class has
been previously derived using the Gelfand-Levitan formal-
ism.
1/In = C n(a*r1/lo = C n( - l)n[ ~'/2 :x: e - X'/2]e - x'/2

(2.10)
II. CLASSICAL FACTORIZATION METHOD
where Hn (x) are the Hermite polynomials given by the Ro-
The factorization method in its most classical form was driguez formula
first used to determine the spectrum of the Hamiltonian of
the harmonic oscillator in one dimension: (2.11)
1 d2 1
H= - - -2 + _x 2 • (2.1) The nonexistence of any other spectrum points and eigen-
2 dx 2
states follows from the completeness of the Hermite polyno-
The method consisted of introducing the operators of mials. The above method was first employed by Dirac. 1 Its
"creation" and "annihilation" extension for the hydrogen atom was found by Infeld and
a -_ - 1 (d- + x ) -_ "",'/2
-1e _ x'/2-----e, d (2.2)
Hull. 2 A generalized presentation is due to Plebafiski. 3 The
group theoretical meaning is owed to Moshinsky,4 Wolf,5
J2 dx J2 dx
and other authors. Yet, there is still one aspect of the method
*
a = J21 ( - d
dx +x = -
)
J21 e
x'/2 d
dx e
_ x'/2
, relatively unexplored. It can be used not only to find the
interdependence between different spectral subspaces of the
(2.3) same operator but also to transform one Hamiltonian into
with the properties another.
a*a =H -~} III. MODIFIED HAMILTONIAN
aa
* _ H 1 =>[a,a*] = 1.
- +2
(2.4)
Consider once more the factorized expression
Hence,
Ha* = a*(H + 1),
H +~ = aa*. (3.1)
(2.5)
Are the operators a and a* here unique? Define the new
Ha=a(H-l). (2.6) operators

alO nleave of absence from Department of Physics, Warsaw University, b= _1 (~+P(X)), (3.2)
Warsaw, Poland. J2 dx

3387 J. Math. Phys. 25 (12), December 1984 0022-2488/84/123387-03$02.50 © 1984 American Institute of PhysiCS 3387

Downloaded 01 Aug 2013 to 132.174.255.3. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jmp.aip.org/about/rights_and_permissions
 ifJl = b *¢o, ifJ2 = b *¢I'"'' ifJ" = b *¢n _ I'''' (3.15)
b * = _1 ( - ~ + {3 (X)), (3.3)
..fi dx are the eigenvectors of H' corresponding to the same eigen-
and demand that H + ! be written alternatively as valuesA n = n +!:
H'ifJn =H'b *¢n-l = b *(H + 1)¢n_1 = b *(n + !)¢n-I
H +! = bb *. (3.4)
This leads to = (n + !)ifJ" (n = 1,2, ... ). (3.16)
The functions t/J" are square integrable because of the asymp-
1 d2 1 2 1 1 ( d2 {3' 2) totic behavior of ifJ (x) for x- ± 00. They are obviously or-
- T dx2 + T X + T = T - dx2 + +{3,
thogonal, as (ifJj,ifJk) = (b *¢j_I' b *¢k- I) = (¢j_I'
(3.5)
bb*¢k_I)=(¢j_I' (H+!)¢k_l) =k(¢j_I' ¢k_d=O,
and so, the condition for {3 is the Ricatti equation for k #j. However, they do not yet span the whole of L 2(R ).
{3' + {32 = 1 + x 2• (3.6) The missing element is the vector ifJo orthogonal to all of ifJ"
(n = 1,2, ..... ),
The occurrence of the Ricatti equation in the factoriza-
tion problems is a typical phenomenon. 2 ,3 In general, the
explicit solution of this type of equation is not known. This is
= (t/Jo,b *¢" - I) = ~(bt/>o'¢n _ I) =
(t/Jo,ifJ,,) °
(for n = 1, 2, ... )~bifJo = 0, (3.17)
not the case in (3.6), where one has one particular solution and so, the "missing vector" is found from the first-order
{3 = x. Hence, the general solution can be obtained putting differential equation
{3 = x + t/J (x). This yields
t/J' + 2tjJx +t/J 2 =~'/t/J2 + 2x(1/t/J) + 1 = 0. (3.7)
bifJo = _1 [~+ {3 (X)]ifJo =
..fi dx
°
Introducing now a new function y = 1/ifJ, one ends up
with a first-order linear inhomogeneous equation
~o = coe - x'/2exp(lX ifJ (x')dx ). (3.18)

- y' + 2xy + 1 = 0, (3.8) By the very definition (3.18), ifJo is another eigenvector of H'
corresponding to the eigenvalue Ao = 1:
whose general solution is
H't/Jo = (b *b + !)ifJo = !ifJo'
y = (r + f e- x" dX)~', rER.
(3.19)
As the system of vectors ifJO,ifJl'" is complete in L 2(R ), the
operator (3.12) is a new Hamiltonian, whose spectrum is that
Hence, of the harmonic oscillator, although the potential is not.
e-~ e-~ Since our initial Hamiltonian is parity invariant, the final
ifJ(x) = ., ~{3(x)=x+ ----- conclusion should remain valid when x_ - x. Indeed,
r + f~ e - x dx' r + f~ e - .<" dx' .
(3.9) though each one of the potentials (3.13) is not parity invar-
iant, the whole class is: V( - x,r) = V(x, - r) (Irl > ![ii).
The introduction of the operators b, b * might seem to offer
The reader can verify that what we have obtained here is the
little new, as we have still bb * = aa* = H + l' However, the
same class of potentials that Abraham and Moses obtained
commutator of band b * is not a number:
by using the Gelfand-Levitan formalism (Ref. 6, Sec. III,
[b,b *] = {3 '(x) = 1 + ifJ '(x). (3.10) starting on the top of p. 1336; see also papers by Nieto and
Hence, the inverted product b *b is not H + const, but it Gutschick 7 and NietoS).
defines a certain new Hamiltonian Remark: Differently than for the oscillator, the eigen-
vectors ifJ" admit no first-order differential "rising opera-
b *b = bb * + [b *,b ] = H + ! - 1 - ifJ ' = H' - !, tor." The ifJ,,'s are constructed not by a rising operation, but
(3.11)
due to their relation to the ¢" 's, which can be schematically
where represented as in Fig. 1. This means, however, that for the
H'=H-ifJ'(x)= - ~d: +V(x), (3.12) ifJ" 's there is a differential "rising operator," but it is of the
2 dx third order: A * = b *a*b. As far as we know, the use of the
with higher-order rising and lowering operators in spectral prob-

= -x - - d[
2
e-
x lems has not yet been explored.
V (x) ' ]
. (3.13)
2 dx r + f~e-x" dx'
If Irl >![ii, the above potential has no singUlarity and be-
haves like x 2 /2 for x- ± 00; and so, one obtains here a one-
H' I H
parameter family of self-adjoint Hamiltonians in L 2(R ). As
!A- ~
one can immediately see, their spectra are identical to that of i ~ A= b·ab
the harmonic oscillator, though their eigenvectors are differ-
··l~~ H'A·=A- (H' + II
ent. Indeed, (3.4) and (3.11) imply
H'b * = (b *b + !)b * = b *(bb * +!) = b *(H + 1).
(3.14)
.,I,J'
400 • 00/0
H'A=A(H'-II

Hence, for ¢n (n = 0, 1, ... ) being the eigenvectors of H, the

functions FIG. I. The relation between the t/Jn's and the <Pn 'so

3388 J. Math. Phys., Vol. 25, No. 12, December 1984 Bogdan Mielnik 3388

Downloaded 01 Aug 2013 to 132.174.255.3. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jmp.aip.org/about/rights_and_permissions
 ACKNOWLEDGMENTS
The author is grateful to Augusto Garcia for his kind
hospitality at the Departamento de Flsica del Centro de In-
vestigaci6n y de Estudios A vanzados in Mexico. Thanks are
due to Jerzy Plebafiski and other colleagues at the Departa-
mento de Flsica for their interest in this work and helpful
discussions. The discussion with Marcos Moshinsky in Oax-
tepec in 1982 is remembered.
3389 J. Math. Phys., Vol. 25, No. 12, December 1984
Downloaded 01 Aug 2013 to 132.174.255.3. This article is copyrighted as indicated in the abstract. Reuse of AIP content is subject to the terms at: http://jmp.aip.org/about/rights_and_permissions
Ip. A. M. Dirac, The Principles of Quantum Mechanics (Oxford U. P., Ox-
ford, 1958).
2L. Infe1d and T. E. Hull, Rev. Mod. Phys. 23, 21 (1951).
3J. Plebaiiski, Notes from Lectures on Elementary Quantum Mechanics
(CINVESTAV, Mexico, 1966).
4M. Moshinsky, "The Harmonic Oscillator in Modern Physics," in From
Atoms to Quarks (Gordon and Breach, New York, 1969).
'K. B. Wolf, J. Math. Phys. 24,478 (1983).
6p. B. Abraham and H. E. Moses, Phys. Rev. A 22,1333 (1980).
7M. M. Nieto and V. P. Gutschick, Phys. Rev. D 23, 922 (1981).
8M. M. Nieto, Phys. Rev. D 24,1030 (1981).
Bogdan Mielnik 3389