self-interacting one-dimensional oscillators
TRANSCRIPT
PRAMANA __ journal of
physics
~,~g: Printed in India Vol. 46, No. 3, March 1996 pp. 203-211
Self-interacting one-dimensional oscillators
MAMTA and VISHWAMITTAR* Department of Physics, Panjab University, Chandigarh 160 014, India *Author for correspondence
MS received 26 October 1995
Abstract. Energy eigenvalues and (x2), for the oscillators having potential energy V(x) = (to2x2/2) + 2 < x 2' > x 2' have been determined for various values of 2, r, s and n using renormalized hypervirial-Pad6 scheme. In general, the results show an improvement over the findings of earlier workers. Variation of the evaluated quantities and of the renormalization parameter with 2, r, s and n has been discussed. In addition, this potential has been employed as an illustrative example of the applicability of alternative formalism of perturbation theory developed by Kim and Sukhatme (J. Phys. A25 647 (1992)).
Keywords. Self-interacting oscillators; energy eigenvalues; perturbation theory.
PACS No. 03.65
1. Introduction
There are a number of important and interesting situations in physical and biological studies where the interaction between the system and its surroundings influence not only the former but also the latter. This results in the modification of environmental field so that the interaction potential acting on the system also depends upon its own state ~,. and is, therefore, written as V(~.). Accordingly, the time-independent Schr6dinger equation for the system becomes non-linear and reads [1-3];
H(~.)~,, = [H 0 + V(q,.)]~. = E,~,. 0 )
with H o as the Hamiltonian for the isolated system. A typical example of such a self- dependent system is one-dimensional oscillator described by the Hamiltonian (h = m = 1)
H = - (1 /2 ) (dZ /dx 2) + (to2 x2/2) + 2 (x 2 " )x 2s, (2)
where both r and s are integers. Obviously, when r = s = 1, (2) becomes Hamiltonian of a self-interacting harmonic oscillator whose force constant is linearly dependent on the mean square of displacement of vibration, and r >/1, s >/2 correspond to the self- interacting anharmonic oscillators whose anharmonicity depends on (x2"). Also for to = 0, r ~> 1, s t> 2 we have the self-dependent oscillators with extremely large magni- tudes of anharmonicity - the so-called infinite-field limit of the oscillators. Further- more, for r = 0 and s = 2 or 3 we get the special cases of well-studied quartic or sextic anharmonic (when to 4: 0) and pure quartic or sextic (to = 0) oscillators.
Since (1) cannot be solved exactly, some iteration and approximation methods have been used to determine the energy eigenvalues. An adaptation of Rayleigh-Schr6dinger
203
Mamta and Vishwamittar
perturbation theory (RSPT) has been put forward for this purpose and applied to the problem of a molecule in a polarizable medium [1,3,4]. Surj/m [3] also pointed out limitations of iterative method, configuration interaction approach and variational technique for solving the time-independent non-linear Schr6dinger equation. How- ever, since perturbation theory suffers from the uncertainty about its convergence, particularly for large perturbations, Cioslowski [5] employed an extension of connec- ted moments expansion to obtain the solution for (1) and illustrated it by finding the values of energy and mean square displacements of vibration <x2 > for the ground state (n = 0) of self-interacting harmonic oscillator for different magnitudes of 2. Vrscay [2] analyzed the perturbation expansion for (2) using hypervirial and Hellmann-Feynman theorems and performed numerical calculations to determine lower and upper bounds to energy for different values of r and s -- 2 corresponding to to = 0 as well as co 4:0 employing a renormalized RSPT wherein summation was carried out by the Pad~ approximants method.
Killingbeck [6-8] presented a variational parameter-based renormalized hyper- virial Pad6 scheme (RHPS) that yields very accurate energy eigenvalues (and also (x 2 ) values) in a simple manner for anharmonic oscillators. We examined the extent to which this technique is successful in finding the energies and the expectation values of x 2 for self-interacting harmonic and anharmonic oscillators and this communication is an outcome of the effort for a wide range of values of r, s, ;t and n. We shall discuss in the sequel the dependence of energy, (x 2) and the variational parameter on various quantities.
It is pertinent to mention that Kim and Sukhatme [9] have developed an alternative formalism for Rayleigh-Schr6dinger perturbation theory (ARSPT) for a one-dimen- sional problem described by linear Schr/Sdinger equation. Being an expansion in the powers of perturbation parameter, this leads to the same results as RSPT but without requiring cumbrous sums over intermediate eigenstates and also reduces to the logarithmic perturbation theory of Au and Aharonov [10]. We have also obtained expression for energy of the self-interacting oscillators in the framework of this approach and compared the results with the findings of RHPS.
2. Renormalized hyperviriaI-Pad~ calculations
2.1 Theoretical f ramework
Following Killingbeck's [6-8] prescription, the Hamiltonian (2) is renormalized by adding and subtracting (2KxZ/2) so that denoting (to 2 + ~,K) 1/2 by to', we get
H = ( - 1/2)(d2/dx 2) + (to'2x2/2) + A ( x Z ' ) x 2s - (2Kx2/2) • (3)
Using the hypervirial theorem together with the relevant commutation relations, expanding E n and (xN)n as
E .= ~ -~lT~J~'~-~, (4) j=O
<xN>= ~, C~N~2k (5) k=O
204 Pramana - J. Phys., Vol. 46, No. 3, March 1996
Self-interacting one-dimensional oscillators
and equating coefficients of the same powers of 2 on the two sides of the resulting equation, we obtain the recurrence relation
P C~N + 2~ = [(2 (N + 1)) /((N + 2)f0'2) "] ~ E(i~n C(N)'p-j _'4- [,,/tt.nllc'/'J2~]l...p_f'(N +12)
j=O
p - I - [ ( ( 2 ( N + s + l ) ) / ( ( N + 2 ) w ' 2 ) ] ~ C '2") C~N+2~) p-k-I
k=O
Here
C~ °1 = ~op and
,2 (N-2) + [ ( N - 1 ) N ( N + I)/4(N + 2)o~ ]C,, . (6)
(7)
°) = (n + ¢8)
The E~ ) forj /> 1 are related with C~ N) by using the Hellmann-Feynman theorem and this gives us
_,E ti~ = (l/j) ( - K / 2 ) C ~ 1 + ~ (k + l~C(2r)c k i - k - t " (9) k=0
Equations (8) and (9) when substituted in (4) give E, in terms of K and C~ N~ and the latter are determined from (6) in a hierarchial manner. Similarly, substitution of(6) in (5) with N = 2 leads to the formula for evaluation of (x 2). . It may be pointed out that the product terms C~2_'~ k_l C~ff ÷ 2s)in (6) a n d C(k2r)c~2_S)k_l in (9) allow the calculation to treat the term 2 (x 2') x 2s in (3) directly, without any need for iteration of any kind.
2.2 Numerical results and discussion
In order to execute the computations the parameter K is determined in such a way that maximum number of stable digits in the value is obtained from the final expression for E, for a particular state. Also it is found that
C(e2M+ 11= 0 (10)
for p, M = 0, 1, 2, 3,..., which is a consequence of even power terms in the potential. The computations for energy as well as (x 2 ) , have been executed with double precision and by terminating the series summation in (4) at j = 32; though in some cases, particularly those for which ). is large, summation had to be extended up to j = 52 (or even 80) to obtain proper convergence. However, in all cases the values being reported correspond to a situation for which the maximum number of digits was stable. The energy values so determined (E~) and those improved upon by finding Pad6 approximants, E,(P), to the series expansion with chosen K parameters are listed in tables 1-3 for different values of r, s, n, co and 2. Also projected in these tables are the relevant (x 2) . obtained by performing sum in (5) by the Pad6 approximants. With a view to compare our results with those of other workers, their values have also been included in these tables. It may be mentioned that the entries in the tables have been kept up to such an unrealistic number of significant figures just to emphasize the degree to which results of the present calculations can be trusted.
Pramana - J. Phys., Vol. 46, No. 3, March 1996 205
O~
T
able
-I.
A c
om
par
iso
n o
f th
e en
ergy
an
d (
x 2
)o v
alue
s fo
r th
e g
rou
nd
sta
te o
f th
e o
scil
lato
r de
scri
bed
by H
= (
- 1/
2)
(d2/
dx 2
) +
(to
2x2/
2) +
2(x
2 )x
2 f
or c
o =
I
and
dif
fere
nt v
alue
s of
2, i
n u
nit
s co
rres
po
nd
ing
to f
i = m
= 1
wit
h th
e fi
nd
ing
s E
o(C
) an
d (
x 2
)o(C
) of
Cio
slow
ski
[5].
-1
¢,.
2 K
E
o E
o(P
) E
o(C
) <
X2)
o
0"01
0"
0 0'
5024
8149
5972
0006
0"
5024
8149
5972
0006
0"
5024
82
0'49
7530
7588
5192
51
0"05
0"
0 0"
5120
6015
0107
5252
0"
5120
6015
0107
5252
0"
5120
60
0'48
8223
8931
2564
87
0'10
0'
0 0"
5233
4026
5902
3011
0"
5233
4026
5902
3011
0"
5233
40
0'47
7700
6782
9382
50
0'50
0'
5 0"
5957
4394
1976
5593
0"
5957
4394
1976
5594
0"
5957
44
0'41
9643
3776
0708
06
1"0
0"5
0"66
2358
9786
2237
30
0"66
2358
9786
2237
30
0"66
2359
0"
3774
3883
3123
3464
10
0"0
0"5
2'35
64
2"35
6698
839
0"10
6080
5885
8 10
00"0
0"
5 5"
02
5"01
6638
0"
0498
3388
10
000'
0 0"
5 10
"44
10'7
737
0'02
3205
8
(x 2)
o(C
)
0-49
7531
0.
4882
24
0.47
7701
0.
4196
43
0.37
7439
Ix
o T
able
2.
Cal
cula
ted
ener
gy e
igen
valu
es a
nd
(x
z) f
or t
he g
rou
nd
sta
te o
f th
e se
lf-i
nter
acti
ng a
nh
arm
on
ic o
scil
lato
rs (
2) w
ith
(o =
1,
s =
2 a
nd
dif
fere
nt v
alue
s of
r a
nd
2.
E0(
V)
are
the
resu
lts
from
Vrs
cay
[2]
whe
re t
he a
ctua
l en
trie
s ar
e th
e lo
wer
bo
un
ds
and
the
up
per
bo
un
ds
are
ob
tain
ed b
y us
ing
k di
gits
in
pare
nthe
sis
in p
lace
of
the
k di
gits
in
the
low
er b
ou
nd
s.
I" 0"0
2 K
E
o E
o(P
) E
o(V
) <
X2>
o
0"1
5"5
0'55
9146
3271
8351
9 0"
5591
4632
7183
5196
0"
5591
4632
7183
519(
21)
0"41
2525
3683
8365
03
0"5
3"5
0"69
6175
8207
64
0"69
6175
8207
6514
5 0"
6961
7581
6(25
) 0"
3058
1365
0717
5871
1"
0 3"
0 0"
8037
7065
13
0"80
3770
6512
3426
0"
8037
705t
7)
0"25
7149
8214
4280
97
5"0
1"5
1"2
24
58
70
36
1"
2245
8703
6059
0 1"
2245
7804
) 0"
1614
5496
0891
8456
10
"0
1"5
1"50
4972
4077
7 1"
5049
7240
7778
891
1"5
04
95
(9)
0"13
0022
0425
5119
37
100"
0 0'
5 3.
1313
842
3.13
1384
1649
3688
3'
1312
6(47
) 0'
0615
6216
4077
664
lO00
"O
0"5
6"69
4220
85
6"69
4220
8505
0403
1 6"
6939
(44)
0"
0286
9929
5440
1213
5 10
000-
0 0-
5 14
'397
70
14"3
9799
5343
518
14-3
97(8
) 0"
0133
3357
8444
69
gJ
Ila J .¢- g .o
I'0
2"0
3"0
0-1
0-5
1"0
5'0
10"0
10
0-0
1000
-0
1000
0-0
0-1
0"5
1'0
5"0
I0-0
10
0"0
lO00
"O
1000
0"0
0"1
0"5
1"0
5"0
10"0
10
0-0
1000
"0
1000
0"0
2"0
1-0
1"0
0"5
0"5
0"5
0-5
0"5
2"0
1"0
1-0
1"0
I'0
1-0
0-5
0-5
2"5
2"5
2-5
2"5
1-5
1"5
1"5
1-5
0-52
9717
4305
6164
20
0'59
7374
410
0-64
9058
4853
0-
8415
397
0"96
303
1"59
7 2"
707
6"0
0"53
6324
6515
445
ff60
0620
224
0'64
44
0"79
0"
899
1"22
2 2"
0 5'
7
0-55
6719
0"
62
0"64
0-
851
0"94
1"
094
2'5
7-7
0-52
9717
4305
6164
15
0"59
7374
4103
16
0'64
9058
4854
4 10
63
0"84
i 539
6038
4427
0-
9630
3104
203
1"59
9640
2'
707
4-63
9
0"53
6324
6515
4444
85
0'60
0619
8960
7489
5 0"
6443
9619
7563
2 0"
7929
23
0"88
0389
1-
2670
1"
795
2"74
7
0"55
6718
7518
3 0"
6272
9014
29
0-67
0213
97
0"80
28
0"87
6 1-
0432
1-
4308
2"
3
0-52
9717
4305
6164
1 0.
5973
7441
0315
(6)
0-64
9058
4854
(5)
0.84
1539
59(6
1)
0.96
3031
00(6
) 1-
5996
590(
6)
2-78
0180
(82)
4.
9075
11 (5
)
0.53
6324
6515
4444
8 (9
) 0.
6006
1989
59(6
1)
0-64
4396
19(2
0)
0-79
2897
0(5)
0.
8805
25(7
) 1.
3039
4(5)
2-
0081
1 (4
) 3"
1455
6(62
)
0"55
6718
7518
0(6)
0-
6272
896(
905)
0"
6702
09 (1
7)
0-80
42(4
) 0"
8788
(91)
1.
2164
(78)
1-
7346
(77)
25
11(7
)
0-44
9970
4234
3286
27
0-37
4348
3196
7968
29
0.33
4482
1496
0758
0 0-
2438
2114
3448
0.
2094
2913
92
0-12
205
0.06
9494
0-
0450
0-44
0752
8022
8028
36
0.37
1505
0862
9111
2 0-
3376
6460
0042
0.
2612
87
0-23
173
0.16
0 0.
1051
0.
0791
4
0.41
5297
9506
4051
28
0.34
9980
5931
0.
3208
7514
0.
2576
0.
2324
4 0-
1795
0.
104
0-05
62
e~
e~ E"
"...4
b~
g~ I
,I Z
Tab
le3.
E
ner
gie
san
d(x
2)
valu
esfo
rthe
~o=
0,,:
.= 1
osc
illa
tors
ford
iffe
ren
tr,
s, a
nd
n.
E (
V)a
reth
elo
wer
bo
un
dso
f en
ergy
rep
orte
d by
Vrs
cay
I'2] w
here
in r
epla
cem
ent
of la
st k
dig
its
in t
he e
ntri
es b
y th
e k
digi
ts i
n th
e pa
rent
hesi
s gi
ves
the
uppe
r bo
und.
s r
n K
E
E
(P)
E(V
) (x
2).
2-0
0-0
0 3"
0 0'
6679
8625
9 0"
6679
8625
9155
77
0-66
795(
801)
0"
2873
3756
6761
75
0-66
7986
* 1
3'0
2-39
3644
0163
2"
3936
4401
6483
2"
3934
7(84
) 0"
7156
0507
3742
2"
3936
44*
2 3"
0 4"
6967
9538
4"
6967
9538
694
4"69
61 (7
6)
0"98
7930
2529
44
4"69
6795
* 1"
0 0
1"0
0"48
906
0"48
9063
9276
0"
4890
636(
40)
0"39
2459
0130
1
1"5
2'20
1 2"
2015
4676
6 2"
2009
(28)
0-
7780
4561
3080
4 2
2"5
4"68
2 4"
6825
5850
243
4"67
(71)
0"
9909
3396
317
2"0
0 1"
0 0"
50
0"49
4648
0"
4946
4(5)
0'
3880
2808
008
1 1"
5 2'
3 2'
2879
5366
2"
27(3
1)
0"74
8661
9301
2 2
2"0
5"1
5'13
7332
9972
4"
9(5'
7)
0"90
3213
0578
8
0 7"
5 0"
6804
1 0"
6807
03
0"25
8929
1
8"5
2"58
00
2"57
972
0"60
3946
2
9"5
5"39
4 5"
3943
0"
7601
0
3'0
0'51
8 0'
5196
0'
3392
1
3"5
2"32
2"
333
0"66
78
2 4'
5 5"
1 5"
112
0'80
4 0
1"5
0"51
0"
508
0'34
7 1
2"5
2"3
2"34
1 0"
667
2 3"
0 5"
0 5'
340
0"76
8
3-0
0"0
1.0
2-0
*Ene
rgy
eige
nval
ues
of o
ther
wor
kers
rep
orte
d by
Vrs
cay.
ex
Se!f-interacting one-dimensional oscillators
Perusing through the computational data displayed in the three tables, we arrive at the following conclusions:
(1) In self-interacting oscillators and the infinite-field limit oscillators (tables 1 and 3) wherever comparison with the findings of other workers is possible, our results for E,(P) as well as (x z). are more accurate. This prompts us to infer that other values found by RHPS are also quite reliable for these categories of oscillators. As far as self-interacting anharmonic oscillators with s = 2 (table 2) are concerned, Eo(P ) are an improvement over the results of Vrscay for r = 1"0 for 2 up to 10.0 and for r = 2"0, 3"0 for 2 up to 1-0, in addition to being so for r = 0 for all 2. However, in all other cases, the convergence of Eo(P ) is worse than the two bounds of energy reported by Vrscay. Furthermore, such a comparison is not possible for (x z )o as these values have not been determined by Vrscay. Nonetheless, the extent to which these sums are convergent, they too must be correct. (2) For self-interacting oscillators, the renormalization parameter K is zero for 2 up to 0.1 and 0"5 for 2 values ranging from 0"5 to 10,000. Thus, for very low 2 values even ordinary hypervirial-Pad~ or simple hypervirial method is successful. In the case of anharmonic oscillators whether to is 0 or not, K does not vary significantly or monotonically with 2, r or n. For a particular set of values of s and r, it decreases with increase in n; and for specific values of r and n, K increases if s is increased. (3) If we consider a particular type of oscillator with to 4 0 then the ground state energy increases while (x 2)o decreases with increase in the contribution of the perturbation term, i.e., in the value of 2. For a particular 2, the values of E o and ( x : ) o for r = 0 are, respectively, higher and lower than the corresponding values for r :~ 0 cases though for r = 1, 2, 3 the variations of E o as well as (x 2)0 with r do not exhibit a well defined trend. In the case ofo9 = 0, 2 = 1 oscillators there is no definite trend in the variation of E, and (x 2) . .
3. ARSPT
In the framework of perturbation theory, the interaction
V(~k.) = 2(xZ') x 2~ (11)
is treated as a perturbation and since it depends upon ~O.(x), similar to ¢. and E. this too is expressed as a power series in the perturbation parameter 2. Thus, following Surj/m and ,~ngy/m [1], Surj/m [3] and Kim and Sukhatme [9], we write
~,(x) = O~°~(x)[1 + 2 f ~ ' + U 7 ~ + ...], (12)
E, = E;o' + 2e~" + 22Ep' + ..., (13) and
V(O.) = 2 V~.'~(x) + 2 2 Vp~(x) + . . . . (14)
Here, the wavefunctions are taken to be real because we are concerned with one- dimensional bound state situation with local confining potential [11, 12]. From (11), (12) and (14) we have
V:"(x)= xZ~ f ~., O~.°'"(x')(x')2"dx' (15)
Pramana J. Phys., Vol. 46, No. 3, March 1996 209
Mamta and Vishwamittar
Table 4. A comparison of the ground state energy for the Hamiltonian (2) with r = 1 determined by ARSPT (Eo) and RHPS Eo(P). All entries are in the units correspond- ing to h = m = o ~ = 1.
s = l s = 2
2 0-1 1-0 0'1 l-0
E o 0.5225 0'5000 0-5361 - 0.5547 Eo(P)* 0'5233 0"6624 0'5297 0"6491
*Eo(P) values have been rounded off to four significant figures.
and
S V~2)(x) = 2x 2~ (x')2'O~°)2(x')f~l)(x')dx'. (16) - - G O
Following the cus tomary procedure, we get
E~ 1' f oo ~1) (o,~ = V, (x)~k. (x)dx, (17)
d - o0
and
E~ 2, f ~ (~) (~) [E. f . (x) V(.2)(x) ") ") (°~ . . . . V. ( x ) f . ( x ) ]¢ . (x)dx (18)
with
f~t)(x) = - 2 [dx'/O~°)2(x')] [E~ l ~ - Vn~t)(X,,)]0,~O~ . . . . . (X)dx . 09) - - o 0
With a view to compare the results obta ined by using A R S P T with those of R H P S we have determined ground state energy for the self-interacting oscillators with r = 1, s = 1,2 keeping co = 1. We have for r = 1, s = 1
E o = 0"5 [ 1 + (2/2) - ( ~ . 2 / 2 ) ] (20)
and for r = 1, s = 2
E o = 0.51-1 + (32/4) - (18322/64)]. (21)
The values found for 2 = 0"1 and 1.0 by two methods are compared in table 4. As expected the E o values differ significantly from Eo(P) and the deviation is higher for higher s and ,~.
References
[1] P R Surj/tn and J G Angyfin, Phys. Rev. A28, 45 (1983) [2] E R Vrscay, J. Math. Phys. 29, 901 (1988) [3] P R Surjfin, J. Math. Chem. 8, 151 (1991) 1-4] J G Angy/m a n d P R Surjfin, Phys. Rev. A44, 2188 (1991) [5-1 J Cioslowski, Phys. Rev. A36, 374 (1987)
210 Pramana - J. Phys., VoL 46, No. 3, March 1996
Self-interacting one-dimensional oscillators
[6] J Killingbeck, J. Phys. AI4, 1005 (1981) [7] J Killingbeck, J. Phys. A20, 601 (1987) [8] J Killingbeck, Phys. Lett. A132, 223 (1988) [9] I M Kim and U P Sukhatme, J. Phys. A25, 647 (1992)
[10] C K Au and Y Aharonov, Phys. Rev. A20, 2245 (1979) [11] L D Landau and E M Lifshitz, Quantum mechanics-non-relativistic theory (Oxford,
Pergamon, 1965) [12] M Razavy and A Pimple, Phys. Rep. 168, 307 (1988)
Pramana - J. Phys., Vol. 46, No. 3, March 1996 21 !