This action might not be possible to undo. Are you sure you want to continue?
IEEE JOURNAL OF QUANTUM ELECTRONICS, VOL. 28 NO 2 , FEBRUARY 1992
Accurate Solutions to Schrodinger’s Equation Using Modified Airy Functions
A. K . Ghatak, R. L. Gallawa, and I. C. Goyal
Abstract-A formalism that utilizes the Airy functions is applied to Schrodinger’s equation for a spherically symmetric potential. We show that the computational procedure is very simple and allows us to have a very accurate description of boundstate wave functions and the corresponding eigenvalues.
rate eigenvalues. Thus we describe a simple method that allows an accurate description of the wave function as well as of the eigenvalue. In comparing our MAF solution to the WKB solution for the wave function, we will use the first-order solution in each case. 11. THEORY For a spherically symmetric potential V(r) the radial part of the wave function R(r) satisfies the equation
I. INTRODUCTION T H E solution of the eigenvalue equation
+ f(r))\k(r) = X\k(r)
is a subject of profound importance in many areas of physics and engineering, the most important being quantum mechanics and optical waveguide theory [l], . When (1) is not amenable to exact solutions, one usually resorts to one of the three approximate methods: perturbation theory, the variational method, and the WKB approximation . Perturbation theory is based on a closely related problem that is capable of exact solution; one usually uses first-order perturbation theory. Even then, it is extremely difficult to calculate the eigenfunctions as it would involve the summation over an infinite series. On the other hand, the variational method can give a good estimate of the lowest order mode by choosing an appropriate trial function and carrying out an optimization; the method becomes quite cumbersome when applied to higher order modes. Finally, the WKB method is applicable to “slowly varying” forms of f ( r ) , but the corresponding eigenfunctions are inaccurate; they blow up at the turning points. In a recent paper  we have shown that a direct solution of (1) based on Airy (Ai) functions gives an accurate description of the exact solution. The solution can be very easily obtained on a personal computer. In this paper, we apply this method to the Schrodinger equation for a spherically symmetric potential. We show that the computational procedure is very simple and allows a very accurate description of all bound-state wave functions corresponding to 1 = 0. We also show here that the eigenvalues can be determined from the tabulated zeros of the Airy function with as much ease as the WKB method affords. We go further, using a first-order perturbation theory [SI (using the present modified Airy function (MAF) solutions), allowing us to obtain very accuManuscript received February 4,1991; revised July 10, 1991. This work was supported by the U.S. Department of Commerce. A. K. Ghatak and I. C. Goyal are with the Physics Department, Indian Institute of Technology, New Delhi 110016, India. R. L. Gallawa is with the Electromagnetic Technology Division, National Institute of Standards and Technology, Boulder, CO 80303. IEEE Log Number 9105195.
where E is the energy eigenvalue and the other symbols have their usual meaning . If we make the transformation
and consider the 1 = 0 case, we readily obtain
d2u dr *
+ r2(r)u(r)= o
Following  we obtain the following MAF solution to (4):
and r = ro represents the turning point [i.e., r 2 ( r 0 )= 01. In writing (6) we have neglected the Bi solution, which diverges at r = 00. The quantity 4 [as defined through (7)] is negative for r < ro (where r 2 ( r ) is positive) and positive for r > ro (where r 2 ( r )is negative). Thus
l(r) = 0 1992 IEEE
r ( r ) dry’’
< r < ro
[see (20)] allows us to obtain very accurate eigenvalues. we get E = 2.886341.64. VOL. V(r) = -v0 ePrla.00160700 19.02265085 10. The eigenvalues are therefore obtained from the solution of the following transcendental equation Ai (((0)) = 0.00244300 13. p = 8. .00295498 11. (21) (16) The exact eigenvalues are obtained by solving the following transcendental equation : JJg) = 0 where (22) v=(?) Clearly. we get the following expression for the first-order correction A E to the energy eigenvalue where which is tabulated in the third column of Table I.78670809 7. which agrees extremely well with the exact result. The reason is seen from the following. FEBRUARY 1992 40 1 and 2 13 r > ro where (9) 1 2 3 4 5 6 7 8 9 10 TABLE I THEFIRST TENZEROS OF THE AIRY FUNCTION (Ai) AND ASSOCIATED CONSTANTS K2(r) = -r2(r) is positive in the region r (10) (1 1) > ro. We can carry out elementary manipulations to obtain (19) Comparing (18) and (4).94413359 9.409 X m. i. (Using ZI = 0. Vo = 40 MeV = 6. 5 . -Zen represent the nth zero of the Airy function.00208210 15. are tabulated in the second column of Table Values of Z I and have been taken from . As we now show. NUMERICAL EXAMPLE We consider the profile and the corresponding WKB quantization condition r(x) dr simplifies to = (n ++ IZ = 1 .00789946 5.00 144240 (12) (13) where NB: Ai (-Zen) = 0 .85873. We note that the solution given by (6) is an exact solution of the following differential equation: d2u - 2 112 g = ( pV&? k ) 1/2 . Recall that for antisymmetric modes corresponding to a symmetric one-dimensional potential. Notice that the numbers are nearly odd integers. dr2 + (r2(r) + F(r))u(r) = 0 (18) If we now solve (22).52055983 6. The corresponding WKB value is given by ZIWKB = 0.84 % .2% .01734924 3. the use of first-order perturbation theory.398 X J. using the MAF solution.e.865815. 2.338 10741 4. (14) and (17) are similar and can be solved with equal ease. which represents the correct binding energy of the deuteron. 28 NO.) The corresponding value of U obtained from (14) is 0. 3. and considering the last term in (18) as perturbation.") = 0.00373725 9. 111.0018 1400 17. Because of (3).08794944 5.858628.IEEE JOURNAL OF QUANTUM ELECTRONICS.04017434 11.85 x kg ( p represents the reduced mass) giving g = 3.223 MeV.00507887 7. for this example. . a = 1. (23) For the deuteron problem. whereas the WKB quantization condition gives an error of 3. 2.93601556 12.. In fact. Let Ai (-2. The perturbation theory [see (20)] yields ZI = 0.858628. our (14) gives an error in the eigenvalue of 0.00852430 11. u(0) = 0. Y(0) = 0 AE= -- A2 2cc sa 0 F(r)u2(r)dr (20) where u(r) is given by (6). we get ZI = 0.82877675 1.
and WKB wave functions for the exponential potential profile. MAF.84%. however. Not surprisingly.402 IEEE JOURNAL OF QUANTUM ELECTRONICS.2% (WKB) versus 0. but additional effort is required because the various solutions must be pieced together to form a wave function that is continuous and has a continuous derivative. the solution is given by  where I . for the Coulomb potential the eigenvalues found from the MAF method for the first two to three modes are in considerable error. This may be due to the singularity of the potential at Y = 0. The difference between the exact and the MAF wave functions of Fig. Unlike the WKB solutions. 1. We chose the exponential potential function here because it has an exact solution. Unfortunately. This method is therefore valid. FEBRUARY 1991 TABLE I1 COMPARISON OF EIGENVALUES FOR THE DEUTERON PROBLEM 0. If the zero is of higher order. CONCLUSIONS We have shown that the MAF solution gives extremely accurate wavefunctions for problems of practical interest with defined by (7) and with ro replaced by the turning point in that region. when it is piecewise continuous. The error in the eigenvalue is only 0. In quantum mechanical and optical waveguide problems. 2 . IV.64. The exact. subject to the constraint that r 2 ( r )should have a zero of order not greater than one at the turning point. The MAF and the WKB solutions both use first-order approximations. but the MAF solution agrees well with the exact solution throughout the region of interest. The multiplying constants are chosen to ensure the continuity of U ( Y ) and u ' ( Y ) at the boundaries of the continuous region. the WKB solution fails at the turning point. and will work with any potential function.032 2 4 p ( = r/a) 6 8 1 0 Fig.012% for the example we used (Table 11). These results. For g = 3. when dealing with a nonuniform waveguide) is not smooth. In that case.858628 1. the MAF solutions hold throughout the region of interest. The method is practical even when the potential function (or refractive index profile. there is only one bound state. see Table 11. the method becomes more complex but not more difficult. r2normally has a first-order zero at the turning points. In that case. normalized to the exact value. For the example we have chosen. The improvement in the eigenvalue. 4'(r) will also be zero at the turning point and the solution given in (6) will not be valid. and should be of use in many cases. . In this regard. the solution is taken as Fig. 28 NO. that is. 1. We have also given a simple expression for the eigenvalues. are given in tabular form in Table 11. even in the case of piecewise continuous potential functions. given by (20). This is analogous to the method used when exact solutions are known for each of the piecewise continuous regions. is very impressive.00012 1.0084 1. Fig. In general. 2. 1 shows the exact wave function along with the WKB and MAF solutions . our accuracy is considerably better than that of the WKB method. The method is quite general. making it possible to test our method. We have carried out a similar analysis of the three-dimensional harmonic oscillator and have found excellent agreement between the MAF solution and the exact wave functions. herein lies the utility and practicality of the solution described here. (14) and (17))./ 2: 4 6 8 1 0 p(=r/a) (25) and v is the order of the zero. In each continuous region. we also show in Fig. 2 the difference between the exact and the MAF solutions. which gives accuracy of about the same order as the WKB method (cf. where the error is seen to be 3. including the turning point. VOL. if there are more bound states the MAF solutions will be more accurate for higher order states. Because of the good agreement.
IEEE JOURNAL OF QUANTUM ELECTRONICS. Abramowitz and I. L.” J . 1989. E.S. Ghatak. Ghatak. “An approximate solution to the wave equation-Revisited. E. K.  R. 3 3 . K. and R. L. A. vol. A. Schiff. 1991. C. Quantum Mechanics. Waves Apvol. VOL. . “On the asymptotic solutions of ordinary differential equations. Washington DC: U. Quantum Mechanics. Applied Mathematics Series 55. ourthanks to M. of Stand.  A. Amer. and to Dr. 1991. pp. B. Lokanathan.. 16. FEBRUARY 1992 403 ACKNOWLEDGMENTS assistance. Lett. 23-64. India: Macmillan. 1984. pp. Electromagn. [ 5 ] I. with an application to the Bessel functions of large order. Cambridge. Pal provided constructive criticism. 1964.” Opt. 623-636. UK: Cambridge Univ.. Ghatak and K.  M. 30-32. “Methods for analyzing planar optical waveguides. Stegun.Depart. Soc. Goyal. New York: McGraw-Hill. of Commerce. and A. Thyagarajan. Nat. C. Langer. Math.  I. New Delhi. Press. Ghatak and S. Bur. ~~n~~~for her REFERENCES [ l ] L. 28 NO.  A. Gallawa. Handbook of Mathematical Functions with Formulas Graphs and Mathematical Tables. Optical Electronics. 1931. Johnson and Dr.” Trans. Gallawa. Goyal. 2. 1968. K. pp. R.. 5. vol. I. K.
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue reading from where you left off, or restart the preview.