Lecture on-line Variation Theory (PowerPoint) Variation Theory (PDF) Handout for this lecture
W > E0
if
f 0
f=
i=1
cii
f * fdv1 = 1
and = f * Hfdv or with
= ci * i
i =1
= ci * i * H c j j dv
i =1 j =1
Now since Hk = E k k
e have = ci * c j E j i * j dv
i =1 j =1
> | c0 |2 Eo + | c1 |2 Eo + | c2 |2 Eo + .. | ci |2 Eo + .. | cn |2 Eo ..
r > Eo | ci |2 = E o
i =1
sin ce : f *fdv = ( ci i )* ( ck k ) =| ci |2 = 1
i k
THE VARIATIONAL METHOD " T he S c h em e " : Let us suppose that we select the trial function f() to be one which contains several parameters 1,2,...n .Then the corresponding energy df*(1,2,...n,)Hf(1,2,...n,) I = df*(1,2,...n,)f(1,2,...n,) Will depend on 1,2,...n
We know that I > Eo , where Eo is the energy corresponding to the exact wavefunction o. We now select 12...n so that I is minimized: se the method of elementary calculus for minimizing everal variables:
S i mu l t a n eo u s l i ne ar e q u a t i on s Consider a set of linear equations a11x1 + a12x2+ ................a1nxn = a21x1 + a22x2+ ................a2nxn =
b1 b1
. . . . . . . . . . . . . . . . . . . . . . an1x1 + an2x2+ .................annxn = bn where the coefficients aij and bi are known and {xi;i=1,n} the set of unknowns The k'th solution is given according to Cramer's rules by Bk xk = A
where A is the secular determinant formed from the set of coefficients of the equation a11 a12 a21 a22 A= a13 .......a1k ...... a1n a23........a2k ....... ann
......................... . ak1 ak2 ak3........akk ....... akn ............................ an1 an2 an3........ank ....... ann
Further the determinant Bk is formed by replacing the k'th column in A by the column b1,b2,....,bk,..bn
......................... . ak1 ak2 ak3........bk ....... akn ............................ an1 an2 an3........bn ....... ann
A special case is the set of homogeneous equations in which the set b1,b2 etc all are zero a11x1 + a12x2+ ................a1nxn = 0 a21x1 + a22x2+ ................a2nxn = 0 . . . . . . . . . . . . . . . . . . . . . . an1x1 + an2x2+ .................annxn = 0 In this case all the determinants Bk are zero and we have 0 xk = A If A is different from zero than we have the set of trivial solutions x1 = x2= .....xk = ... = 0
If on the other hand A=0 o Then xk = o = ? It turns out that in this case the set of homogenous equations might have non-trivial solutions.However it is clear that they can not be found using Cramers rule Assume now that we have such a set of solutions {e1,e2,...,ek...en} such that a11e1 + a12e2+ ................a1nen = a21e1 + a22e2+ ................a2nen = 0 0
It is clear by multiplying each equation by c a11ce1 + a12ce2+ ................a1ncen = 0 a21ce1 + a22ce2+ ................a2ncen = . . . . . . . . . . . . . . . . . . . . . . an1ce1 + an2ce2+ .................anncen = that {cek ;k=1,n} is also a set of solutions where c is any number. In fact we can give any value to one of the x's,say xn. 0 0
Suppose that we give xn the value co If {e1,e2,...,ek,....en} is a set of solutions so is co co co co { ene1, ene2,......., enek,........, enen } thus co co co { ene1, ene2,......., enek,........,co } is the solution set for which xn = c0. How do we find the remaining set {x1,x2,...,xk,..xn-1}
Substitute into the set of equations a11x1 + a12x2+ ................a1nxn = 0 a21x1 + a22x2+ ................a2nxn = 0 . . . . . . . . . . . . . . . . . . . . . . an1x1 + an2x2+ .................annxn = 0 the solution xn= co.Thus, a11x1 + a12x2+ ................a1nco = 0
a11x1 + a12x2+ ................ = -a1nco a21x1 + a22x2+ ................ = -a2nco . . . . . . . . . . . . . . . . . . . . . . an1x1 + an2x2+ ................. = -annco We now have n-equations from which to determine n-1 unknown {x1,x2,....,xn-1}. Let us take the first n-1 equations a11x1 + a12x2+ ................+a1n-1xn-1 = -a1nco a21x1 + a22x2+ ................+ a2n-1xn-1 = -a2nco . . . . . . . . . . . . . . . . . . . . . . an-11x1 + an-12x2+ .......... an-1n-1xn-1 = -an-1nco we can now use Cramer's rule to solve this set of equations .
2. Understand the general procedure for solving linear equations. Understand what a secular determinant is and why it must be zero for homogeneous linear equations