Scribd
Upload
425 views23 pages

Chem 373 - Lecture 30: Linear Variation Theory

The lecture discusses the linear variation method for approximating the ground state energy and wavefunction of a Hamiltonian. The trial wavefunction is written as a linear combination of known functions with coefficients to be optimized. Taking the derivative of the energy expression with respect to the coefficients and setting it to zero produces a set of linear equations. The secular determinant of the equations must equal zero, yielding a polynomial whose lowest root provides an approximation to the ground state energy. The coefficients can then be determined by back-substituting into the equations.

Uploaded by

Nuansak3
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
425 views23 pages

Chem 373 - Lecture 30: Linear Variation Theory

The lecture discusses the linear variation method for approximating the ground state energy and wavefunction of a Hamiltonian. The trial wavefunction is written as a linear combination of known functions with coefficients to be optimized. Taking the derivative of the energy expression with respect to the coefficients and setting it to zero produces a set of linear equations. The secular determinant of the equations must equal zero, yielding a polynomial whose lowest root provides an approximation to the ground state energy. The coefficients can then be determined by back-substituting into the equations.

Uploaded by

Nuansak3
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 23

Lecture 30: Linear Variation Theory The material in this lecture covers the following in Atkins. 14.

7 Heteronuclear diatomic molecules (c) The variation principle

Lecture on-line Linear Variation Theory (PowerPoint) Linear Variation Theory (PDF) Handout for this lecture

The linear variation method The Linear Variation Method ^ We have a Hamiltonian H with the eigenfunctions n
nd eigenvalues En given by the SWE ^ Hn = Enn Let us look at the groundstate 1 with the energy E1 . We would like to find a wavefunction 1 nergy for which the

= H 1d / 1d 1 1
is close to E1

The linear variation method


We write 1 as a t r i a l wavefunction in terms of a inear combination of known functions {fi } hat depends on the same variables as 1 and have the ame boundary conditions j=n 1 =

Cjfj

j=1

Or 1 = C1f1 +C2f2 +....+Cjfj + Cnfn

The linear variation method

We shall now vary all the coefficients {Cj ,j=1,n} in such a way that W has the smallest possible alue. That is ,we shall find the absolute minima of the unction W(C1,C2,C3,.....,Ci,..Cn). et the values of the coefficients {C1,C2,C3,.....,Ci,..Cn} t the minimum be given by

1 1 1 1 1 C1 ,C2,C3, ...,Cj ,..Cn }

The linear variation method


What do we know about this particular coefficients ? We know that if we look at the derivatives of W

W = W(C1,C2,.....,Cn) Ci
1 1 1 1 1 Then W(C1 ,C2,C3, ...,Cj ,..Cn ) = 0 1 1 1 1 1 We shall use this fact to find { C1 ,C2,C3, ...,Cj ,..Cn}

The linear variation method


irst let us substitute the expression for 1 into the xpression for W. The denomenator of I1 is j=n k=n Cj*fj* Ckfk j=1 k=1

1 1 d =

We shall assume that all functions {fj j=1,n} are real and hat all coefficients {Cj=1,n} are real

The linear variation method


After multiplication of the two parantheses j=n k=n 1 d = Cj Ck fjfkd 1 j=1 k=1

Let us introduce : fjfkd = Sjk

Thus :

1 1 d = Cj Ck Sjk j=1 k=1

j=n k=n

The linear variation method


For the numerator in the expression for I1 we have k=n j=n ^ ^ Cj*fj* H Ckfk 1 H 1d = j=1 k=1 Or after multiplication of paranthesis

^ H 1d = = Cj Ck fj H fkd 1 ^ j=1 k=1

j=n k=n

The linear variation method


We shall now introduce

^ fj H fkd =

Hjk

^ We know that H is Hermetian


^ fj* H fkd =

^ fk (H fj)*d (1)

^ ^ We shall also assume that it is real H = (H )*. hus since {fi=1,n} are real functions it follows from (1) Hjk = Hkj

The linear variation method


We have j=n k=n

^ 1 H 1d = = Cj Ck Hkj j=1 k=1

j=n k=n

Cj Ck Hkj

j=1 k=1 We can now write W = j=n k=n

Cj Ck Skj

j=1 k=1

The linear variation method


Or j=n k=n j=n k=n W Cj Ck Skj = Cj Ck Hkj j=1 k=1 j=1 k=1 It is important to observe that (C1,C2,..,Ci,.. Cn) are ndependent variables We shall now differentiate with respect to one of hem,say Ci ,on both sides of the equation.

The linear variation method


We have j=n k=n j=n k=n W + W Cj Ck Skj = Cj Ck Skj Ci Ci j=1 k=1 j=1 k=1
j= n k = n C j Ck H kj Ci j=1 k =1

j=n k=n Let us now look at Ci Cj Ck Skj = j=1 k=1

The linear variation method


ince we differentiate a sum by differentiating ach term from rules for differentiating a product j=n k=n ( Cj Ck Skj ) = Ci j=1 k=1 j=n k=n [Cj Ck Skj +Ck Cj Skj] Ci Ci j=1 k=1 Since (C1,C2,..,Ci,.. Cn) are independent variables Cj Ci Cj Ci Cj Ci = ij

= 0 if ij

= 1 if i=j

The linear variation method


k=n j=n j=n k=n [jiCk Skj +kiCj Skj] = CkSik + CjSjk j=1 k=1 k=1 j=1 k=n

k=n CkSik +

k=n CkSki = 2

CkSik

k=1

k=1

k=1

Since : Sik =

fi fk d =

fk fi d = Ski

k=n j=n k=n hus Ci Cj Ck Skj = 2 CkSik k=1 j=1 k=1

The linear variation method

Now by replacing Skj with Hkj k=n j=n k=n Cj Ck Hkj = 2 CkHik Ci k=1 j=1 k=1 j=n k=n W hus from: Cj Ck Skj Ci j=1 k=1 j=n k=n j=n k=n WCi Cj Ck Skj = WCi Cj CkH kj j=1 k=1 j=1 k=1

The linear variation method we get by substitution


k=n j=n k=n W Cj Ck Skj + 2 W CkSik = Ci j=1 k=1 k=1 k=n 2

CkHik

k=1 his equation is satisfied for all {Cj ,j=1,n}. The optimal et for which W is at a minimum must in addition satisfy

W =0 Ci

The linear variation method


hey must thus satisfy k=n k=n 2 W CkSik = 2 CkHik k=1 k=1 k=n Or by combining terms: written out C1 [ Hi1 2WSi1 ]+C2 [ Hi2 2WSi2 ] ,.Cn [ Hin 2WSin ] 0

Ck [ Hik 2WSik ] = 0

k=1

The linear variation method


We have obtained this set of equations from

W =0 Ci
However i= 1,2,3,....,n We can as a consequence obtain the set of n equations k=n

Ck [ Hik -WSik ] = 0

i=1,n

k=1

The linear variation method


C1[ H11 -WS11 ]+C2 [ H12 - WS12 ] ,...Cn [ H1n - WS1n ] 0 C1[ H21 - WS21 ]+C2 [ H22 - WS22 ] ,...Cn [ H2n - WS2n ] 0 C1[ H31 -WS31 ]+C2 [ H32 -WS32 ] ,...Cn [ H3n -WS3n ] = ...... C1 [ Hi1 -WSi1 ] +C2 [ Hi2 -WSi2 ] ,...Cn [ Hin -WSin ] = 0 C1 [ Hn1 -WSn1 ] +C2 [ Hn2 -WSn2 ] ,...Cn [ Hnn -WSnn ] 0 his is a set of n homogeneous equations

The linear variation method


his set of equations has only non-trivial solutions rovided that the secular determinant [ H11 -WS11 ] [ H12 - WS12 ] [ H1n - WS1n ] H21 - WS21 ] [ H22 - WS22 ] [ H31 -WS31 ] ...... [ Hi1 -WSi1 ] ... [ Hn1 -WSn1 ] [ Hn2 -WSn2 ] [ Hnn -WSnn ] [ H32 -WS32 ] [ Hi2 -WSi2 ] [ H2n - WS2n ] [ H3n -WS3n ] [ Hin -WSin ] =0

The linear variation method By expanding out the determinant we will obtain an norder polynomial in W.This polynomial has n-roots
I En I
[n]

[1] [2] [3] [i] [n] I1 <I1 < I1 < .....,<I1 <..I1 [1] The lowest root I1 s the approximation to the actual groundstate energy E1

1
[i]

1
[3]

Ei E3 E E1

1
[2]

I I

1
[1]

The linear variation method


[i] [i] urther I1 is an approximation to Ei. In all cases I1 arger than or equal Ei. Having obtained the roots we an now find {Ci,i=1,n) by substituting into the set of quations
1[ H11 -WS11 ]+C2 [ H12 - WS12 ] ,...Cn [ H1n - WS1n ] = 0 1[ H21 - WS21 ]+C2 [ H22 - WS22 ] ,...Cn [ H2n - WS2n ] = 0 . 1 [ Hi1 -WSi1 ] +C2 [ Hi2 -WSi2 ] ,...Cn [ Hin -WSin ] = 0 ... 1 [ Hn1 -I1Sn1 ] +C2 [ Hn2 -I1Sn2 ] ,...Cn [ Hnn -I1Snn ] = 0

[i] where W = I1

i=1,n

What you should learn from this lecture 1. You should understand that in linear variation theory the trial wavefunction is written as a linear combination of KNOWN functions where the relative contribution from each function is optimized.

2. You should know how the set of linear equations are generated and why the secular determinant must be zero and how this is used to determine the energies.

3. You should be able to derive the equation in the case where n = 2

You might also like