Scribd
Upload
30 views11 pages

Lab1 Notas

Uploaded by

Samuel Días
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
30 views11 pages

Lab1 Notas

Uploaded by

Samuel Días
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 11
COMPLEMENT Bs, Complement By REVIEW OF SOME USEFUL PROPERTIES OF LINEAR OPERATORS 1, Trace of an onerator a. Definition The srace is invariant ©. Important properties 2, Commutator algebra a. Definition . Properties 3. Restriction of an operator to a subspace 44. Functions of operators 8. Definition; simple properties ’. An important example: the potential operator € Commutarors involving functions of operators ‘5. Differentiation of an operator 8, Definition b. Differentiation rules ©. Examples 4. An application: a useful formula The aim of this complement is to review a certain number of definitions and useful properties of linear operators. 1. Trace of an operator a. DEFINITION The trace of an operator 4, written Tr A, is the sum of its diagonal matrix elements. When a discrete orthonormal basis, {| u; > }. is chosen for the space &, one has, by definition’ TA = TC lAluy> 0 For the case of a continuous orthonormal basis { | w, > }, one has: Tea = fae cw latns> (2) When & is an infinite-dimensional space, the trace of the operator A is defined only if expressions (1) and (2) converge. 166 REVIEW OF LINEAR OPERATORS b. THE TRACE IS INVARIANT The sum of the diagonal elements of the matrix which represents an opera- tor A in an arbitrary basis does not depend on this basis. Let us derive this property for the case of a change from one discrete ortho- normal basis {| u, > } to another discrete orthonormal basis { | 4 > }. We have: [zla>PaiJaim> @ (where we have used the closure relation for the |, > states). The right-hand side of (3) is equal toz Zduler @ (since it is possible to change the order of two numbers in a product). We can then replace ¥| u; > < u; | in (4) by (closure relation for the | u, > states), and we obtain, finally: Tiulalur=Ldhlale> ©) We have therefore demonstrated the property of invariance for this case. COMMENT: If the operator 4 is an observable, Tr A can therefore be calculated in a basis of eigenvectors of 4. The diagonal matrix elements are then the eigen- values a, of A (degree of degeneracy g,) and the trace can be written: TrA=Yonay 6 IMPORTANT PROPERTIES TrAB =TrBA (7a) Tr ABC = Tr BCA = Tr CAB (7b) In general, the trace of the product of any number of operators is invariant when a cyclic permutation is performed on these operators. Let us prove, for example, relation (7-a): Tr AB = = FC] Aly) > Cail Ald =F 6a] BA | = Tr BA ®) (twice using the closure relation on the { | u; ) } basis). Relation (7-a)is thus proved; its generalization (7-b) presents no difficulty. 167 Come Tawnovon, ~ Quantim mechani. — > COMPLEMENT By, 2. Commutator algebra DEFINITION The commutator [4, B] of two operators is, by definition: [4,8] = 4B - BA 0) b. PROPERTIES [4, 8] = ~ [B, 4] (10) [4, (B + ©] = [4,8] + [4,€] (ay [4, BC] = [4, B]c + BLA, C] (2) [4 [2.c]] + [B, [¢. 41] + [¢, [4, 8] = 0 (3) [4 B]' = [B44] (4) The derivation of these properties is straightforward : it suffices to compare both sides of each equation after having written them out explicitly. 3. Restriction of an operator to a subspace Let P, be the projector onto the g-dimensional subspace &, spanned by the q orthonormal vectors | 9; ): a= Sled is an arbitrary ket, it follows from this definition that: Ay|W> = PA b> an where \¥>=Palv> 3 (18) ‘] Consequently, to make A, act on an arbitrary ket Tw >, one begins by projecting this ket onto 6; then one'lets the operator A act on this projection, retaining only the projection in 6, of the resulting ket. The operator 4,, which transforms any ket of &, into a ket belonging to this same subspace, is therefore an operator whose action has been restricted to &,. What can be said about the matrix which represents 4,? Let us choose a basis, { | 4 > } whose first g vectors belong to &, (they are, for example, the | >), the others belonging to the supplementary subspace. We have: = Cui | PeAPy | uy > a9) 168 that is: =
    ithjsa 0 if one of the two indices i or greater than g (20) Therefore, the matrix which represents 4, is, as it were, “elldut” of the one which represents A. One retains only the matrix elements of A associated with basis vectors | u, ) and | u, >, both belonging to &,, the other matrix elements being replaced by zeros. 4. Functions of operators 2. DEFINITION; SIMPLE PROPERTIES Consider an arbitrary linear operator A. It is not difficult to define the opera- tor A": it is the operator which corresponds to n successive applications of the operator A. The definition of the operator 4™', the inverse of A, is also well known A-* is the operator (if it exists) which satisfies the relations AA = AA =T (21) How can we define, in a more general way, an arbitrary function of an opera- tor? To do this, let us consider a function F of a variable z. Assume that, in a certain domain, F can be expanded in a power series in z Fe) = 5 fet 22) By definition, the corresponding function of the operator 4 is the operator F(A) defined by a series which has the same coefficients f,: F(A) = Y fA" (23) a For example, the operator e4 is defined by: ea z4 DHA A + + AMAL + 4 We shall not consider the problems concerning the convergence of the series (23), which depends on the eigenvalues of 4 and on the radius of convergence of t ‘Note that if F(z) is a real function, the coefficients f, are real. If, moreover, Ais Hermitian, we see from (23) that F(A) is Hermitian, Let | @, > be an eigenvector of A with eigenvalue a: Ala.>=a|o.> es) Applying the operator n times in succession, we obtain : A" | G4) = a"| > (2) 169 COMPLEMENT 8, ‘Now let us apply series (23) to | g, >; we obtain Fayle.> = ¥ hatloe>= Fla)l en en ‘This leads to the following rule : when | @, > is an eigenvector of A with the eigen- value a, |g, > is also an eigenvector of F(A), with the eigenvalue F(a). This property Ieads to a second definition of a function of an operator. Let us consider a diagonalizable operator A (this is always the case if A is an observable), and let us choose a basis where the matrix associated with A is actually diagonal (its elements are then the eigenvalues a, of 4). F(A) is, by definition, the operator which is represented, in this same basis, by the diagonal matrix whose elements are F(a,). For example, if o, is the matrix 1 0 ( _‘) (28) it follows directly that: e 0 (6 ww) 1) COMMENT: Care must be taken, when functions of operators are used, with respect to the order of the operators. For example, the operators e4e® e#e4, and e4*# are not, in general, equal when A and B are operators and not numbers. Consider : a x pT = 30) rae en 6 When A and B are arbitrary, the right-hand sides of (30), (31) and (32) have no reason to be equal (see exercise 7 of complement H,,). However, when A and B commute, we have : [4, B] = 0 => ete? = ofet = of*? 3) (a relation which is obvious, moreover, if the diagonal matrices which repre- (Fy 4 sent e4 and e# are considered in a basis of eigenvectors common to A and B), 170 b. AN IMPORTANT EXAMPLE: THE POTENTIAL OPERATOR In one-dimensional problems, we shall often have to consider “potential” operators V(X) (so called because they correspond to the classical potential energy V(x) of a particle placed in a force field), where V(X) is a function of the position operator X. It follows from the preceding section that V(X) has as eigenvectors the eigenvectors | x > of X, and we have simply: VX) |x> = VOx)Ix> (G34) The matrix elements of V(X) in the { | x > } representation are therefore = V(Qx)d(x — x’) a (35) Applying (34) and using the fact that V(X) is Hermitian (the function V(x) is real), we obtain : 6x1 MOIY> = Me cxly> = Mow E) (36) This equation shows that in the { | x) } representation, the action of the opera- tor V(X) is simply multiplication by V(x). The generalization of (34), (35) and (36) to three-dimensional problems can be performed without difficulty; in this case, we obtain : VR) |r > V(r) |r > (37) = Mir)(e - ¥’) (38) = Peru) (G9) © COMMUTATORS INVOLVING FUNCTIONS OF OPERATORS Definition (23) shows that A commutes with every function of A: (4, F(4)] = 0 (40) Similarly, if A and B commute, so do F(A) and B: [B, 4] = 0 = [B, F(4)] =0 (41) What will be the commutator of an operator with a function of another operator which does not commute with it? We shall restrict ourselves here to the case of the X and P operators, whose commutator is equal to : [YP] =a (42) Using relation (12), we can calculate: Lx, P?] = [x PP] = Lx, PP + PLY, P] = 20 P (43) More generally, let us show that: LX, P*] = ihnP! (44) If we assume that this equation is verified, we obtain: [x PY!) = [PP] = [x PP + PL, = ihP* + ihnPPY! = itn + (45) COMPLEMENT By Relation (44) is therefore established by recurrence. ‘Now let us calculate the commutator [X, F(P)]: (x, Fy] = DEX, 4 Pe =D ihing Prot (46) 0 Lo If F'(z) denotes the derivative of the function F(z), we reco} of the operator F’(P). Therefore: [X%. AP) = mh FUP) (47) ‘An analogous argument would have enabled us to obtain the symmetric relation: in (46) the definition [2,600] = - nox (48) COMMENTS: (i) The preceding argument is based on the fact that F(P) (or G(X)) depends only on P (or on X), It is more difficult to calculate a commutator such as Lx, ©(X, P)], where (x, P) is an operator which depends on both X and P: the difficulties arise from the fact that X and P do not commute. (i) Equations (47) and (48) can be generalized to the case of two operators and B which both commute with their commutator. An argument modeled on the preceding one shows that, if we have [4c] = [B,c]=0 (49) we c=[4,8] o (50) [4, #(B)] = [A, 81 F'(4) (1) 5. Differentiation of an operator DEFINITION Let able an operator which depends on an arbitrary variable 1. By defi- nition, the derivative Bor alo with respect to / is given by the limit (if it exists): A jig Alt + 41) = AW ae At 7 (62) The matrix elements of A(t) in an arbitrary basis of independent vectors |u, > are functions of ¢ = Ayko ° (53) m nevew oF uncan ovenarons @ (dA Let us call (2) = = ECL Fae > Cul Ely > (57) We have seen that the matrix elements of d(FG)/ds are the derivatives with respect to 1 of those of (FG), Thus we have, differentiating the right-hand side of (57): d aF cml IH) =F [Cw IGE la >
      + dG + cml elu?
        and | ¥ > of &, and their transforms |D, > and | J, > under the action of U: Wi >=Ulv> Iv2> = UlW2> (2) Let us calculate the scalar product < J, | J, >; we obtain: 1102.> = = Cr be? @) The unitary transformation associated with the operator U therefore conserves the scalar product (and, consequently, the norm) in &. When @ is finite-dimensional, moreover, this property is characteristic of a unitary operator. COMMENTS: (i) If 4 is a Hermitian operator, the operator T = e'4 is unitary, since: T'= @ and therefore: TT = eet a4 TT' = ele" 4 = (5) (obviously, — id commutes with iA). 176

You might also like