Received: 17 November 2015 | Revised: 9 November 2016 | Accepted: 31 December 2016 DOI: 10.1002/cmr.a.21385
RESEARCH ARTICLE
Wigner active and passive rotation matrices applied to NMR tensor
Pascal P. Man
Sorbonne Universités, UPMC Univ Paris 06, CNRS, FR2482, Institut des Abstract matériaux de Paris-Centre, F-75005 Paris, NMR Hamiltonians are double contraction of two spherical rank-2 tensors, their France space parts are represented by a spherical tensor and their spin parts are composed Correspondence of spherical tensor operators. The comprehension of modern NMR experiments is Sorbonne Universités, UPMC Univ Paris very often based on the rotation of these tensors. We present the active and pas- 06, CNRS, FR2482, Institut des matériaux sive rotations in a progressive way from position vector to spherical tensor opera- de Paris-Centre, F-75005 Paris, France Email: [email protected] tor via space function, spherical harmonic, and vector operator. The passive rotation of a physical quantity is described by the rotation of coordinate system. Both the left- and right-handed rotation conventions are applied whereas the right-handed rotation convention is mainly used in the literature. Throughout the article, we explore the equivalence between the active rotation of a physical quan- tity in one direction and the rotation of the coordinate system in the opposite direction. The article presents redundant mathematical demonstrations between active and passive rotations, but they clarify the meanings of some important expressions not well developed in the literature.
KEYWORDS active rotation, canonical transformation, Euler angles, passive rotation, Wigner rotation matrix
1 | INTRODUCTION In this article, we present the active and passive rota- tions in a progressive way from position vector to spherical NMR Hamiltonians may be defined with Cartesian or tensor operator via space function, spherical harmonic, and spherical tensors.1 In the latter case, they are double con- vector operator. The passive rotation of a physical quantity tractions of two spherical rank-2 tensors, one representing is described by the rotation of coordinate system. the space part and the other the spin part. Very often, the Section 2 describes two equivalent points of view for space part has to be expressed in several coordinate sys- the rotation of axes by an angle h in a plane of 3-D Eucli- tems (principal-axis system, MAS rotor system, laboratory- dean space using left- and right-handed conventions. axis system), whereas the spin part is mainly described in Throughout the article, right-handed coordinate systems are the laboratory-axis system in simple pulse experiments. For used for left- and right-handed rotations. example, the electric-field gradient tensor of quadrupole Active, passive, and canonical transformations are spins in powder is well studied in static, MAS, DOR, extensively discussed in the literature. As we mainly deal DAS, and MQMAS experiments. Multi-pulse experiments with spin operators and the density matrix operator in 1 fi on spin-2 systems reduce the homonuclear dipole-dipole NMR, we de ne not only the active and passive rotations Hamiltonian. These methods aim to narrow the line widths of operators but also the canonical transformation of opera- using rotations. Descriptions of rotation in classical and tors in Section 3. quantum mechanics are numerous in literature. We explore Section 4 details the active and passive rotations of the active and passive points of view, which are sources of space function Ψ(r) about a single axis using active RA(a, confusion. z) and passive RP(a, z) rotation operators. We also use
| Concepts Magn Reson Part A. 2017;45A:e21385. wileyonlinelibrary.com/journal/cmr.a © 2017 Wiley Periodicals, Inc. 1of64 https://doi.org/10.1002/cmr.a.21385 2of64 | MAN
Dirac notations,2 which have the advantages to represent not only the space function by ket state jiW but also the coordinate system by bra position basis hjr . This is not the case in standard notation about the rotation of space func- tion such as RA(a, z) Ψ(r). We show that right- and left- handed rotations of coordinate system about z-axis by angle a are expressed mathematically by the application of the operators RP(a, z) and RA(a, z) to bra position basis, respectively. In contrast, right- and left-handed active rota- tions of space function about z-axis by angle a are expressed by the application of RA(a, z) and RP(a, z) to ket state, respectively. In Section 5, active and passive rotation operators for space function are expressed in terms of Euler angles about fixed or rotated axes. These rotation operators are exten- sively discussed in the literature. Some examples are pro- vided for clarifying the notations used in these rotation operators. We recall in Section 6 the reasons for which rotations of spin components or those of spherical tensor operator components should be performed about fixed axes. As a result, we only present active and passive rotations about fixed axes in the remaining of the article. Section 7 presents the Wigner active and passive rota- FIGURE 1 (A) Right-handed counter-clockwise rotations about 0 tion matrices because they are involved in the active and the axis n; (B) left-handed clockwise rotations about the axis n passive rotations of spherical harmonics, spherical tensors, and spherical tensor operators. with spherical components. In each case we provide a sim- Section 8 presents the spherical harmonics, which are ple example. space functions. Properties of spherical tensors are Section 11 provides the definitions of spherical tensor deduced from those of spherical harmonics. Furthermore, operator and an application example about the excitation of spherical harmonics are related to Wigner rotation matri- a spin by an off-resonance RF pulse using Wigner rotation ces via simple relations. These relations allow us to matrices. determine the type (active or passive) of Wigner rotation matrix. Section 9 details the active rotations of spherical har- 2 | ROTATION IN 3-D EUCLIDEAN monics about fixed axes and the rotations of coordinate SPACE system about fixed axes, which allows us to define the rotations of spherical tensors. For simplicity, we replace A right-handed rotation in 3-D Euclidean space is defined the passive rotations of spherical harmonics by rotations of by an axis of rotation (n) through the origin (O) and an coordinate system. We explore the two points of view (ac- angle (h) about that axis (Figure 1A). This rotation is tive and passive) of the two rotation operators denoted as R(n, h). The positive sense of rotation3 is coun- fixed a; b; c fixed c; b; a RA ð Þ and RP ð Þ in Dirac notations. In each ter-clockwise when looking down the axis n toward the ori- case we provide a simple example. gin O. This description of rotation only involves the right Section 10 details the rotations of vector operator and hand. A second description of rotation involves the two those of coordinate system. A vector operator is the sim- hands: right-handed rotation (Figure 1A) and left-handed plest spherical tensor operator, whose rank is 1. In NMR, rotation (Figure 1B). In left-handed rotation, the positive the spin I is a vector operator. The study of rotation of sense of rotation is clockwise when looking down the axis vector operator is more complex than that of space func- n0 toward the origin O. tion because vector operator is sandwiched by the rotation Figure 2 adopted from Jackson4 shows a right-handed operator and its adjoint operator. In contrast, the rotation orthonormal Cartesian coordinate system (O, x, y, z). The operator is sandwiched by a bra position basis and a ket point B in the fgx; y plane is specified by the position vec- state in the case of space function. We discuss the rota- tor V with components (x1, y1, 0) along the axes as shown tions of vector operator with Cartesian components and in Figure 2A. The latter shows that a left-handed rotation MAN | 3of64
FIGURE 3 Two equivalent points of view for a rotation of axes by an angle a in a plane: (A) the vector V is fixed in the plane while a FIGURE 2 Two equivalent points of view for a rotation of axes the axes are right-hand rotated about z-axis by a positive angle in fi by an angle a in a plane: (A) the vector V is fixed in the plane, passive point of view; (B) the axes are xed while the vector V is 0 whereas the axes are left-hand rotated about z-axis by a positive angle left-hand rotated to a new position V about z-axis by a positive angle a a in passive point of view; (B) the axes are fixed while the vector V in active point of view is right-hand rotated to a new position V0 about z-axis by a positive angle a in active point of view Heine10 presents a similar example (Figure 3). Fig- ure 3A shows that a right-handed rotation about z-axis by about the z-axis by angle a applied to the coordinate sys- angle a applied to the coordinate system transforms (O, x, tem transforms (O, x, y, z) into (O, x0, y0, z0), keeping the y, z) into (O, x0, y0, z0), keeping the point B fixed. Fig- point B fixed. Figure 2B shows that a right-handed rotation ure 3B shows that a left-handed rotation about z-axis by about the z-axis by angle a applied to the position vector angle a applied to the position vector transforms V into V0, transforms V into V0, keeping the coordinate system (O, x, keeping the coordinate system (O, x, y, z) fixed. In either fi 0 ; 0 ; 0 ; 0 ; y, z) xed. In either case, the primed coordinates ðx1 y1 0Þ case, the primed coordinates ðx1 y1 0Þ are given by are given by 0 1 0 10 1 x0 cos h sin h 0 x 0 1 0 10 1 @ 1 A @ A@ 1 A 0 h h y0 ¼ sin h cos h 0 y : (2) x1 cos sin 0 x1 1 1 @ 0 A @ h h A@ A: 0 0010 y1 ¼ sin cos 0 y1 (1) 0 0010 Equation (2) is deduced from Eq. (1) by changing the We replace h in Eq. (1) with the positive angle a used sign of h as the senses of rotation described by angles a in in Figure 2. This means that left-handed rotation of the Figure 3 are the opposite of those in Figure 2. This means coordinate system (O, x, y, z) keeping the position vector that right-handed rotation of the coordinate system (O, x, y, V fixed is equivalent to right-handed rotation of the posi- z) keeping the position vector V fixed is equivalent to left- tion vector V keeping the coordinate system (O, x, y, z) handed rotation of the position vector V keeping the coor- fixed. dinate system (O, x, y, z) fixed. A single transformation matrix (Eq. (1)) can represent In practice, Figure 2 is suitable for the analysis of rotation either left-handed rotation of coordinate system or right- of vector because Figure 2B describes the right-handed rota- handed rotation of position vector. An important point to tion of vector.9,11 In contrast, Figure 3 is suitable for the note is that we use the same symbol a for two opposite analysis of rotation of coordinate system because Figure 3A senses of rotation.5–9 These opposite senses are shown by describes the right-handed rotation of coordinate system.12,13 arrows. This point is not emphasized enough in the litera- Figure 4 shows the typical rotation angle notations suitable ture. for rotation of vector. In this case, all rotations are described 4of64 | MAN
x, y, z) keeping the position vector V fixed. Another way to express the same thing is the following. Right-handed counter-clockwise rotation of the position vector V keeping the coordinate system (O, x, y, z) fixed is equivalent to right-handed clockwise rotation of the coordinate system (O, x, y, z) keeping the position vector V fixed.
3 | ACTIVE, PASSIVE, AND CANONICAL TRANSFORMATIONS OF OPERATOR
Thaller17 suggests a test of the isotropy of space in per- forming an experiment with a physical system S in a fixed coordinate system C and then repeating the experiment in a rotated coordinate system C0. The response of S to an experiment should not depend on its orientation in space. This can be done in three ways. 1. Rotate the system S but not the observer (Figure 5A). The experiment on the rotated physical system S0 is described by an observer sitting in the fixed coordinate system C. FIGURE 4 Two equivalent points of view for a rotation of axes 18 by an angle a in a plane: (A) the vector V is fixed in the plane while the The observer represents the measuring apparatus. axes are right-hand rotated about z-axis by a negative angle a in 2. Rotate the observer but not the system S (Figure 5B). passive point of view; (B) the axes are fixed while the vector V is right- The experiment on the system S in the fixed coordinate hand rotated to a new position V0 about z-axis by a positive angle a in system C is described by an observer sitting in the active point of view rotating coordinate system C0. 3. Rotate the system S and the observer (Figure 5C). The procedure consists in rotating the whole experimental 1,3,11,14–16 by right-handed rotations in right-handed coordinate setup (the physical system and the observer). The exper- systems. The sense of rotation is either negative as shown in iment on the rotated system S0 is performed and Figure 4A by angle a or positive as shown in Figure 4B by described in the rotated coordinate system C0. If the angle a. As a result, two rotation matrices are involved. That space is isotropic, the rotated system S0 in the rotated corresponding to Figure 4A describes a right-handed rotation coordinate system C0 behaves exactly as the system S 3 of coordinate system about z-axis by angle h: did in the fixed coordinate system C.
0 1 0 10 1 In classical mechanics, “rotate the system S but not the x0 cos h sin h 0 x 1 1 observer” is called an active transformation, whereas “ro- @ y0 A @ sin h cos h 0 A@ y A: (3) 1 ¼ 1 ” 0 0010 tate the observer but not the system S is called a passive transformation.17 This angle h is negative ( a) in Figure 4A. That corre- In quantum mechanics, the physical system is repre- sponding to Figure 4B describes a right-handed rotation of sented by a state vector in Hilbert space of the physical vector about z-axis by angle h:3 system and observables are provided by measuring appara- tus (or an observer in classical mechanics) sitting in a coor- 0 1 0 10 1 dinate system. As a result, moving a measuring apparatus 0 x cos h sin h 0 x 19 1 1 means moving the coordinate system. Observables (or @ y0 A ¼ @ sin h cos h 0 A@ y A: (4) 1 1 more generally operators) T defined before moving the 0 0010 measuring apparatus becomes observables T 0 after moving This angle h is positive (a) in Figure 4B. the measuring apparatus. In other words, observables are Replacing h by a in Eq. (3) and by a in Eq. (4) yield defined for a coordinate system. 0 ; 0 ; 20 21 5 the same primed coordinates ðx1 y1 0Þ. Therefore, right- Shankar, Rembold, and Auletta and Wang provide handed positive rotation of the position vector V keeping the definition of active and passive transformations. The the coordinate system (O, x, y, z) fixed is equivalent to state vectors in Hilbert space are affected in an active trans- right-handed negative rotation of the coordinate system (O, formation and left alone in the passive case. The operators MAN | 5of64
FIGURE 5 Transformations of physical system: (A) In an active transformation, the physical system S is transformed with respect to a fixed coordinate system C. (B) In a passive transformation, the coordinate system C is transformed while the physical system S is left unchanged. (C) In a canonical transformation, both the physical system S and the coordinate system C are simultaneously transformed in the same way. C and C0 are called body-attached coordinate system or the measurement apparatus are left alone in an active system that remains fixed in an active transforma- transformation and are affected in the passive case. This tion. But the matrix elements of operators are usually presentation of active and passive transformations opposes changed:25 transformed states against transformed operators. The coor- RA W T W ! R W TR W : (7) dinate system is implicitly related to measuring apparatus 1 2 A 1 A 2 or operators. In calculating matrix elements, the transformation opera- 22 23 Thompson and Lipkin also distinguish three types of tor RA can always be shifted from the state vectors to the transformations as does Thaller:17 active, passive, and operators T. Equation (7) becomes9 canonical transformations. State vectors are represented by R kets. Dynamical operators are denoted by T and transfor- W W A W y W : hj1 Tji2 ! hjð1 RAÞ TRA 2 (8) mation operators by Ri. 26 9 In NMR, we essentially manipulate spin operators and Sakurai and Napolitano and Brink and Satchler also distinguish two approaches to describing RA. density matrix operators, scarcely the state vector of the R W A W spin system. Therefore, we have to define the active and Approach 1: ji ! jiRA is a transformation of state passive rotations of an operator. We focus on operators and vector with operators T unchanged (Eq. (7)). RA y coordinate systems, whereas state vector is hidden. The Approach 2: T !ðRAÞ TRA is a transformation of active transformation may be involved in one or two coor- operators T with state vectors unchanged (Eq. (8)). 1 In other words, we can either apply the active rotation dinate systems. Two coordinate systems are involved in R W A W passive transformation. operator to state vector (ji ! jiRA ) and leave opera- tors unchanged, or we can leave state vectors unchanged R and rotate operators [T A R yTR ]. Both yield the 3.1 Active transformation of operator !ð AÞ A | same matrix elements: Three entities are involved in our approach: state vector, operator, and right-handed coordinate system. An active RA y hjW1 TjiW2 ! hjRAW1 TRjiAW2 ¼ hjðW1 RAÞ TRA W ; transformation operator R changes the physical system but 2 A (9) leaves the coordinate system unchanged.24 That is,22 so the same physics.
If RA is a right-handed counter-clockwise (Figure 1A) R W A W W W0 ; ji ! RAji¼ jiRA ¼ ji (5) active rotation operator, the transformation defined in Eq. (7) describes a right-handed counter-clockwise rota- jiW0 is the transformed state vector and tion of the physical system, keeping the operators T R T !A T: (6) fixed. It is equivalent to saying that the transformation defined in Eq. (8) describes a right-handed clockwise The operators T are unchanged by an active transfor- rotation of the operators T, keeping the physical system mation, because operators are defined for a coordinate fixed. The same matrix element describes either the 6of64 | MAN
RP 0 y right-handed counter-clockwise rotation of the state T ! T ¼ðRPÞ TRP as a left-handed rotation of the coor- vectors or the right-handed clockwise rotation of the dinate system, keeping the operators T fixed. It is also operators. called left-handed passive rotation of operators T. To distinguish these two ways to describe a matrix ele- RA ment, we call the first way (jiW ! jiRAW ) a right-handed (Figure 1A) active rotation of state vector and the second 3.3 | Canonical transformation of operator RA y way [T !ðRAÞ TRA] a left-handed (Figure 1B) active A canonical transformation by the operator R changes both rotation of operators T. In both ways, the coordinate system A the physical system and the coordinate system. We have,22 remains fixed.
RA 0 jiW ! RAjiW ¼ jiRAW ¼ jiW ; (16) 3.2 | Passive transformation of operator and
A passive transformation operator R leaves the physical R P T !A T0: (17) system unchanged, but changes the coordinate system. The change in coordinate system is also called the change in Recall in quantum mechanics, observables are provided description.27 This means two coordinate systems are by a measuring apparatus. Moving a measuring apparatus involved in passive transformation. We have22 means moving the coordinate system. The transformation defined by Eqs. (16) and (17) is R jiW !P jiW (10) called canonical because it transforms state vector and operators but leaves the matrix elements invariant:28 and
R T !P T0; (11) W W RA W0 0 W0 W W : hj1 Tji2 ! 1 T 2 ¼ hj1 Tji2 (18) T and T 0 being the same operators expressed in terms of the old and new coordinate systems, respectively. Matrix The important point is that matrix elements do not elements are usually changed under a passive transforma- change by a unitary transformation of state vector and tion: a unitary transformation of observables. An important class of canonical transformations are those which R 23 hjW Tji W !P hjR W TRji¼W hjðW R ÞyTR jiW correspond to symmetries of the physical system. 1 2 P 1 P 2 1 P P 2 They are translations, rotations, space inversion, and ¼ W T0jiW : 1 2 time-reversal. (12) Equation (16) allows us to rewrite the matrix element in Eq. (18) as As it is well-known (see Section 4) that y y W0 T0 W0 R W T0 R W W R T0R W ðRAÞ ¼ RP; (13) 1 2 ¼ A 1 jiA 2 ¼ hj1 ðÞA Aji2 ¼ hjW TjiW : the operators T 0 in Eqs. (11) and (12) are defined by19,23 1 2 (19) As a result, Eq. (17) becomes RP 0 y y T ! T ¼ðRPÞ TRP ¼ RATðRAÞ : (14) RA 0 y: Therefore, Eq. (12) becomes T ! T ¼ RATðRAÞ (20) R W W P W 0 W W y W The transformation in Eq. (20) differs with hj1 Tji2 ! T ¼ ðRPÞ TRP R 1 2 1 2 T !ðA R ÞyTR discussed in active transformation of oper- W y W : A A ¼ 1 RATðRAÞ 2 (15) ator. As the latter is a left-handed active rotation of opera- tors T in a fixed coordinate system, the transformation in If RA is a right-handed counter-clockwise active rotation operator, it follows that R is also a right-handed counter- Eq. (20) is a right-handed active rotation of operators T in P fi clockwise, but passive rotation operator. We have just pro- a xed coordinate system. R posed that the transformation T A R yTR deduced In quantum mechanics textbooks on symmetry operations !ð AÞ A 29 from Eq. (8) describes a left-handed active rotation of oper- (Chapter 3 in Ballentine, Section 14 of Chapter 3 in 30 ators T, keeping the coordinate system fixed. We deduce Biedenharn and Louck, Complement BVI in Cohen-Tan- 31 32 that the transformation T RP T0 R T R y in Eq. (14) noudji et al., Pages 99-101 in Davydov, Chapter 1 in ! ¼ A ð AÞ 27 28 describes a right-handed active rotation of operators T, Haywood, Chapter 29 in Hecht, Section 7.4 in Konishi 33 keeping the coordinate system fixed. Thanks to Eq. and Paffuti, Chapter 2 and Section 6.3 in Mathur and 14 34 35 (14), it is equivalent to describing the transformation Singh, Chapter 13 in Messiah, Section 4.10 in Miller, MAN | 7of64
Section 2 in Rose,36 Page 3 in Tung,37 Chapter 3 in Zare,38 We first provide the definitions of terms such as Hilbert Pages 102 and 116 in Zettili39), the active transformation is space, scalar product, position representation, and wave func- defined as the following: In conjunction with the transforma- tion used in quantum mechanics that are important for the tion of the state vectors induced by a symmetry operation comprehension of some common notations. Especially, scalar (unitary transformation) given by Eq. (16), each operator T product and wave function have two possible interpretations. undergoes the transformation defined in Eq. (20). In the canonical transformation by the operator RA, both 4.1 | Hilbert space operators and coordinate system rotate together. The right- handed active rotation of operators T in a fixed coordinate A Hilbert space is an abstract vector space with a scalar system is followed by a right-handed rotation of the fixed product and a positive definite norm.28,39,43 In quantum coordinate system. The rotated coordinate system is called mechanics, physical properties of a system are represented body-attached coordinate system. by linear operators R, L, ... and a state vector Φ or state In fact, the active transformation corresponds to our for short by ket state jiU or wave function Φ(r), and by active rotation in one coordinate system, the fixed coordi- bra state hjU or complex conjugate U ðrÞ of wave function. nate system associated with the unprimed operator. The While ket state belongs to a Hilbert space, its bra state canonical transformation corresponds to our active rotation belongs to the corresponding dual Hilbert space. Wave in two coordinate systems:1 the fixed coordinate system is functions, complex-valued functions of space, are square- associated with the unprimed operator and the body- integrable functions. Their space possesses the properties attached coordinate system is attached to the operator and of an infinite-dimensional Hilbert space.39 rotates with it. Equation (18) means that the primed opera- The quantum state is an abstract concept, but it contains tor in the body-attached coordinate system has the same all the information about the system. Observables are Hermi- matrix elements as those of the unprimed operator in the tian operators, which take real eigenvalues only. A wave fixed coordinate system before the active rotation. function is associated with a state, whereas a space function is not necessary a state. An operator acting on a ket produces another ket. However, the result of an operator on a state is 4 | ROTATION OF SPACE not necessary a state, it may be a space function.35,44 In our FUNCTION ABOUT A SINGLE AXIS case, a rotation operator converts an old state into a new state because rotation operators are unitary operators. The study of active rotation of space function about a single One of the key benefits of Dirac notation2 is that operators, axis involves one or two coordinate systems. When one coor- kets, and bras are independent of any representation basis. dinate system is used,40 it corresponds to a space-fixed coor- Very often it is convenient to use some less abstract represen- dinate system (O, x, y, z).38 When two coordinate systems tations of states and operators to get numerical results. are used,36 one corresponds to the space-fixed coordinate In finite-dimensional Hilbert space, we expand any ket system (O, x, y, z) and the other is attached to the space func- as a linear combination of the eigen kets of an observ- tion, rotates with it, and is called body-attached coordinate able.45 If Q is such an observable in a 2-D vector space system (O, x0, y0, z0) or molecule-fixed frame.38,41 Viewed in with eigen kets {jii , jij }, then an arbitrary state jiU can be its body-attached coordinate system, the space function expressed as a linear combination of {jii , jij }. These two remains motionless. We previously shown1 that Rose36 used eigen kets are basis vectors in an orthonormal coordinate this approach for deducing the active rotation operator system. That is, involving Euler angles. As Rose did not use figures to sup- 40 42 U : port his presentation, Wolf and Bouten understood that ji¼aijiþi ajjij (21) Rose dealt with passive rotation of space function. The components are given by hii j U ¼ a and hij j U ¼ a . In this section, we formulate the rotation of a space i j Therefore, the state Φ may be written as an abstract ket function in one direction and the equivalent rotation of state jiU , a superposition of ket basis vectors (Eq. (21)), or coordinate system in the opposite direction using right and a set of numbers (expansion coefficients hii j U ¼ a and left hands. Two approaches are used: the standard one i hij j U ¼ a in Eq. (21)). The three kets jiU , jii , and jij involving rotation operator and space function [R (a, j A belong to the same Hilbert space. The set of components z)Ψ(r)] and that involving Dirac notations [hjr RAða; zÞjiW ]. {a , a } of jiU along the eigen kets of Q completely The second approach has the advantage to exhibit both the i j describe jiU . They are called the Q representation of coordinate system represented by bra position basis and the jiU .29,45 Changing representations does not change the space function represented by ket space function. In the state ket. The ket state jiU itself is representation-free. Sim- standard approach [R (a, z)Ψ(r)], the coordinate system is A ilarly, the scalar product is also representation-free.35,46 not explicitly defined. 8of64 | MAN
When we take a scalar product in quantum mechanics, we Ket position basis: jir ; 8r 2 R3: (25) use a common basis in both states. In Eq. (21), the ket state is expressed in a countable A scalar product or a bra-ket pair can be interpreted in basis, identified by an integer. On the other hand, the posi- two ways.39 First, a bra-ket pair corresponds to a vector pro- tion basis is uncountable, identified by real numbers. jection. The projection of jiU onto the eigen ket jii is hii j U , The ket position bases verify26,28,39 the closure relation that on jij is hij j U . Therefore, Eq. (21) can be written as Z X 3 jiU ¼ hii j U jii þ hij j U jij ¼ jia hia j U ; (22) jir hjr d r ¼ 1; (26) a¼i;j P and the orthonormality condition because the closure relation jia hj¼a 1 is available. a¼i;j Second, the quantity hiW j U represents the probability hr0jir ¼ dðr r0Þ; (27) amplitude that the system’s ket state jiU before a measure- the latter being the highly localized Dirac delta func- ment will be found to be in another ket state jiW after the tion5,28,33,44,48 defined by measurement. Some Hilbert spaces are of infinite number of dimensions, Z where the observables can take on an infinite number of pos- f ðrÞ¼ dðr r0Þf ðr0Þd3r0 (28) sible values. The typical examples are function spaces where vectors are functions. In quantum mechanics, wave functions for any function f that is continuous at r. Suppose the parti- are often space functions.32,43 For states with continuous cle is at position r . The corresponding ket is r . The eigenvalues, such as two wave functions W ðrÞ and W ðrÞ, 0 0 1 2 position representation of r 35,49 or its wave function is the scalar product is defined by integral33,47 0 d Z ðr r0Þ. fi 3 The norm of the continuous ket position bases is in - W W ¼ W ðrÞW ðrÞd r: (23) 1 2 1 2 nite.29,39 Therefore, the kets jir are not square inte- 28,39 fi grable. As Hilbert space admits only vectors of nite W W 29 ’ In other words, 1 2 corresponds to the integral of norm, the kets jir are not elements of the particle s Hil- W W 46 34,39,49 1ðr Þ 2ðrÞ. As we mentioned above, the scalar product bert space but remain a valid basis that spans the W W W W ’ 28,39 1 2 is the projection of 2 onto 1 . However, particle s Hilbert space. W W U the more general interpretation of 1 2 , independent of The scalar product hir j of jir with an arbitrary state U U 49 representation, is that it represents the probability amplitude ket jiis a square-integrable function of r, ðrÞ. The W W 26 U U for the ket state 2 to be found in the ket state 1 ; result of an operator L on jiis Lji. Its position repre- operators are not explicitly involved in this case. sentation is hjr LjiU . If an observable N is involved, we are interested in cal- culating the mean value hjW NjiW , which can be checked by performed the same experiment on the same state many 4.3 | Wave function times. The state Ψ appears as both a bra and a ket. Space function Ψ(r) of a particle in 3-D space is defined hjW NjiW is also called expectation value. by24,25 WðrÞ¼hir j W ; (29) 4.2 | Position representation which is not an equation in the usual sense. Ψ(r) is also known as normalized Schrodinger€ wave function for ket We focus on the position representation corresponding to state jiW 14,28,39,47 with measuring position operator R for a spinless particle in 3-D Z Z space. The eigen kets of R are jir with corresponding 0 0 3 0 0 0 3 0 eigenvalues r,26 the position vectors of the particle. The jiW ¼ jir hjr Wid r ¼ jir Wðr Þd r : all space all space eigen kets are labeled by the eigenvalues. That is, (30) A particle0s wave function defines not only its position Rjir ¼ rrji: (24) but also all the spatial properties of the particle. The pro- These eigenvalues are continuous as every value of duct of the wave function with its complex conjugate is the position is measurable. The position representation may be density probability25,44 for finding the particle at position r. defined by the continuous set of ket position basis jir .So Equation (29) relates the conventional notation of the we have abstract wave function Ψ in position representation with MAN | 9of64
Dirac notation of the wave function.46 It shows that the Before the rotation, the physical system is described by wave function is the projection of the ket state jiW onto the ket state jiW , whose space function Ψ(r) is the r-posi- the bra position base hjr .39,50 tion representation of jiW in (O, x, y, z). Ψ(r) is a mathe- The use of the position eigenstates as basis states to rep- matical function of r, the first meaning (1) of a space resent operators and states is referred to as the position rep- function. After the rotation, the physical system is 0 resentation. Equation (30) is the expansion of an arbitrary described by the new ket state jiW A ¼ RAða; zÞjiW , whose W 0 Ψ0 ket state jiin terms of the ket position bases jir , where space function A(r) is the r-position representation of fi W0 the expansion coef cient is the wave function in position A in the same coordinate system (O, x, y, z). As whole representation. The summation over the index of a com- functions, the mathematical expression of Ψ(r) differs with 0 0 plete set of eigenvalues in Eq. (22) becomes an integral that of Ψ A(r), that is, Ψ(r)6¼Ψ A(r). over continuous eigenvalues in Eq. (30). Ψ(r) is also the As the rotation operator RA(a, z) is a unitary operator, 13,33,44,51 0 probability amplitude for finding a particle at posi- the value Ψ A(r) of the space function at the position r is tion r in an orthonormal basis of 3-D space. equal to the value Ψ(r1) of the space function before rota- Therefore, the symbol Ψ(r) has two different meanings tion at the position r1 (Figure 6A): in different contexts:44 0 W ðrÞ¼Wðr1Þ: (31) 1. The first meaning is WðrÞ¼jiW , Ψ(r) means a mathe- A 0 matical function of r, the whole function without being The second meaning (2) of space functions Ψ(r) and Ψ A(r) explicit the actual value of r. It is the position represen- is used in Eq. (31). 3 tation of the abstract ket state jiW . As the list of compo- The two positions r1 and r in (O, x, y, z) are related by nents of a vector in a basis of 3-D vector space is a the right-handed active rotation matrix Az(a): particular representation of that vector, Ψ(r) and jiW are similar to the components of a vector in a particular 0 1 cos a sin a 0 basis and the vector, respectively. Similarly, WðpÞ¼ r ¼ A ðaÞr ¼ @ sin a cos a 0 Ar : (32) jiW is the momentum representation of the abstract ket z 1 1 001 state jiW , but Ψ(p) is the Fourier transform of Ψ(r).26 It is important to note that the mathematical expression of Formally, we can introduce the right-handed passive Ψ Ψ 3 (r) differs with that of (p). rotation matrix Pz(a): 2. The second meaning of Ψ(r) is Eq. (29), that is, the A a P a : r1 ¼ zð Þr ¼ zð Þr (33) probability amplitude Ψ(r) for finding the particle in the state jiW at the some specific position r,52 or more gen- Equation (33) relates the components of position vectors erally the value Ψ(r) of the space function Ψ at some using rotation matrices. 31 specific position r, not the whole function. A scalar Equation (33) does not mean the two positions r and r1 product is a number, and hir j W is the value of Ψ at are the same position vector viewed in two different coordi- P a the position r.Ifhir j W ¼ 0, it does not mean jiW and nate systems. It suggests that the two positions r and z( )r jir are orthogonal because they do not belong to the (Figure 6B) are viewed in two different coordinate systems, P a 0 0 0 same Hilbert space. It just means the value of the wave that of r is (O, x, y, z) and that of z( )r is (O, x , y , z ). The function Ψ at the specific position r is zero. latter is generated by the left-handed rotation of (O, x, y, z) about z-axis by angle a. Notice that we use the same symbol A particle in state jir is located at position r. In con- a for two rotation angles of opposite signs. In other words, trast, a particle in state jiW can be found at different posi- the components of position r in (O, x, y, z) are identical to 0 0 0 tions with corresponding probability amplitudes. The action those of position Pz(a)r in (O, x , y , z ). of an operator L on a space function is related to its action The left-handed rotation of coordinate system is related to 29 P a on the abstract vector space by the rule LWðrÞ¼hjr LjiW the fact that the position r1= z( )r is deduced from the posi- P a with WðrÞ¼hjr Wi. tion r by a left-handed rotation. As a result, z( )maybe either a right-handed passive rotation matrix or a left-handed passive rotation matrix, depending on the circumstance. 4.4 | Active rotation of space function in one We illustrate this property of P (a) with Figure 7. Fig- coordinate system z ure 7A shows the space function Ψ(y) in the fixed coordi- Consider the right-handed active rotation9,11,22,32,34,38,40,53 nate system (O, x, y, z). The action of a right-handed active a Ψ a p about z-axis by angle of the space function in the space- rotation of (y) about z-axis by angle ¼ 2 is shown in W0 fixed coordinate system (O, x, y, z). The rotation operator Figure 7B where the space function becomes Að xÞ. The a P a p is denoted by RA( , z). action of zð ¼ 2Þ on r is 10 of 64 | MAN
FIGURE 6 (A) Right-handed active rotation of space function about z-axis by angle a shown in a single coordinate system (O, x, y, z). 0 Ψ(r1) is the initial space function before rotation and Ψ A(r) is the rotated space function. (B) Equivalent left-handed rotation of coordinate system from (O, x, y, z)to(O, x0,y0,z0). Position vector r is attached to (O, x, y, 0 0 0 z) and position vector Pz(a)r is attached to (O, x ,y,z)
0 10 1 cosða ¼ pÞ sinða ¼ pÞ 0 x B 2 2 CB C P a p @ a p a p A@ A zð ¼ 2Þr¼ sinð ¼ 2Þ cosð ¼ 2Þ 0 y 001z 0 10 1 0 1 010 x y B CB C B C ¼ @ 100A@ y A ¼ @ x A: (34) 001 z z
In fact, the position vector r represents the coordinate FIGURE 7 Equivalence between right-handed active rotation of system. That is, x- and y-axes become y- and –x-axes, space function about z-axis by angle a=p/2 transforming the space 0 respectively. In Figure 7C, the original x- and y-axes are function Ψ(y) in (A) to the space function Ψ A( x) in (B) and left- black lines and the rotated axes are green lines. The green handed rotation of coordinate system about z-axis by angle a = p/2 coordinate system is deduced from the black one by a left- transforming black (O, x, y, z) to green (O, x, y, z) in (C); the space a p function is Ψ(y) in black (O, x, y, z) and Ψ( x) in green (O, x, y, z) handed rotation about z-axis by angle ¼ 2. The space function becomes Ψ( x) in the green coordinate system like the space function in Figure 7B. 0 0 24 W r r Equation (31) becomes The space function ðÞis the -position representa- tion of the initial ket state jiW in (O, x0, y0, z0). The same W0 W W P a : space function Ψ designates the initial space function in AðrÞ¼ ðr1Þ¼ ðÞzð Þr (35) (O, x, y, z) and that in (O, x0, y0, z0). This may confuse us, P a The position z( )r is described in the left-hand rotated but it is correct, because the ket state jiW is viewed in two 0 0 0 coordinate system (O, x , y , z ). coordinate systems. Its space function is Ψ(r)in(O, x, y, MAN | 11 of 64