Bras, Kets, and Matrices When quantum mechanics was originally developed (1925) the simple rules of matrix

arithmetic was not yet part of every technically educated person’s asic !nowledge" #s a result\$ %eisen erg didn’t reali&e that the (to him) mysterious non' commutative multiplication he had discovered for his quantum mechanical varia les was nothing other than matrix multiplication" (i!ewise when )irac developed his own distinctive approach to quantum mechanics (see his oo! *+he ,rinciples of -uantum .echanics/\$ first pu lished in 1901) he expressed the quantum mechanical equations in terms of his own customi&ed notation 2 which remains in widespread use today 2 rather than availing himself of the terminology of matrix alge ra" +o some extent this was (and is) 3ustified y the convenience of )irac’s notation\$ ut modern readers who are already acquainted with the elementary arithmetic of matrices may find it easier to read )irac (and the rest of the literature on quantum mechanics) if they are aware of the correspondence etween )irac’s notation and the standard language of matrices +he entities that )irac called *!ets/ and * ras/ are simply column vectors and row vectors\$ respectively\$ and the linear operators of )irac are simply square matrices" 4f course\$ the elements of these vectors and matrices are generally complex num ers" 5or convenience we will express ourselves in terms of vectors and matrices of si&e 0\$ ut they may e of any si&e\$ and in fact they are usually of infinite si&e (and can even e generali&ed to continuous analogs)" 6n summary\$ when )irac refers to a * ra/\$ which he denoted as 7#8\$ a *!et/\$ which he denoted as 89:\$ and a measurement operator α\$ we can associate these with vectors and matrices as follows

+he product of a ra and a !et\$ denoted y )irac as 7#889: or\$ more commonly y omitting one of the middle lines\$ as 7#89:\$ is simply the ordinary (complex) num er given y multiplying a row vector and a column vector in the usual way\$ i"e"\$ (i!ewise the product of a ra times a linear operator corresponds to the product of a row vector times a square matrix\$ which is again a row vector (i"e"\$ a * ra/) as follows and the product of a linear operator times a !et corresponds to the product of a square matrix times a column vector\$ yielding another column vector (i"e"\$ a !et) as follows 4 viously we can form an ordinary (complex) num er y ta!ing the compound product of a ra\$ a linear operator\$ and a !et\$ which corresponds to forming the product of a row vector times a square matrix times a column vector .

We can also form the product of a !et times a ra\$ which gives a linear operator (i"e"\$ a square matrix)\$ as shown elow" )irac placed the ras and the !ets into a one'to'one correspondence with each other y defining\$ for any given !et 8#:\$ the ra 7#8\$ which he called the con3ugate imaginary" 6n the language of matrices these two vectors are related to each other y simply ta!ing the transpose and then ta!ing the complex con3ugate of each element (i"e"\$ negating the sign of the imaginary component of each element)" +hus the following two vectors correspond to what )irac called con3ugate imaginaries of each other" where the over ar signifies complex con3ugation" )irac chose not to call these two vectors the complex con3ugates of each other ecause they are entities of different types\$ and cannot e added together to give a purely real entity" .evertheless\$ li!e ordinary complex num ers\$ the product of a ra and its (imaginary) con3ugate !et is a purely real num er\$ given y the *dot product/ .

+he con3ugation of linear operators is formally the same\$ i"e"\$ we ta!e the transpose of the matrix and then replace each element with its complex con3ugate" +hus the con3ugate transpose of the operator α (defined a ove) corresponds to the matrix +his is called the adjoint of α\$ which we denote y an over ar" <ecall that the ordinary transpose operation satisfies the identity for any two matrices α and β" 6t’s easy to verify that the con3ugate transpose satisfies an identity of the same form\$ i"e"\$ +he rule for forming the con3ugate transpose of a product y ta!ing the con3ugate transpose of each factor and reversing the order is quite general" 6t applies to any num er of factors\$ and the factors need not e square\$ and may even e ordinary num ers (with the understanding that the con3ugate transpose of a complex num er is simply its con3ugate)" 5or example\$ given any four matrices α\$β\$γ\$δ with suita le dimensions so that they can e multiplied together in the indicated sequence\$ we have .

+his applies to the row and column vectors corresponding to )irac’s ras and !ets\$ as well as to scalar (complex) num ers and square matrices" 4ne minor shortcoming of )irac’s notation is that it provides no sym olic way of denoting the con3ugate transpose of a ra or !et" )irac reserved the over ar notation for what he called the con3ugate complex\$ whereas he referred to the con3ugate transpose of ras and !ets as con3ugate imaginaries" +hus the con3ugate imaginary of 7#8 is simply written as 8#:\$ ut there is no sym olic way of denoting =the con3ugate imaginary of 7#8α’\$ for example\$ other than explicitly as " We might e tempted to define the *real part/ of a matrix as the matrix consisting of 3ust the real parts of the individual elements\$ ut )irac pointed out that it’s more natural (and useful) to carry over from ordinary complex num ers the property that an entity is *real/ if it equals its *con3ugate/" 4ur definition of *con3ugation/ for matrices involves ta!ing the transpose as well as con3ugating the individual elements\$ so we need our generali&ed definition of *real/ to ta!e this into account" +o do this\$ we define a matrix to e *real/ if it equals its ad3oint" >uch matrices are also called selfadjoint" ?learly the real elements of a self'ad3oint matrix must e symmetrical\$ and the imaginary parts of its elements must e anti' symmetrical" 4n the other hand\$ the real components of the elements of a purely *imaginary/ matrix must e anti'symmetrical\$ and the imaginary components of its elements must e symmetrical" 6t follows that any matrix can e expressed uniquely as a sum of *real/ and *imaginary/ parts (in the sense of those terms 3ust descri ed)" (et αmn and αnm e symmetrically placed elements of an ar itrary matrix α\$ with the complex components .

+hen we can express α as the sum α @ ρ A η where ρ is a *real/ matrix and η is an *imaginary/ matrix\$ where the components of ρ are and the components of η are 9y defining *real/ and *imaginary/ matrices in this way\$ we carry over the additive properties of complex num ers\$ ut not the multiplicative properties" 6n particular\$ for ordinary complex num ers\$ the product of two reals is real\$ as is the product of two imaginaries\$ and the product of a real and an imaginary is imaginary" 6n contrast\$ the product of two real (i"e"\$ self'ad3oint) matrices is not necessarily real" +o show this\$ first note that for any two matrices α and β we have %ence if α and β are self'ad3oint we have +hus if α and β are self'ad3oint then so is αβ A βα" >imilarly\$ for any two matrices α and β we have %ence if α and β are self'ad3oint we have .

+his shows that if α and β are self'ad3oint then so is i(αβ − βα)" #lso\$ notice that multiplication of a matrix y i has the effect of swapping the real and imaginary roles of the elementary components\$ changing the symmetrical to anti'symmetrical and vice versa" 6t follows that the expression αβ 2 βα is purely imaginary (in )irac’s sense)\$ ecause when multiplied y i the result is purely real (i"e"\$ self'ad3oint)" +herefore\$ in general\$ the product of two self'ad3oint operators can e written in the form where the first term on the right side is purely *real/\$ and the second term is purely *imaginary/ (in )irac’s sense)" +hus the product of two real matrices α and β is purely real if and only if the matrices commute\$ i"e"\$ if and only if αβ @ βα" #t this point we can egin to see how the ideas introduced so far are related to the physical theory of quantum mechanics" We’ve associated a certain o serva le varia le (such as momentum or position) with a linear operator α" .ow suppose for the moment that we have diagonali&ed the operator\$ so the only non'&ero elements of the matrix are the eigenvalues\$ which we will denote yλ1\$ λ2\$ λ0 along the diagonal" We require our measurement operators to e %ermitian\$ which implies that their eigenvalues are all purely real" 6n addition\$ we require that eigenvectors of the operator must span the space\$ which is to say\$ it must e possi le to express any state vector as a linear com ination of the eigenvectors" (+his requirement on *o serva les/ may not e logically necessary\$ ut it is one of the postulates of quantum mechanics") .

erhaps the most natural way of forming a real scalar value from the given information is the product 6f we normali&e the state vector so that its magnitude is 1\$ the a ove expression is simply a weighted average of the eigenvalues" +his motivates us to normali&e all state vectors\$ i"e"\$ to stipulate that for any state vector # we have 4f course\$ since the components #3 are generally complex\$ there is still an ar itrary factor of unity in the state vector" 6n other words\$ we can multiply a state vector y any complex num er of unit length" i"e"\$ any num er of the form eiθ for an ar itrary real angle θ\$ without affecting the results" +his is called a phase factor" 5rom the standpoint of the physical theory\$ we must now decide how are we to interpret the real num er denoted y 7#8α8#:" 6t might e tempting to regard it as the result of applying the measurement associated with α to a system in state #\$ ut )irac asserted that this can’t e correct y pointing out that the num er 7#8αβ8#: does not in general equal the num er 7#8α8#:7#8β8#:" 6t isn’t entirely clear why this inequality rules out the stated interpretation" (5or example\$ one could 3ust as well argue .ow\$ given a physical system whose state is represented y the !et 8#:\$ and wish to measure (i"e"\$ o serve) the value of the varia le associated with the operator α" ..

that α and β cannot represent o serva les ecause αβ does not in general equal βα") 9ut regardless of the 3ustification\$ )irac chose to postulate that the num er 7#8α8#: represents the average of the values given y measuring the o serva le α on a system in the state # a large num er of times" +his is a remar!a le postulate\$ since it implicitly concedes that a single measurement of a certain o serva le on a system in a specific state need not yield a unique result" 6n order to lend some plausi ility to this postulate\$ )irac notes that the ra'!et does possess the simple additive property >ince the average of a set of num ers is a purely additive function\$ and since the ra'!et operation possesses this additivity\$ )irac argued that the stated postulate is 3ustified" #gain\$ he attri uted the non'uniqueness of individual measurements to the fact that the product of the ra'!ets of two operators does not in general equal the ra'!et of the product of those two operators" +here is\$ however\$ a special circumstance in which the ra'!et operation does possess the multiplicative property\$ namely\$ in the case when oth α and β are diagonal and # is an eigenvector of oth of them" +o see this\$ consider the two diagonal matrices +he product of these two o serva les is .

ow\$ ta!ing the state vector to e the normali&ed eigenvector B1 1 1C for example\$ we get +herefore\$ following )irac’s reasoning\$ we are 3ustified in treating the num er 7#8α8#: as the unique result of measuring the o serva le α for a system in state # provided that # is an eigenvector of α\$ in which case the result of the measurement is the corresponding eigenvalue ofα" 5rom this it follows that quantum mechanics would entail no fundamental indeterminacy if it were possi le to simultaneously diagonali&e all the o serva les" 6n that case\$ all o serva les would commute\$ and there would e no *%eisen erg uncertainty/" %owever\$ we find that it is not possi le to simultaneously diagonali&e the operators corresponding to every pair of o serva les" >pecifically\$ if we characteri&e a physical system in %amiltonian terms y defining a set of configuration coordinates q1\$ q2\$ D and the corresponding momenta p1\$ p2\$ D\$ then we find the following commutation relations +hese expressions represent the commutators of the indicated varia les\$ which are closely analogous to the .oisson rac!ets of the canonical varia les in classical physics" +hese commutators signify that each configuration coordinate q3 is incompati le with the corresponding momentum p3" 6t follows that we can transform the system so as to diagonali&e one or the other\$ ut not oth" .ote also that in this special case the multiplication of operators is commutative" ..

>hortly after %eisen erg created matrix mechanics\$ >chrEdinger arrived at his theory of wave mechanics\$ which he soon showed was mathematically equivalent to matrix mechanics\$ despite their superficially very different appearances" 4ne reason that the two formulations appear so different is that the equations of motion are expressed in two completely different ways" 6n %eisen erg’s approach\$ the *state/ vector of the system is fixed\$ and the operators representing the dynamical varia les evolve over time" +hus the equations of motion are expressed in terms of the linear operators representing the o serva les" 6n contrast\$ >chrEdinger represented the o serva les as fixed operators\$ and the state vector as varying over time" +hus his equations of motion are expressed in terms of the state varia le (or equivalently the wave function)" +his dichotomy doesn’t arise in classical physics\$ ecause there the dynamic varia les define the state of the system" 6n quantum mechanics\$ these two things (o serva les and states) are two distinct things\$ and the outcome of an interaction depends only on the relationship etween the two" %ence we can hold either one constant and allow the other to vary in such a way as to give the necessary relationship etween them" +his is somewhat reminiscent of the situation in pre'relativistic electrodynamics\$ when the interaction etween a magnet and a conductor in relative motion was given two formally very different accounts\$ depending on which component was assumed to e moving" 6n the case of electrodynamics a great simplification was achieved y adopting a new formalism (special relativity) in which the empirically meaningless distinction etween a solute motion and a solute rest was eliminated\$ and everything was expressed purely in terms of relative motion" +he explicit motivation for this was the desire to eliminate all asymmetries from the formalism that were not inherent in the phenomena" 6n the case of quantum mechanics\$ we have asymmetric accounts (namely the views of %eisen erg and >chrEdinger) of the very same phenomena\$ so it .

seems natural to suspect that there may e a single more symmetrical formulation underlying these two accounts" +he asymmetry in the existing formalisms seems to run deeper than simply the choice of whether to ta!e the states or the o serva les as the dynamic element su 3ect to the laws of motion" +his fundamentally entails an asymmetry etween the o server and the o served\$ whereas from a purely physical standpoint there is no o 3ective way of identifying one or the other of two interacting systems as the o server or the o served" # more suita le formalism would treat the state of the system eing *o served/ on an equal footing with the state of the system doing the *o serving/" +hus\$ rather than having a matrix representing the o serva le and a vector representing the state of the system eing o served\$ we might imagine a formalism in which two systems are each represented y a matrix\$ and the interaction of the two systems results in reciprocal changes in those matrices" +he change in each matrix would represent the a sorption of some information a out the (prior) state of the other system" +his would correspond\$ on the one hand\$ to the reception of an eigenvalue y the o serving system\$ and on the other hand\$ to the *3ump/ of the o served system to the corresponding eigenvector" 9ut oth of these effects would apply in oth directions\$ since the situation is physically symmetrical" 6n addition to it’s central role in quantum mechanics\$ the ra'!et operation also plays an important role in the other great theory of 21th century physics\$ namely\$ general relativity" >uppose with each incremental extent of space'time we associate a *state vector/ x whose components (for any given coordinate asis) are the differentials of the time and space coordinates\$ denoted y super'scripts with x1 representing time" +he ra and !et representatives of this state vector are .

.ow we define an *o serva le/ g corresponding to the metric tensor #ccording to the formalism of quantum mechanics\$ a measurement of the *o serva le/ g of this *state/ would yield\$ on average\$ the value +his is 3ust the invariant line element\$ which is customarily written (using Finstein’s summation convention) as %owever\$ unli!e the case of quantum mechanics\$ the elements of the vector x and operator g are purely real (at least in the usual treatment of general relativity)\$ so complex con3ugation is simply the identity operation" #lso\$ ased on the usual interpretation of .

in!ows!i metric of flat spacetime (in geometric units so that c @ 1)\$ we would expect a measurement of g on x to yield A1\$ '1\$ '1\$ or '1\$ with pro a ilities proportional to (dx1)2\$ (dx1)2\$ (dx2)2\$ (dx0)2 respectively" 6t isn’t o vious how to interpret this in the context of spacetime in general relativity\$ although the fact that the average of many such measurements must yield the familiar squared spacetime interval (ds)2suggests that the individual *measurements/ are some !ind of quanti&ed effects that collectively com ine to produce what we perceive as spacetime intervals" .quantum mechanics\$ we would expect a measurement of g to always yield one of the eigenvalues of g\$ leaving the state x in the corresponding eigenstate" #pplying this literally to the diagonal .