|| 1¥) {i -norm of vector jy) lliw) (viv) Important for normalization of |y) i.e. |w)/ || (w) || (0/4ly) - inner product of |p) and 4|y). ‘or inner product of A'\@) and |y) 2 Postulates of Quantum Mechanics distinction needs to be made between quantum mechanics, quantum ph puting. Quantum mechanics is a mathematical language, much like calculus. Just uses calculus to explain nature, quantum physics uses quantum mechanics to _ Just as classical computers can be thought of in boolean algebra terms, quantum ‘reasoned about with quantum mechanics. There are four postulates to quantum isa complex vector space with inner prod- the state space of the system. The system is which is a unit vector in the system's state space. Let |) and |) be orthonormal basis In quantum computing we usually label the basis is only a name. For example, |o) = |0) and gine that “(0)” is being represented by an(spin state or other quantum phenomena) and the logical meaning. This same detachment occurs classically where we traditionally call a high positive voltage “I” and a low ground potential “O”. Note that |y) = a|0) +5]1) must be a unit vector. In other words, (yly) = 1 or |a|? + |b|? = 1. For quantum computing {a,} € € ‘This formalism for a quantum bit is a direct extension of one way to describe a classical computer. is, ay may write that a classical bit |) is in the state |) = x|0) +-y|1). The only difference |» are defined not over the complex numbers but rather from the set {0,1}. That is {x.y} € ‘The same normalization condition applies |x| + |p|? = 1. This normalization condition is erty of quantum mechanics but rather of probability theory. Evolution of quantum systems of the system at time /1 is related to the state of |y’) of the system id which depends only on times /, and f." is 1) ‘and only on fy and fy is a subtle and disappointing fact. ef |) then quantum computers could easily solve NP ‘as something you can apply to a quantum bit but you ‘occurs without any regard to the current state of |).Important: U must be unitary, that is Utu = 7Note that M}My = Mo, hence P(0) = (y|Moly) =.) [ 4 + [ =I # ~wei [6] =r Hence the Probability of, ‘measuring (0) is related to its probabilit ity amplitude a by way of |a\2, Itimportant to note that the state after ‘surement is related to the ‘outcome of the measurement, _ For example, suppose |0) was measure ther State of the system after this measurement is _Te-normalized as: note We are forced to wonder if Postulate 3 can be derived from Postulate 2. It seems ‘that measurement in the physical world is just interacting a qubit with other qubits ‘Strange to have measurement be its own postulate. At this Point though physicists three, so we shall just have to be of @ composite physical system is the tensor product of the state ponent physical systems. [sic] e.g. suppose systems 1 through n 2), then the joint state of the total system is |yi) © |yo) @ ) = e|0) +), then +a-d0)|1) ++ cl1)|0) +b-d)1)|1) ba\1) is a simple concatenation). For quantum systems ition, thats if system 4 is in state |) and B in state of A and alittle of B. Tensor product exposes3 Entanglement Entanglement isa uniquely quantum phenomenon. Entanglement is a property of a mult-qubit state space (mult-qubit system) and canbe thought of asa resource. To explain entanglement we"! ‘examine the creation and destruction of an EPR pair of qubits named after Einstein, Podolsky, and Rosen, Rosen, Suppose you begin with a qubit|yi) in zero 0) state, ia mav=H=4[ +) Then let |y) = Hivs) = $90) + 3511) = Jp(l0) +1) ‘Now take another qubit |Wz) also in the zero {0) state. The joint state-space probability vector is the tensor product ofthese two: |i) @v2) = Iiva) = 45100) + 0101) + 75110) +0111) ‘a measurement just prior to ee (for obtaining a |0) or a |1)) are:0 0 0 1 0 1 0 0 coc] 1-100) + 0/01) + Recall that just prior to the CNot the system is in the state |y} ys {10 +01, stems he ssn i wilson i) M}, Mo, = Mo,.. Therefore: PC) = (wi Wal, Mo, |v W2) = (jal Mos live) = 1o00)(% toto {9999 [odo] oo10 0000 4,0, 5,0] Sh oS J quent 2 a ‘mei mort. Ma) After measurement: Boia (ada) VE sere —= Sees She oat = |00) loooe abstract in quantum computing. The difficulty is the lack of any classical analog. ¢ useful, but imprecise way to think about entanglement, superposition and measurement is that “js” quantum information, Entanglement links that information across quantum bits, eS not create any more of it. Measurement “destroys” quantum information turning it into ‘Thus think of an EPR pair as having as much “superposition” as an unentangled set of ‘one in a superposition between zero and one, and another in a pure state. The superposition ¢ pair is simply linked across qubits instead of being isolated in one. fuzzy way of thinking about these concepts is useful when we examine telepor- we insert an unknown quantum state (carrying a fixed amount of “quantum informa tem of qubits. We mix them about with additional superposition and entanglement t the superposition we just added. The net effect is the unknown quantum ‘qubits, albeit migrated through entanglement to another physi- J. Since the information (ie. state) ends up info Generate EPR pirand isnt 0 each end “Transmit sia information re-transmitting an EPR pair tothe source and destination. The ‘ interacts with onchalf of this EPR pair creating a joint and only classical information is transmitted to the destination. to “fixup” the destination qubit -what happens when we apply the Hntum circuit depicting teleportation. Note that in this diagram single lines represent | while double lines represent classical information. [1]=al -1] have: [11)) + J6(10)—|1))(/10) + |01)) = Sosa oes |10) (alo) —8|1)) + |11) (a\2) —b)0))) yein[ a] m]3100)71y) + J01)xIy) + (10)Z\y) +|11)i¥Iy)} And of interest to us with teleportation: 1@) = $[l00)rIy) + \01)xiy) + [10)z}y) + {11)xzZ\y)] This implies that we can measure the fist and Second qubit and obtain two classical bits. These ‘two classical bits tell us wh: lat transform was applied to the third qui s ward, either applying nothing, X, Z or both Lets work through 09000000 0000000 09000000 0000000 0001000 0000100 0000000 0000000 01) = (@|Mio|9), since here M,Mio. Thus: =Haa*+b-5"]5 Super-dense Coding Super dense coding is the less popular sibling of teleportation. It can actually be viewed as the process in reverse. The idea is to send two classical bits of information by only sending one quantum bit, The process starts out with an EPR pair that is shared between the receiver and sender (the sender has one half and the receiver has the other). a’ o. eC Figure 3: Super-dense coding works by first pre-communicating an EPR pair. To send two bits of classical information one half of this EPR pair (a single qubit) is manipulated and sent to the other side, The two qubits (the one pre-communicated and the one sent) are then interacted and the resulting two bits of classical information is obtained. |) = 35 (100) +[11)) To send information apply particular quantum gates: 1 na0 . usi[§ Fy | a0) + 111)) — 4, (01) |10)) = | 0 wey key to super-dense coding is that they are or- ishable by a quantum measurement. ae AI subtle, and disappointing ‘than any classical computerever can. It illustrates, the subtle Finally, unfortunately, is solves a Deutsch’s algorithm answers the following question: suppose f(x) is either constant or balanced, which one is it? If F(X) Were constant then for all + the result is either 0 or 1. However, if f(x) ced then for one half of the inputs (x) is 0 and for the other half it is 1 (which se Fort t9 O.0r 1 is completely arbitrary). To answer this question classically, we clearly need 10 query for both x = 0 and x. <1, hence two queries are required, Quantum mechanically though ‘We will illustrate how this can be solved in just one. query. interaction of | Superposition, completely pointless problem. Phase-kick back, and interference, TY ¥®fx~)— Uy implements f, x input as (|0) +-|1)) / 2 and y as |0), then the remarkable Simp aparece GOO een two qubits since “un-changed”. The second qubit, y, acts as a unitary transform that implements this. This is illustrated in Figure 5. n we could input x as |0), and y, our output register, as |0) ‘the results of 0 and 1 simultaneously //2 and y equal 0.‘The output ja) = Silas —— Remarkable: Uy is applied to |0) and |1) simultaneously! This is known as quantum paral- ~~ Problem: sounds good, but measurement produces ether (0, f(0)) or |1 : " . .f(1)). Hence w be clever about what type of question we ask, and how we go about extracting the ane F pa uy w—fahy Algorithm uses quantum parallelism and interference to extract information ‘the solution space. ‘known quantum algorithms that provide exponential ns do, answers a question about a global property of a often called promise problems, whereby the structure of d. Usually we work with problems that of f(x) with the desired property?” of the system prior to applying Uy. Let ys) be the state of the system prior toSuppose f(x) = 0 then y@ f(x) =¥80= J, (|00)—|10)) = 4 ((0) -|1)) Suppose fx) = | then y@ f(x) =y@1 = 44 (01) -|11)) = -4(-(0)+|1)) ‘We can compactly describe this behavior inthe following formula: y@ flv) = (—1)!") ((0)~[1)) ‘Thus, Us transforms (x) 45 (0) —|1)) into: (-1/ 2) $5 ((0) —|1)) Or we can say that: [F500 +19) (0) 9] = §[(19"o (0) - 1) + (1/1) 0) -)] ‘Suppose f is constant, that is f(0) = f(1), then: 3[-070) (0) — 1) +141) (0)- 119] : 0)—|1)) +11) (10) ~ |1))] +11) (10) = |1))]iysy= SOO -1) YF £0)= 70) oe 11)((0)-|) if £0) 401) Since in our case /(0) @ f(1) = 0. f(0) = f(1) we can write this as Iva) = #17(0)@,7(1)) [2] Hence it is possible to measure x to find /(0) © f(1). Key: Note that /(0) © f(1) is a global property of f(x). Classically it would require two evaluations of /(x) to find this answer. Using a quantum computer we are able to evaluate both answers simultaneously and then interfere these answers to ‘combine them together, Another more subtle point is that the phase of the result qubit transfers to the query qubit, This is a special case of phase kick back. In effect, the query qubit acts as a control of whether or not to flip the result qubit. While the result qubit is potentially flipped by the state of the query qubit, the phase of the query qubit is altered by the phase of the result (or target) qubit! We will explore this property in more detail late, since itis the key to Shor’s algorithm. 6.1 Deutsch-Jozsa Algorithm ‘The Deutsch-Jozsa algorithm is a generalization of Deutsch’s algorithm. aa | + {0,1} and that fis either constant or balanced. The goal is determine which y it is trivial to see that this would require (in worst case) querying just over "/2+ 1 queries. The Deutsch-Jozsa algorithm answers this question we y Or} multi-qubit generalization of Deutsch’s algo-We then apply the £7®" transform. This symbol means to apply the /7 gate to each ofthe qubits in parallel, although this does not matter, The key is only that the H’ gate is applied once to each qubit). One way to define this transform is: HP") =5), Ep) This notation is rather terse, but what itis sayin, composed of components is that given any arbitrary state vector, it will be 0) +6)1) Apply H®" to get: ab) + ME 1) + HEY Ia) + 01) = Jp (a+) 0) + Jp (a2) 1) ‘When we look at the actual transform as we have been writing it in the past we find: [i a] [s]-sorem ring back to |yo) we transform it by: jt strings of size n, For example, if n = 2, this would be f(x) to obtain the state |y2): the probability ampli-tude of 2 = |0)®". Consider when f(x) is constant. Since z= (0)2", we know that (~1)**"/") is cither —1 of +1 for all values of x (since zis equal to zero x-z must be equal to zero. Further f(*) ‘was constant). Hence, if f(x) is constant the probability amplitude forz = 0)" is expressed as: 1 py =tl efor Hence when you measure the query register you will obian a zero, Since postulate one tls that the probabilities must sum to 1, if /(x) is constant, then we must measure a Ze10- On the other hand, lets consider if f(x) is balanced. Then (~ 1)#+/@) will be +1 for some x snd i for other x’s. This is where the balanced requirement comes into play. Since aif x's fare considered, and the function is perfectly balanced, the probability of obtaining 7 = {0)2" is ‘expressed as: ee Hiyoao e Where xi the set of x's such that f(x) is equal to 0 and x2 are those x's where J) is equal fo Hence you will not measure O when /() is balanced since the probability amplitudes perfectly destructively interfere to produce a probability of zero for that quantity.8: Clearly e'¥cos$ is a fully general way of expressing all values of a that can still be normalized to 1, Lets examine |a|?: \al? = ecos$-e-cos$ = cos? $ | Hence b= 1 —cos?§ = sin’, therefore: |b)? = elt) sin 9. e-itt+#) sin$ = sin? $ er Way to express an arbitrary single qubitis: ‘does not matter. Itis not observable, under any measure~ tum bit as:Any single qubit state is a point on the surface of the Bloch sphere. Any unitary transform becomes a rotation on this sphere: U = e'TR, (0). This is a rotation across some arbitrary angle 7i and a global phase shift. More usefully this can be broken down into: U=e7R.(B) Ry (9) R:(3) Rotations about the X, Y, and Z axis are related to the Pauli operators: o1 _fo -i ela a alg) r= al a fe Ei « discussing this precisely lets build upto the R, rotation, Lets assume that i (A) sin (B) ierRy(0) =e? = cos$! —i-sin$Y = cos $ ]-[ 0 -ii-sing cos? —sing | 0 cos$ ii-sing 4-534 ae] Re (0) = eI? = cos $1—i- sin$Z = [<3 3]- i-sin§ oj_fe®? 0 0 cos$ 0 -i-sing | 0 et? “Any single ub unity transform can be devormposd int the folowing —R, (B) Ry (1) R(8)I From the last two equations it should be clear that b = =e sin (J) and d = e~!cos (J) (it is possible to invert where the negative sign is placed, but itis equivalent), which gives us: cod} tsa? =1 (tin +efsn() Zen) ~ei#cos (3) -e sin (4) +e“ sin (3) e“ cos (f) = 0 —e¥ sin(f) -e cos $) +0! cosf-e-*’sin($) = 0 Focusing on the last two equations we have: eld-¥ 095 J) sin (f) = el“ cos (f)sin (3) Sadat f a) 22ers) “Or quite simply: a! —6/—c = —d" ae many solution wi three free variables, bu a lever one (fr our Purposes) is: —e-te8/2¢- sin y/2) elt ei8/2e18/2c0s (y/2) =e Hsin (2) A) arena | (/2) ~8/?sin(y/2) (1/2) cos(y/2) ] =sin(y/2) ][e*? 0 cost?) [ a 2] [ox*]-[0 5]/ R (R= (5)=| —isin(n/4) cos(n/4) cos (/4) see eal 1 “4 -t R(DR)RGD =| 9 Hie alls °] Beem yea! | a{o me N= al1 ail Note: is there a similar visualization fortwo or n qubits? 7.1 Phase traveling backwards through control operations ‘can illustrate a fundamental aspect of quantum ‘erations the target qubit is amplitude flip! but the control qubit is phase flipped by the target. This is @ key algorithm and as we will see shortly Shor’s algorithm as well. -ePsin0)|)] [cosoo) +e*sino) Asino}1)] +esin81) [cosjo)+e*sino')] =c0s(0+§) 0) resin (o-+8)|1)] Boe=cos6|0) [eosoio) +eMsino|l )| — el singlt) [oos (0+ $)|0) ~e sin (0 + $)|! | = cos60) [eosolo) +e*sino)t)] 4 ell0/44%) sind |1) [cos (a+ 3)|0) +e Ysin (0+ pI] ‘Observe the phase shift that occurs on the first (control qubit), The phase of the target, A, has now become part of the phase of the control. eflips versus bitflips gates is an X rotation