Transcript
the following content is provided under a Creative Commons license your support will help MIT open courseware continue to offer highquality educational resources for free to make a donation or to view additional materials from hundreds of MIT courses visit MIT open courseware at ocw.mit.edu so to finish up so today I want to finish up a couple of loose ends in the class the first is I want to pick up we're going to by the end of the day we'll talk about epr that we've picked up on the last couple of lectures we're going to close up Bells theorem from the very beginning so we talked about Bells inequality and we'll reprove Bell's inequality and then we'll see what quantum mechanics has to say about it um but first I want to show you something neat so a couple of times we've talked about entanglement and what it can do uh and and I want to spend the first good chunk of the class talking about what entanglement can do and also what it can't so before we get started I want to remind you of a couple things in the last lecture so the first is we have so particles can have spin half integer intrinsic angular that has nothing to do with being a rigid object that rotates but it's just some intrinsic property of angular momentum and the spin is represented it's observable it's represented by an operator in fact a vector of operators as XX Y and a c that satisfy the same commutation relations is the orbital angular moment now last time one thing we showed is that a state which is up uh or for example down at an angle Theta so if I measure the spin along this axis I will always find it's either up or down spin 12 either plus h upon two or minus H bar upon two but I can express this state in terms of the states with spin which are up or down along the z-axis and we found uh or we gave an expression and I challenge you to to follow the logic given to derive this expression um an expression for the state which is in this case down along the angle Theta uh down when measuring spin along the angle Theta um in terms of up and down along uh uh of spin um along the z-axis or along the angle zero Theta here being the declination from the vertical okay yeah yes yes yeah there's a there's a general thing for this that says at for a general angle and it's actually in the notes um which I guess weren't posted from last time but there it's in the notes um and it's also easy to dve right you just you know what the so what's the point here the point is you have a spin vector and it's Vector SX YZ and you can compute this the operator representing spin along a particular direction by taking the unit Vector in that direction and taking the dot product so that gives you a particular linear combination of SX s y and s z but from last time you know that's just some stupid Matrix and you can express that Matrix in any basis you want for example the bases of up and down in the Z Direction which is what we used last time then you can find the igen vectors of that and that's how you find these states yeah so there's a very simple generalization it just involves an extra angle Theta which is the angle around the equator okay so then sorry fi exra angle fi um so we then did the following thing I talked about the stern gash experiment and here I'm going to abstract away from the stern gash experiment the core of the Nugget the the core of it and the core of it is this suppose I have some some uh some Spin and I put it in a particular State let's say for Simplicity up in the Z Direction plus down in the Z Direction with equal amplitudes so they're equally likely to be measured and then so These Guys these two states should be degenerated energy because the system is rotationally invariant it's just a spin sitting there there's rotational invariance so we know that the energy can't depend on the Z component of the angular momentum but we can break that Symmetry and break the degeneracy by turning on a magnetic field so for stern Gish experiment we turn on a magnetic field that had a gradient but I just want to look at a constant magnetic field for the moment see what it does so we turn on a constant magnetic field that's a contribution to the energy which is minus the magnetic field do the Magnetic Moment which is a constant times the spin but if the magnetic field is in the Z Direction then this is just bz SZ so we then found yesterday that this led to a splitting that the upstate had energy plus h bar Omega and the down state had energy minus H bar Omega where Omega is given by mu bz upon 2 yesterday we had an additional term which involved the Z dependence the beta Z term but here I'm just looking at a constant magnetic field but if we know what the energies are these are still the igen states of the energy operator if we know what the energies are we know how the system evolves in time here's the initial State these are the IG States the energies are these so we can let the system evolve in time and we find that all we do is we evolve the system with phases this is as usual in Quantum Evolution if you have an expansion of a state in energy I States time evolution is just phases for each term but note that there's a simple thing this Omega the thing that determines the rate of evolution of the phase is controlled by the Magnetic Moment which is something intrinsic to the spin and the magnetic field which is something I can dial in an experiment I can really tune it meanwhile I can decide how long to leave the magnetic field on I could turn the magnetic field on for a while I could turn it off a while later so suppose I'm I leave the field on for a Time t such that Omega T is equal to Pi upon 2 okay so I turn on a magnetic field for a Time t such that Omega T is pi upon 2 or t is equal to Pi upon 2 Omega what is the state then afterwards the state subsequently s after is equal to well it's this guy with Omega t uh equal Pi upon 2 but what's E the I Pi upon 2 I right and eus I pi/ 2 is minus I so this is an i and a minus I I can pull out the I the f after is I a phase Upon < tk2 Time up Z minus down Z yeah that's cool and this is a state that we that we saw last time it's actually up to normalization which is just an I uh up or down X down in the X Direction okay so what this lets us do is something really quite cool what this tells us I all these calculations are just the stern Gish calculation but even easier because we don't have a gradient if we turn if we have the spin uh in some superp position which incidentally was initially up and down in the Z Direction This is equal to up in the X Direction and we turn on a so we have uh in in in in X we have it up we turn a magnetic field in Z and what happens is the spin pre recesses around the z-axis and we get down okay we get procession here's the important point the important point for everything that's going to follow is this if I have some spin State and I want to take it to some other spin State how do I do so well I can build a machine that does it by turning on the appropriate magnetic field and having the magnetic field press the spin okay here I found exactly what magnetic field I need to turn on with what amplitude and for how long such that I take the state up X to down X with the known phase I everyone cool with that so any unit here's the thing I want to convince you of any unitary operation that takes you from one vector to another vector or sorry from one spinner to another spinner up Z down y linear combination of up and down Z some particular linear combination of up and down X any such pair of vectors can be related by unitary transformation and any such unitary transformation can be built by some series of magnetic fields that cool okay and you can just you can prove this to yourself quite quickly but let me just state it okay so now this gives us the ability to do the following one um we have so what do we have here we have when we have spins one spins that can be put in a system that can be put in States like up Z and down Z right and two the ability to um to rotate States ability to evolve States uh from state to state from spin State One S one to S two by turning on some magnetic fields so by suitably tuning choosing our magnetic fields everyone cool with that so here's the question I want to ask what do we do with this imagine we really had this equipment in front of us what power what awesomeness could we realize and this is the entry point to Quantum Computing okay the answer to what can you do with this cool Machinery is you can build a quantum computer so let's talk about what that means so quick question do quantum computers exist yeah are they very big no okay the biggest quantum computer that's been built to my knowledge is one that has factored the number 15 although they might have done 21 at this point I'm not exactly sure 21 done yeah okay 21's done good so that's that's the upper end at the moment yeah well they're large physically it's true they take up a room but you know or at least a large tabletop an Optics table but but they're not very big in a useful sense okay and there's actually interesting reasonable people not just reasonable intelligent people um and reason that's that's a non-trivial okay the in the people who live at the intersection of intelligent and reasonable okay um uh which admittedly as as my colleague is pointing out is is not a large overlap um the uh there are people in that overlap who suspect uh that it's for some fundamental reason impossible to build a useful quantum computer in some deep deep sense I don't know whether that's true or not but we're about to engage in a discussion of what happens when you build quantum computers for n Bits where n gets large like you know millions and billions as you need for codes or for for studying images um and of course it's a little strange because these computers such a computer doesn't exist which which leads to uh my colleague and friend Scott Aronson's uh lovely comment that his job as a someone who studies quantum computers is to prove things that can't be done with computers that don't exist which is what we're about to do so so let's talk about Quantum Computing so what do I mean by qu Computing here's the basic idea let's let talk about classical Computing classical Computing says I have these things called bits which are binary objects they're either zero or they're one and we realize them in very particular way we realize them with a little magnet which you know Points North or South now it's not always exactly a magnet it isn't like on an old style hard disk uh in your computer it's something a little more fancy than just a magnet but but it's still basically a magnet and you have something north south and whether it's pointing North or whether it's pointing South down here which I'll call a zero or whether it's pointing North uh down here uh which I'll call a one uh determines whether you call this Z one and comp classical computation is your data are a set of numbers you know what's pointing down and uh qu uh the process of doing a classical computation is build a machine which is governed by uh classical mechanics okay that takes this initial configuration and replaces it with a new one F of 01 which is also some number so it's maybe it's 0 0 I don't know it's whatever f is and what you want to do is you want to perform some calculation which is some function a known function of your input variables function in this case of two bits which produces the output that you want and you build a machine that does that okay so let me give you an example I want to build a a a classical computer that takes uh takes a string of of of of binary integers 0 0 1 0 0 um and uh and performs a logical knot on each bit independently so I need to build that computer out of objects available to me okay and I need to use nothing other than the laws of classical mechanics uh that has to suffice to describe the system that I'm going to so let me give you an example of such a computer why Allan why don't you do this calculation okay so that's the input and I am now the classical computer 0011 right now that's not actually what you do in your cell phone or on your laptop that uses transistors but it's the same basic idea you build a set of machine you build a set of of uh objects you know magnetic fields electric currents that affect this calculation and so there's some relationship between what electric Fields you turn on what currents you induce to flow and the calculation you want to perform F of the input okay this is the basic idea of a classical computer with a quantum computer we want to oh and one last thing suppose we have n Bits suppose we have n uh classical bits um then we have 01 com to the n and how many numbers do you have to specify to specify the configuration of n Bits this is sort of OB well no you just need to specify the number of each bit right so we need n Bits right so n binary numbers every cool that I specify what each register is and I'm done okay let's treat imagine this is quantum mechanical let's now take instead of having bits let's take um quantum mechanical bits by which I'm going to mean A system that has two possible observable States zero and one okay so these are the possible observable States um and um to specify a general configuration I say that s is equal to um and I'm going to represent this by uh well s is equal to some linear combination Alpha 0 plus beta 1 now if I measure we know that we'll find either zero or one right if we measure the spin in the Z direction we will measure either up or down however at a given moment in time the system be in a state corresponding to a definite value of that it could be in a general superposition agreed so now here's the question how do you how many numbers does it take to specify the state of a single Quantum bit two complex numbers right it takes two complex numbers to specify the state of bit and if we have n cubits then there n of these guys well then how many numbers does it take well I have to specify the state of every possible uh I have a possible superp position for every possible configuration of the system so for example they could be all Z 0 0 dot dot dot uh zero uh plus beta the first one is one and the rest are zero plus and how many terms are there they two to the end terms right so how many numbers do I need to specify this the state of of n cubits I need 2 to the N complex numbers yeah everyone see that so this immediately leads to a really frustrating reality we have to face suppose you want to simulate using a classical computer such as sits on my desktop I wanted to simulate a quantum mechanical system of n Bits or n spin one half States evolving According to some energy function evolving According to some hamiltonian how many variables and how much memory will it take two to the N right if I've got n Bits that I want to describe it's going to take two to the N complex numbers so that's how many variables I have so if I have 10 bits literally 10 little spin one half objects to accurately specify the quantum configuration of the system I need 2 to the 10 complex numbers and that's for 10 spins how many spins make up this piece of chalk right so could you ever in an honest way simulate on a computer on a classical computer like sits on my desktop a quantum mechanical system the size of this chalk absolutely not it's completely and utterly intractable you have to come up with some sort of approximation scheme so simulating Quantum systems directly by just taking the spins and representing them with a wave function is wildly inefficient incredibly difficult interestingly the converse is almost certainly not so difficult it's almost certainly possible for a Quantum system to simulate classical Evolution quite easily and how do you know that we are yeah here we are exactly right nature apparently has no trouble simulating classical mechanics with underlying quantum mechanics quantum mechanics is the operating system and classical mechanics is the simulation it's running okay yeah in a very in a very useful sense very real sense so at this point um uh Fineman I think was the first person who really pointed this out but I don't I don't know the details of the history but the lore is that this was really F's observation that look if that's true this opens up a real possibility for computation if C if things like spins you know hard problems like calculating how spins evolve or even the motion of chalk in the room um can be done pretty efficiently by Nature if we could build a computer that used quantum mechanical bits whose variables were quantum mechanical and involved all the superpositions and interference effects of quantum mechanics perhaps we could build computers that run exponentially faster and with exponentially less memory and less resources than a classical computer would because apparently it works right because we're here as was LE said so this is the question of quantum Computing can you use quantum mechanics can you use in particular the quantum evolution of a system to perform calculations that you'd care about rather than uh uh rather than classical computation and if you do so do you gain anything do you get any sort of speed UPS is there an enhancement is anything better so here's the basic gig the basic gig is that we're going to take our system consider the following kinds of quantum systems our systems are going to be n cubits or n spins but because I don't I want to emphasize that the substrate doesn't matter the underlying material doesn't matter I'm going to refer to them purely and this is totally typical in the field as cubits rather than spin systems and the reason is it might work to use little isolated you know spinning particles with intrinsic spin that might turn out to be technologically INF feasible it it shouldn't matter at the end of the day in the same way that if I ask you how does your computer work or if I ask you to write a code you write some code in C or python or whatever with the hip kids are using these so you you write some you know some little scheme code I still love scheme um you you write some little scheme code and it performs some calculation it's defined in in just logic in in Lambda calcul abstract logic um do you need to know the substrate do you need to know whether your transistors are built out of you know silicon or germanium or or indeed whether it's running on vacuum tubes no that's the whole point of abstracting away computation that the substrate doesn't matter so we can use spin one half and that's going to be useful for us at various moments but I want to emphasize that the important thing is the actual logic of the process not the underlying substrate so here's what I need my systems are going to be n copies or n cubits okay where each bit is specified by either one represented by up or zero represented by down I'll use these interchangeably so that the the full system s is a superposition of sum over all possible well sorry I won't even write that so this is my system it's going to evolve According to some energy operation and so my input is going to be some wave function s in for these n cubits okay comput is going to be evolve with some energy operator which I've chosen to implement an algorithm in the same way that the precise series of magnetic fields that we turn on in a classical computer or currents that we allow to flow in a classical computer implement the algorithm we want to affect we evolve with some energy operator implementing our algorithm we get some output wave function s again for our n Bits And Quantum bits out okay so this is a basic gig with a Quantum computation we're just replacing strings of binary numbers and function and functions Evolution according to classical mechanics in Maxwell uh to other strings of numbers with superpositions of states superpositions of strings as it were um evolving according to the shinger equation into again General superpositions so how does this differ what are the key differences from classical computer so the key things are first off the input and output are not definite values they could in general be superpositions they do not have to correspond to a definite state in the one zero basis right these in general are superpositions of 0 0 0 one Z 0 okay so the input and output are superpositions are superpositions of the value vales will measure okay second when we do measure at the end of the day the output we get interference from the various different terms in the superposition the different terms in the superposition will lead to interference effects and so our output will be sensitive to that interference between the different terms in the superposition right it may be that we're in some pure state but it may be in general more generally we'll be in some superposition of the states we want to measure so we're going to get interference so um naively this is a disaster because this means we get probabilistic outcomes that sounds bad and so this leads to the the key difference in in Quantum or the key move in Quantum computation given that we're going to have interference and given that those interference effects are going to lead are going to um uh are going to affect the probabilities of various outcomes what we're going to want is to tune or orchestrate the interference to get a definite outcome we want to get the right out the the correct answer out rather than just probably the correct answer out now there's a slight variation of this it's not obvious at the beginning how to do that it's not even clear that you can so I'm going to show you that you can I'm going to show you an explicit algorithm that realizes this but okay that's obviously going to be tricky here's something else you can do if you if you can also focus on special problems focus on checkable problems okay what I mean by this imagine we have some algorithm that gives us an output and that output is probably uh the right answer but we're not sure there's some quantum mechanical probability that this is the correct answer to our competition there's some probability that it's not so if the calculation was difficult but it's easy to check whether you have the right answer then that's great imagine it's 10% you get the right answer but it's trivial to do to check so for example factoring numbers right if you multiply the numbers that's easy you can check whe to see whether you got the right thing so for example if imagine I I build a computer that is supposed to factor numbers I almost said Factor prime numbers that would be a boring so imagine that we build a machine that factors numbers right and so imagine its output is probabilistic so you say 15 and I say 3 * 5 you say 15 I say 5 * 3 you say 15 I say 7 * 2 at that point you're like well which one is right well you can check by multiplying right so if you have a problem which is easy to check but hard to do then probabilistic is already okay and that's just as true classically as it is quantum mechanically those are basic moves okay and so the key thing the key thing for us is that when we have n Quantum bits we have these interference effects and in particular as discussed as we started talking about last time we get entanglement effects what we're going to discover is that the key to making a good Quantum algorithm is going to be to attune the entanglement appropriately so I'm going to have to Define that for you in just a minute yeah the fact that is also sorry the checking process it may or may not be so the the check for example could be like for example imagine the the process I just gave you so the question is how do what do you do if the checking process is probabilistic but you can use a classical computer to check if you have an easy checking algorithm for example multiplying numbers together point no because good and so the question then becomes doesn't that defeat the point of using quantum computer if using classical computer to check and so here's the thing if I ask you to factor a gigantic number what do you have to do well you have to take all numbers up to its square root and you have to multiply them together and that takes forever right but if I tell you do these two numbers multiply to give you a a third number that's easy right so I use a quantum computer for the hard part and a class computer for the forward part for the checking and that's a special class of problems which are checkable these all have fancy names in information Theory I'm not going to use the fancy name Nam yeah yeah exactly I mean that when we compute suppose I give you some output and let's say the output is 0 0 plus 1 one right what are you going to get if you measure the first bit well you're either going to get zero or one with one probability or another right um when we take uh when we um when we compute the probabilities in general when we have many terms in our superposition we're going to get interference effects from different terms in the superposition right and those interference terms will tune the probability so they're not just the naive probability you know the the probability of that one thing but you know that you get these Quantum infer terms the norm squ of one Norm square of two and then the real part twice the real part of the Cross term and that twice the real part of the Cross term didn't exist classically Quantum mechanically it's important and it can change the final probability that's what I mean by the interference effects okay so let's make all this much more explicit so so far I'm just giving you philosophy and you should be deeply suspicious of philosophy in a physics classroom so again to be explicit a Quantum bit or Q bit is equal to a two-state system and again the substrate doesn't matter I could be talking about spin one half systems or I could be talking about you know ropes on a cylinder mod with winding mod two I could be talking about all sorts of things but it's it's some system with two Quantum States uh zero and one I want to emphasize this is not going to be on the final so this is this is for your moral well-being okay so so Quantum bit is a two-state system we have these two states which I'll call 0 and one and a general State the general wave function is s is equal to Alpha 0 plus beta 1 okay and the probability what this means is the probability that I measure zero is equal to Norm Alpha squ Etc okay now again many systems can be used many different substrates this is what I'm going to mean by a cubid so that's one cubit the much more interesting system is two cubits so let's study two cubits okay so in the case of two cubits what's a general State well a general state is going to be well the first particle could be zero and the second particle could also be zero or we could have the first particle zero and second particle one with some coefficient plus beta plus uh gamma uh 1 Z Plus Delta 01 okay so this is just a general Superstition now you might worry at this point look are these identical are these non-identical spins right but here's the thing I've got a spin here it's in a box I've got another spin here it's in a box I've got another Cubit over here it's in a box so they're distinguishable because they're in different places in my laboratory so these are distinguishable particles right the particle in this box is either up or down okay so these are distinguishable I don't have to worry about uh symmetrization bosonic or fironic statistics or any of that they're distinguishable and we need normalization so Norm Alpha squ plus dot do dot is equal to one and what does this mean well what this means is that for example the probability that the first particle which I'll call a is equal to zero is equal to well it's all we sum all the possible ways we could do that we have Norm Alpha squ plus Norm beta squar but these guys are whoops um this was one one thank you um and uh and more to if we're more specific the probability the first uh Cubit is zero and the second cuit uh is one is equal to I need 01 that's this guy is nor beta Square okay what I mean by two cubits but this immediately leads to a funny thing there are two kinds of states of that a that a pair of cubits could be in a very special set of states which are called separable States and these are special and these states corresponds to the full system being in the state where the first particle is in one state and the second particle is in the second state okay so for example um this could be EG uh the state 1/ < tk2 uh up plus sorry uh yeah uh 0 + one for the first particle and the second particle to be in the State 1/ < tk2 uh Z uh minus one oops is not the example I want to oh let's just call these General coefficients um a plus b and time C plus d okay so what is this equal to well this is of the form well there's going to be a term that's 0 0 and that's uh a plus a term that's 01 and that's a d with a minus minus a d uh uh plus a term that's 1 Z and that's BC and plus a term whoops minus a term that's one one and that's um BD okay and this is clearly not generic because it implies relationships amongst the alpha beta Delta and Gamma apart from just normalizability everyone to see that so it's a pretty non-trivial thing that you can write a thing as the first particle is in one state and the second particle is in the second state and here's physically what that means physically what that means oh and well actually so physically what that means is that if you're in a state like this and I ask you what's the probability that I measure the first particle to be zero do I need to know anything about the state of B no the relative if I want to know the probability of the first particle is zero I just take Norm a squ because I'm going to get plus Norm C S Plus norm d^2 and if I want to know whether the first particle is going to be B okay so that's fine so imagine so the probability of the first particle being up or down is independent of any information about the second particle right but there's another thing that's important suppose I tell you aha I've measured in the first particles in the state zero okay cool what is the state subsequent to that measurement so if we measure a is equal to zero what is the state subsequent s is equal to well that's zero and we know we've lost this so This particle is now this particular subsystem this particular Cubit has been collapsed to the state zero so we have 0 time C 0us D1 for the second particle have we learned anything about the state of the second particle have we learned anything about the state of the second particle absolutely not right beforehand what was the probability that the part the second particle was Zero Norm c s and the second particle one Norm d^2 now what's the problem same Norm C Square Norm d^ so when you have a separable system measuring one cubit tells you nothing about the other Cubit cool on the other hand consider a state which is not separable so the generic State generic states are not separable and let me give you an example of a state which is not separable which one do I want to pick yeah what the hell um s is uh one to uh 0 0 + 1 1 okay can this be written as the product of the first particle in one state the second particle in the other no because there would have to be cross terms which don't exist here right just compare this to that form so we have that the coefficient of 0 0 is a and and both so a and C must both be non zero and the coefficient of 1 one is BD so the coefficient BD must be both of those must be non zero so a c and b d are all non zero that means these terms have to exist in order for it to be separate yeah CU A and D can't vanish and B and C can't vanish and these are orthogonal States they're independent so this is not a separable state we call these states entangled and it's funny because uh entangled Ed yeah I think there's an in there um okay that's better um uh look I'm not a professor of spelling so um so we call these things it's a little bit funny to call these to give these a special name and call them entangled because the generic state is entangled right it's sort of like calling mice uh and calling mice mice and calling all other mammals non- mice right oh look well that was a bad example oh look at mice ran across the room so um hearkening back to an earlier lecture um so uh so but in any case we give these a special name and the reason we give them a special name is that these separable states do more or less what you expect classically they're no strange correlations between the state of one and the state of the other they are independent quantities but when you have a generic State there something funny happens they're entangled and here's the funny thing that happens suppose uh given this state what's the probability that I measure the first particle to be up or to be zero2 what's the probability that I measure the first particle to be one2 okay so so before doing any measurements uh the first particle is equally likely to be either 0 one but suppose I measure the second particle to be uh up okay then the probability that I measure the first particle is zero is equal to zero and the probability that the second the first particle is one is equal to one because I've collapsed onto this term in the wave pack in the wave function so measuring the second Cubit Alters the probability distribution for the first Cubit the these guys aren't independent they are correlated in a way that you studied on the problem side they are correlated they're in a correlated State we call this correlation entanglement and here's the thing that's most spooky about this and we'll come back to this in a few minutes but I didn't tell you anything about the geometry of this setup but in fact when I was thinking of that measurement I built the two little cubits in my lab and then I took one and I sent it to France and I took the other because France you know um and I took the other one and I sent it to the planet zorg right and so off on the planet zorg is some poor experimentalist huddling in the cold and uh and the the our French experimentalist makes a measurement altering the probability distribution instantaneously on zorg that should make you a little worried that sounds a little crazy we'll come back to why that is or isn't crazy and and uh and the epr uh analysis which which sort of puts Flesh on it sounds crazy uh in a little bit but for the moment let me just emphasize that while the generic state is entangled the generic state is also Al different than what your classical intuition would expect there are correlations between the particles so this is something that happens with quantum mechanical particles that doesn't happen with classical particles and that means it's something that can be used in a quantum computer that can't be used in a classical computer so let's see what using entanglement gives us for computation so let's come back to the quantum Computing problem and um how do we compute with c so how to compute okay so again as usual the way we take our input to our output is we build a machine that implements an algorithm of our choice by arranging the physical evolution of the system under time so that means picking an electric field or sorry an energy operator an electric field good Lord um so it means picking an energy operator okay so Computing is shading Evolution with a chosen with our chosen or our tuned energy operator so for example I want to build for you now a couple of basic circuit elements that you might want to use in a quantum computer so example one first example is I want to build a not gate and what not means is it takes the state zero and gives me one and it takes a state one and it gives me zero okay this is what does so how do I build an an well I can realize this uh in a in a nice way um I want to say it that way yeah good so if I realize zero as one Zer and one as 01 from last lecture then one Z oh sorry 01 01 to 1 Z so we need an operator we need time Evolution to affect multiplication by an operator that takes this Vector to this vector and this Vector does this Vector but we know what that unary operation is that unary operation which I'll call not must be equal to 0 1 1 0 and this operation takes this guy to this guy and it takes this guy to this guy yeah but I can write this as I like this I can write this as um shoot there's a phase minus i e to the I pi over 2 uh 0 1 1 0 I mean you can't stop me right so this expanded out as we've done as we've done before expanding this out uh gives me with the exponential 1 plus the thing and then all the other terms this becomes minus I * cosine of pi/ 2times the identity the 2x2 identity uh plus I S of Pi / 2 * 01 1 0 okay but what's cosine of pi upon 2 yeah zero I come on everyone that tired so cosine pi over 2 is zero up um s of Pi upon 2 is 1 that gives us I * minus I that's 1 * 0 1 1 0 pretty solid but what is this well this is the form this is trading Revolution with a magnetic field and this is just the unitary uh transformation unary transformation uh for uh for um a magnetic field well okay we can say this this is um the the X poly Matrix from last time so this is like SX unitary transformation for a magnetic field in the X direction for some time t for a Time T times the frequency Omega which is given by mu B upon 2 um is equal to uh Pi upon 2 okay just like before but for slightly different one slightly different magnetic field so my point here is that we can turn we can pick a magnetic field that does this we turn on a magnetic field for with a known amplitude with known amount of time doesn't details here don't matter so much the point is we can do it we turn on a magnetic field in the X Direction and it takes Z to one and one to zero everyone cool with that so here is a substrate an actual physical system that affects this particular Evolution I can build a knot crucial thing is I can build a knot gate and I'll call I'll represent that knot with some unitary transformation uh U sub knot so that's a useful one but it's not the most useful game because gate if you only ever impose logical knots you just get everyone angry but you don't actually get anything done so second example uh let that be a lesson to you Congress so the second example uh if we turn on the magnetic field in the y direction for a Time t for a particular time T what we find is that 0er goes to 1/ < tk2 0 + one okay and one goes to 1 upon < tk2 uh 0 minus one okay and this should be familiar this is the up X State and this is the down X State just as we talked about before so we turn on some b field and we get this so this this operation is called it has a name because it's going to turn out to be very useful for us it's taking a system that's definit in the state zero for sure right and it puts us in a superposition of 01 it's a definite superposition so it's not like we don't know what happened but it's a superposition and and you've lost certainty that you'll measure up in the Z Direction you've gained certainty that you'll measure up in the X Direction but if we do all our measurements in Z we've just taken ourselves from definite to superposition cool so that's useful because we know that something a quantum computer that can do that a classical computer can't do something a quantum computer can take advantage of that a classical computer can't take advantage of is this process of putting things in superpositions so here we've got an operation that puts things in superpositions and I'll call this hadamard and I don't know the history of why that's called hadamard but presumably there's some guy with the last name hadamard anyway the U hadamard does this and as a matrix it's represented as 1/ < tk2 time 1 one one minus one okay and there's a last one which is going to be useful for me which is called uh which is uh oh and this is represented in the following um well let me not worry about that another one is called cot control knot control knot does a really cool thing it takes 0 0 and 0 1 and 1 Zer and one one what it does is it says I'm going to apply a KN to the second Cubit if and only if the first cuit is a one so this takes me to 0 0 goes to well do I perform a KN on this bit no so 0 0 do I perform the knot on this bit no 01 now I do perform a knot on zero so I get 1 one and one one I perform a knot on this bit which gives me one zero okay so this is called a controlled knot it's a very useful thing and I'm going to uh I'm going to represent this in the following way um oh sorry I should represent all of these so this not gate first I take some initial bit and it's in some State and it and then I impose U not and it gives me an out State similarly with the hatar I take an initial state in and I apply U hadamard and I get um U out and for controll knot I now have two Q bits and uh I take the two Q bits and I do a control knot which is represented in this form which says do a knot on this guy controlled by this first bit okay and so this acts as u c okay and the key thing here is that while it's always possible to find a physical real representation of some particular unitary transformation at the end of the day all we're going to need is some truth table at the end of the day all we need is the logic that's being affected and so the details of the substrate can be abstracted away okay so what can we do with this oops right so what what can we do before we actually talk about what we can do with it let's briefly talk about what you can't do with it so what are things you can't do with these sorts of operations what can't you do and to me this is among the more surprising so here's a really surprising thing remember that what we're doing here is going to be evolving a system forward with straing Evolution and straing evolution is linear it respects superpositions it's unitary it preserves probability um and and it's well let's just focus on that it's linear unitary it has lots of other properties time reversal invariant um unless you turn a magnetic field which we do but in particular it's linear and unitary and those facts are going to constrain powerfully the kinds of operations we can effect on a Quantum system so in particular when we look at just two cubits there's a beautiful beautiful theorem that says there's no cloning and here's what that means the no cloning theorem is pretty high fluen for what it really is says the following suppose we have a system which is input x y and I want to build a machine that says look I've got this first cubit in the state X and what I want to do is I want to make a copy I want to make another quantum system that's in exactly the same state as X whatever X is so you hand me a system where the first bit's in state in state X the second cubits in state Y and I want to make a copy of X and Y is just the I mean who cares what's in it okay so I want this to go to x x okay for all y so regardless of what what data was in here I want to overwrite that data and rewrite it or that datam and rewrite it with X can you do this no right why it excellent it would violate linearity and also unitarity indeed so to see that quickly so it's easiest to see the unar I think well it violates them both but for unitarity you take a linear combination of these guys that are where the two states y are orthogonal you take the norm squar so you normalize it to one that linear combination each one goes to XX but where the coefficient is the sum of those two terms so we have uh for example x z uh Alpha XZ plus beta X Y goes to Alpha plus beta XX yeah and that's really bad because if these if x z and y are orthogonal then normalization is Alpha squ plus beta square is one but XX the normalization is Alpha plus beta quantity squar is one and in general that's not true in fact this could be zero so this violates linearity and unitarity rather badly so you can't clone this is really disturbing that means if you have a if you have a Quantum State and you want to make a copy of it you can't you can't ever make a copy of your Quantum system one copy one chance that's it no cut and paste so as you can imagine that pretty powerfully constrains things you can do so a related thing here is that there's no forgetting Quantum evolution is uh unlike an elephant it is highly uh it has a very uh well it's like an elephant I guess it it remembers very well never forgets anything um and you can see that from this this would be an example of forgetting you forgot what was in the state y you can't ever do that okay so I leave this as a challenge to you to prove this show it's a simple extension of the same logic okay so what can you do what you can do is you can entangle two CU bits and that's really the juice of everything you can whoops you can entangle so let me show you entanglement good no e um is there a question no okay good um so you can intangle and here's how you do it let's take the state zero zero so we have two cubits the first one's in the state zero and the second one's in the state zero and now I'm going to do the following set of operations to it I'm first going to impose a hadamard operation on the first Cubit nothing on the second and then I'm going to apply a controled knot okay and I'm going to see what I get out so the initial state is 0 0 right I'll write this as 0 0 after I hadamar the state is well the first bit is no longer in zero hadamar on zero gives me 0 + 1 1 so this is now the state 1 upon < tk2 0 + 1 * 1 or sorry time 0 okay also known as 1/ < tk2 0 0 plus 1 Z Now is this a separable state yes there it is separated okay and now I'm going to perform a controled n and with the control not does is it switches the second bit if the first bit is a one so what is the state after we've done this the state after we've done this is well from the first term one upon root2 from the first term 0 0 what what happens to that when we control KN well we KN this if this is one this is not one so we don't KN it we leave it alone on the second term 1 Z Plus well we we flip this if this is one and not if it's not this is one we flip it we get one one and now this is the prototypical entangled State um that I think I just erased but this is our entangled State it's not separable right if I measure the first one I know what the state of the second one is which is to say it's entangled cool so by performing this series of operations which is nothing other than a series of magnetic fields which I'm going to impose to the system I've taken a state with initial condition 0 and put it into an entangled State 0 0 plus 1 one cool and that's all we need for the first basic algorithm of quantum computation so this idea that quantum computers might be able to do things faster than classical uh computers uh floated around for a while it took a while for people to to make that sharp and uh David Deutch who is uh um very entertaining and bombastic uh speaker and he write he wrote I guess several now uh pretty entertaining books um on the topic and he sounds crazy you listen to the guy talk and he sounds nuts he sounds like you know he he's way he's just way out there the thing he's he's just God you know as a theorist you listen to him you're like go just slow down there buddy right and um so for a long time I thought the guy you know because I only knew his sort of public Persona I thought you know he's a little bit crazy I'm not exactly sure and this is why everyone thinks he's such a damn genius because this is beautiful so here's I don't know how he came up with this but he's clever so here is what's now called the Deutsch and it's really the one part or one b version of the Deutch um T TS C joa algorithm so there was a first an algorithm by deuts it wasn't didn't quite do what it was supposed to do then it was improved together with josa and they made an N particle version and then everything was awesome but here's the deuts josa algorithm okay and what it is it's a series of rules for how to make a a quantum mechanical system evolve so as to affect the calculation you wanted to to to calculate so you have to Grant to begin you have to let me pose a problem to solve that can be solved in this fashion and this problem is going to sound a little contrived and in fact it's wildly contrived it was contrived so it could be solved in this fashion but it's actually it's one that that it's one that pre-exists uh the algorithm itself so it's not quite as ridiculous so here's the problem so the statement of the problem is that someone has a function f ofx so uh let's say Matt knows a function f ofx okay now the thing is um it's extremely expensive to evaluate this function f ofx right so the way you evaluate it involve putting Quantum putting 20 kilom mirrors in superposition states with each other you have to run a whole experiment and it costs uh a lot of money to run so he charges but he charges $1 million in order to uh thank you you guys are not quite old enough so um so he he knows the function effect he charges a million dollar in order to evaluate the function you say hey Matt well look I know this is a function I should tell you um f is uh is a function uh that takes a single bit Z or one to another single bit Z one so it sounds like how hard this possibly be but in fact it's a very hard function to evaluate so you say hey Matt what's F of zero and he's like give me a million bucks and you give him a million bucks and he's like one you're like damn that cost a lot of money so okay so and now here's the question so this is not yet the problem the problem is this is f of0 equal to F of one or not okay is f of Z so F of Z is either Z One F of one is either Z or one are they equal to each other so this is easy right classically this is stupid you've calculate the function f twice you evaluate F of zero and you get a number you evaluate F of one you get a number you look at your piece of paper you say it's either the same or different right how much does that cost you two million bucks better have good funding so this is an expensive and here's what deuts and josa have to say and this is really deuts uh at the beginning it's all it's really quite spectacular deut says actually I tell you what um give me a million and a half and I'll do the computation and give you the answer at which point you think like I did previously the guy's clearly raving but um in fact he's going to make a profit and here's how he's going to do it he's going to build not a classical computer but a quantum computer using Quantum interference and entanglement to do this calculation with one evaluation and here's how it's going to work now the first thing we have to do is uh a preview or uh sorry setup in order to do this calculation uh you need two things first off you need to you need uh Matt to be able to build to to evaluate his function in a way that respects quantum mechanics okay so in particular Matt had better be able to do his experiment if I give him a superposition so he' better be able to affect the calculation in a quantum mechanical way the same way that we implemented a not Quantum mechanically or the controlled not Quantum mechanically or the hadamar with some set of magnetic field he must be able to implement it quantum mechanically otherwise it's not an interesting function and let me just point out that any function you can think of can be implemented Quantum mechanically because you are quantum mechanics okay you're just not elegant implementation no offense you're just not an elegant implementation of the quantum mechanical uh uh computation so what I want so setup is Matt needs to be able to give me Matt provides a unitary transformation unitary operation U subf that takes two cubits X and Y y okay to X and F ofx + Y where what this means f ofx plus Y is plus is addition mod mod 2 so what this says is if Y is Z then this gives me f ofx plus 0 if f ofx is z that's 0 plus 0 so that gives me zero if FX is one this gives me 1 plus Z that's one on the other hand if y or sorry if Y is 1 then this gives me if f ofx is z this gives me 1+ 1 which is z and if F ofx is0 this gives me 0 plus 1 which is one everyone cool with that okay yeah question we know that actually how do how we know that actually mechs you mean what if quantum mechanics is only an approximate description of the sorry no sorry like um what if quantum mechanics like in is inescapable isim yeah this is an interesting question so it so it's tempting to think that this is a philosophical question but it it turns out not to be in a way that will be made sharp in about 10 minutes with bells inequality but um a good answer a complete answer to that question remains uh open and I think it probably always will but let me rephrase the question slightly tell me if this is in acccurate statement um look um at the end of the day uh what we're doing is we're going to develop a model where quantum mechanical calculation does something in particular and that may or may not be a good model of the real world and in particular whatever the actual thing the actual system that we're studying does may or may not be well described by um by that quantum mechanical model um so can we check whether or not it you know it is and I is that more or less the question yeah it's it's a hard thing to make for so the problem is all we can ever do is say that our model is a good model or a bad model on the other hand we can do the following and this is the really neat thing you might say look underlying quantum mechanics is going to be something more fundamental that's going to lead to slightly different results in exactly the sort of situations where we're going to care about Quantum computation with large numbers of bits okay um and if you tell me just a little bit a little tiny bit about what properties that underlying description will have that becomes an empirical question so for example if you say look I suspect that underlying the Quant mechanical probabilities is some classical probability distribution over an hidden variable that you have not actually measured what we're going to find out is that we can rule that out experimentally just that extra little assumption that there's an underlying hidden variable a secret probability distribution on some variable that we just haven't observed yet that is enough information about the the system to rule out that model amazingly so I think we'll never have a full answer to your question um but it all we can do is work and see uh how well our model fits and so far nothing's ever disagreed with the quantum mechanical description um let me hold off on questions just now because I want to um but it's a good and interesting question that's a hard one to deal with um by which I mean it's an open question so Matt provides for us an operator that allows us to calculate F ofx Now remind remember you might have said well look why not just take X and why not have build a machine that takes X and gives you f ofx could you have done that no well it's not exactly no cloning but let me leave this to you as a fun thing to think about why are we why do we need this carrier bit as well okay so there's our setup Matt provides this function for us U and here's the algorithm so the algorithm and it's a series of steps one by one we do them we perform these operations on our on our Cubit um so here's what de Deutch says de says start input a state s is equal to0 one first Cubit is zero in the state zero the second Cubit is in the state one for sure we Implement that with our boxes or however we want to implement it so we find ourselves in a definite State you know hard soft so we take a hard box and we take a soft box and we pull out the hard one and the soft one okay one hatam each had on on both bits both cuits okay so what does this take us to it takes us to S is equal to well the zero goes to 1/ < tk2 time 0 + 1 did I erase hamard no good uh there's hadamard um so it does this does this so it takes uh the first one to 0 + 1 and it takes the second one to 1/ < tk2 uh 0 minus one cool so at this point this isn't very interesting what we've done is we've taken it from one superposition to a different superposition and it doesn't seem to have anything to do with f in fact we haven't measured F two apply F so we apply our operation u u subf and this takes us two ah well this is a sort of entertaining one um if we take this one overun so I'm going to rewrite this um in in a slightly simple for this is 12 uh 0 times uh 0 * 0 - 1 1 plus 1 * 0 - one okay and the reason I'm doing that is we're going to apply UF and UF our function f uses that first bit as um a control bit for the second so here's the control bit for the second so we play UF and um this gives us2 I'm going to actually do this on the next board because it's going to be gigantic okay so this gives us20 so that first one this is going to take this and give it 0 plus F of 0 and 1 plus F of 0 so times uh F of0 uh plus 0 minus F of 0 time plus one plus for the one uh this is going to be times uh F of f of one now plus 0 minus F of 1 + 1 okay now here's a crucial step this is equal to and want to note the following look at look at this particular guy okay so for that particular guy suppose F of Z is one or sorry is 0 if F of 0 is 0 then this is 0 plus 0 which is 0 so f f of 0 is equal to 0 then this gives me zero and this gives me 0 + 1 which is one 0 - one on the other hand if F of 0 is equal to 1 then we get 1 + 0 which is 1 and here we get 1 + 1 which is 0- 0 which is equal to minus uh 0us one yeah that so I can write this as -1 to the F of 0 * 0us one cool that this is just a little exercise in binary arithmetic so I can write this first term this is 12 min-1 to the F of 0 0 time um 0 minus 1 okay so that's from the first one and exactly the same logic is going to apply to the second but now F of one instead of f of zero so plus - one to the F of one * 1 * 0 - 1 okay now I want to point something out to you if f of0 is equal to F of one then what's true of f of0 plus F of one well if they're the same exactly then either it's 0 plus 0 which case we get zero or it's 1 plus one in which case we get zero so this is0 if it's the same and one if it's not okay so we could either know them both or we can measure F of 0 plus F of one so notice what happens here this is equal to 12 and now I'm just going to pull out a factor of f of0 minus one to the F of0 times well both of these terms have a 0 minus one on the second bit so the second Cubit is in the state 0 minus one right everyone cool with that so this is equal to for the first Cubit Z Plus 1 * -1 to the F of 0 I've pulled out to square it plus F of one 1 time 0 - 1 and here's the quantity we wanted to measure if this is zero then they're even then they're the same and if it's um if it's one then they're not the same so now at this point we just forget three forget about the second Cubit oh Lord forget about second Cubit and we get the state so forget about the second Cubit just does this just focus on this guy and now if f is Z Plus F of 1 is one or sorry is zero so that they're the same this is 0 - 1 0 is 0 so we get 0 Plus + 1 so same then our state is 0 + 1 and if they're different then we get the state 0o Plus or sorry 0 minus one everyone agree with that so if they're the same we get 0 plus one we get different we get 0us one still doesn't work for us because if we measure what's the probability we get zero here2 and probability we get one is 1 12 similar if we measure here 0 1 12 1 12 on the other hand these states are familiar to us because they're what you get by hatam Maring so why don't we take these from these states to these by doing the inverse of the hadamard which it turns out is hadamard itself so four hadamar the first bit had the first bit and we get the output is uh the state s out is equal to 12 1 + -1 to the uh F0 + FS1 z uh plus 12 1 - -1 to the F0 plus FS1 one and now if F0 plus if F0 one are the same this is a zero we get 1 + 1 and we get just zero and this is zero because this is one this is one they subtract we get zero they're the same you get this state zero properly normalized if they're not the same you get this state one properly normalized now if we measure if we measure zero we know they're the same and if we measure one we know they're different and so with absolute certainty now five measure the first CU bit and we we get zero implies the same and one implies different and we did all of this with a single evaluation of our function f right here right this is where we applied our function evaluation we've applied the function evaluation once and we deterministically get the result whether they're the same or different so with one call to Matt to my Oracle with one call to mat which cost me $1 million we get the answer to whether it's the same or different and that's a factor of two faster than the best classical algorithm but that's not so satisfying this was supposed to be exponentially better and so that's where josa comes in and together with Deutsch deut and josa show The Following that there's an exactly analogous problem for n cubits wow there's an exactly analogous problem for incits deuts Joseph problem okay and now how many different how many different strings of integers could you put in there are now two to the end possible States and if you want to know whether f is the same for all of them the worst case scenario is you measure that you evaluate F on the first possible combination 00 0 and you get some number you measure F on 0 one and you get the same number and you just keep doing that forever and you still don't know if they're all the same until you get to the very last one right so it's 2 to the N it's order 2 to the N is worst case scenario but typical scales are of order 2 the N so classically it takes an enormous number whoops of observations but the quantum do joa algorithm sorry that was n bit D josa and now the N Cubit D joa problem one Quantum operation and you get a deterministic result and all of this one right you evaluate it once and you know you've solved the problem so instead of two to the N operations it takes a single one and now for a large number n of bits for example for large integers dealing with very large numbers uh this is dramatically exponentially more efficient than the classical algorithm so this is the basic at this point you know people start really seriously thinking about Quantum computation whether you could get it to work and um you know how much how much juice you could get out of actually building such a quantum computer and this is developed into a whole theory in in the theory of comput whole field of the theory of computation um the thing I want to emphasize is that the crucial move is observing that in quantum mechanics you can entangle degrees of freedom The crucial move is observing that you can entangle degrees of freedom Quantum mechanically and that's what gave us all of the nice effects we have these interference effects these interference effects lead to the deterministic outcome being correlated with the result of the computation it's the interference is is is crucial and this brings us to the last the last point which is exactly what's so troubling about entanglement and so here is where EV araman and Einstein pkin Rosen say the following they say look there are two things that are deeply upsetting about this entanglement Story the first and let me just give you a prec a precise experiment they say they say let me give you a precise experiment that embodies all the weirdness of this suppose I take two of these cubits and I put the first one in this and I put the two cubits in an entangled State up up plus down down okay and let's normalize this with one upon Ro tk2 so there's our state and then we take the first CU bit so there's our two bits and we take the first CU bit and we send it some far away and someone named Alice is sitting here and is holding on to that bit we take the second one we take it far away and someone named Bob conventionally is sitting here and holds a second bit now given this initial configuration what is the probability that Alice will measure the spin to be up her spin to be half right and half down similarly Bob half up and down once Alice has measured the state to be up right immediately she knows something about Bob's spin Bob's state will be up because I chose this one I could have chosen the other one which is the more popular e pair so it's so Bobs will also be up any now if you look at the list if you do this over and over and over again their list is just some random list of ups and downs ups and downs but they're exactly correlated amongst each other so at this point epr are really upset because they say look there are two possibilities either there was an answer to the question all the way along of whether alyses was up and Bobs was up or Alice's was down and Bobs was down or there's some deep non-locality in in the universe such that a distant measurement causally disconnected can uh can be uh can have an influence on Bob's states such that they're correlated this may seem random but it's certainly not random because it's correlated with Alice even though Alice is wildly disconnected distant distant distant Observer nothing could have traveled across that time in you know across that distance in the time it took uh to do the measurement so there are sort of three responses you could take to this the first response is look there isn't a problem here it's just saying that quantum mechanics is insufficient there really there secretly there's a hidden variable a variable you haven't observed yet a property of an electron that determines whether it's going to be up or down early on and the fact that it looks probabilistic just means that there are some classical Dynamics for this hidden variable that effectively is is probabilistic like a particle moving in a fluid like dust pollen grains in a fluid it just moves around randomly so but it just looks random it's not actually random and it it's that's because there's an underlying classical mechanism controlling the probability distribution the second version is a Quant mechanical version the the second interpretation is to say look this may look upsetting and I grant you that it looks upsetting but I'm a Quantum mechanic and Quantum mechanic works like a champ and I'm not about to throw it out and say that there's some secret hidden variables um it just works so just give up on your naive Notions of locality Let It Go and just do the quantum mechanical calculation practicing physicists look at this in just yawn right if if you're a practicing physicist you just forget it like obviously it works so there's no more conversation to be had so um so meanwhile there's a second version of this which is slightly more disturbing suppose Alice measures up and this is all in the Z Direction suppose Alice measures up and Direction she thus knows that Bob's particle is up in the Z Direction but simultaneously Bob could measure spin in the X Direction and determine that his spin is up in the X direction as well at that point epr say look we've measured empirically we've measured that the particle is both up in the Z Direction and up in the X Direction it's just we meas we did that measurement using entanglement but at the beginning of the day at the very beginning oh shoot I erased it dang that's not what I wanted so at the very beginning of the day we had that SX and x uh s z uh don't commute so you can't have a state with definite SX and definite s z you cannot possibly it doesn't mean anything to say so Einstein wants to say that this is because qu mechanics is great but incomplete the rest of us want to say that no it's not but that sounds like a that sounds like um you know that sounds like a philosophical question and that's the way it was treated for a very long time until you come along to Bell and B made a beautiful observation B said look telling me that there's an underlying classic description telling me that there's an underlying probabilistic classical description tells me enough to make this an empirical question because it's saying that the the statistics the random statistics for Bob and Alice are correlated uh by a classical Dynamics rather than independent so here's Bell's version so remember at the very beginning we talked about Bell's experiment we said consider three binary properties a b and c the number that of some set that are a and not B plus the number that are B but not C is always greater than or equal to the number that are a but not C and the way we proved this was just by noting that if these are classical deterministic binary properties then a not b means a not b and c or a not b and not C and ditto for each of these other guys and we ended up with an expression which was number that are a so using that logic leads to number that are a not b and c plus the number that are not a b and not C is greater than or equal to zero and that's clearly true you can't have a number of elements be negative so this is trivially true Quant mechanically but now here's the experiment I want to do I want to do actually the epr experiment and here's the experiment I want to run Alice is going to measure up or down at zero and Bob is going to measure up or down at Theta Alice is then going to measure up and down at Theta and Bob is going to measure up and down at 2 Theta and the third experiment is going to be Elsa is going to measure up and down at zero and Bob is going to measure up and down down at 2 Theta so a is up or down at um up at zero B is up at Theta um B is up at Theta and or sorry this is going to be down at Theta not B is down at Theta B is going to be up at Theta and C is going to be down at 2 Theta and a again up at zero and not C is down at 2 Theta okay so we can rephrase this as the probability that given that we are um up at zero what is the probability that we are subsequently down at theta plus the probability that we are up at Theta and uh what is the probability that we are up at Theta and subsequently down at 2 Theta and then the probability that we are up at zero and down at 2 Theta using an e measurement where one is performed by Alice and the other is performed by Bob exactly as epr wanted and we computed did this last time in fact I just erased CU I'm excited about this I guess I just erased the wave function that we needed the state we needed and the state that we needed was that if we are down at the angle Theta then this is equal to cosine of theta upon 2 uh down at zero plus I sin uh Theta upon 2 up at zero and this is enough to answer our question this is the Quant mechanical prediction what's the probability that given that we're down at the angle Theta we're up at the angle zero well if we're down at Theta the probability that we're up at zero the coefficient is I sin Theta upon 2 and the probability is the norm square of that expansion coefficient so the probability is sin squ of theta upon 2 similarly the probability that we're up at Theta and down at 2 Theta well by a rotation by Theta this gives me exactly the same thing so it's again going to be sin^2 sin^2 Theta upon 2 probability we're up at zero and down at 2 Theta well just stick in a factor of two for Theta everywhere and that gives me sin squ of theta now I ask you is the left hand side always greater than or equal to the right hand side and this is easy to check let's do this for very small Theta a very small Theta sin s Theta so for small Theta much less than one sin Theta s is Theta upon 2 the angle squared which is equal to Theta upon 4 and the next one is the same thing plus Theta upon 2^ 2 which is equal to Theta 2 upon 4 Theta s upon 4 so Theta 2 upon 2 and the right hand side is sin which is and is upon 2 greater than equal to Theta certainly not so quantum mechanics predicts that if we do the epr experiment using these observables repeatedly and build up statistics what we will find is an explicit violation of the Bell inequality and what that would represent if it were actually true if we actually observed it would be a conclusive empirical proof that there are no classical definite configurations underlying the probability of Quant mechanical events it would say that it's impossible to build a classical the with hidden variables that are randomly distributed such that you reproduce the predictions of quantum mechanics we see already that it doesn't agree with classical mechan or with quantum mechanics the question is does it agree with the real world so someone has to build this experiment and check and this was done by Al spay and it violates BS inequality there is no classical description underlying quantum mechanics the universe around you is inescapably probabilistic it evolves in a deterministic fashion through straing our Evolution but when we measure things we meure results with probabilities and those probabilities cannot be explained through some underlying classical Dynamics if there's something else underlying quantum mechanics whatever else we know about it is it is not classical and this property of probabilistic evolution or probabilistic measurement is an inescapable empirically verified property of the reality around us and that's quantum mechanics thanks guys yeah