You don’t know how Quantum Computers work!


Ever since I first heard about quantum computing, there was something that bugged me about it. The story goes like this: All operational
computers today are classical computers. They store and operate on information represented
as 1s and 0s (bits) to solve intensive, mathematical problems. Quantum computers, which are being
developed in labs right now, are fundamentally different. They take advantage of quantum
mechanics to do multiple calculations at once. This is done by putting quantum bits into
superpositions so they can be 1 and 0 at the same time. But you have to be careful with
superpositions, because as soon as you read them, they collapse to either a 1 or 0, with
a known probability. If all that was a little too fast, feel free to check out Building the Bits and Qubits, as it explains all of this more clearly. Don’t worry, I’ll wait here. Done? Good! Now we’re on the same page.
Here’s my problem: if you put a qubit in a superposition to do many computations at the same time, wouldn’t you only obtain one of those many calculations when you measure your answer? You can’t get many answers at once, since reading the answer to a quantum computation would destroy most of the information your qubits are storing! How can quantum computers
even be practical if this happens? The problem is that this story doesn’t completely explain what quantum computers actually do. Why don’t we just take a look? To understand what’s happening here, let’s
actually build a quantum algorithm. Qubits can be visually represented as Bloch spheres.
1, 0, and everything in-between. Today, we won’t be needing all of these qubit states, so let’s simplify this representation to a slice of this sphere. Let’s call it the Bloch circle. And just a little aside for those of you following the math, the Bloch circle illustrates all possible qubit states where the probability amplitudes have no imaginary
component. And don’t worry if none of that made sense. this is just so those who already
know the math can follow along. Now let’s begin. Say that I hand over to
you this quantum circuit. It takes in two qubits, and spits out two qubits. Let’s call these x and y. Qubit x always passes right through this circuit unaffected, so if it was a 1, it comes out as a 1, and if it was a 0, it comes out as a 0. Now, here’s where you come in. I’m not gonna tell you what this circuit does to qubit y. I will only tell you that it will do one of 4 possible things. Case A: it does absolutely nothing to it,
just like qubit x. Case B: it flips it, so if it was a 1, it comes out a 0, and if it was a 0, it comes out a 1. Case C: it flips it like before, but only if qubit x is a 1. If instead qubit x is a 0, it does nothing to it. Case D: it flips it only if qubit x is a 0.
If instead it’s a 1, it does nothing to it. So to review: In case A, never flip y. In
case B, always flip y. In case C, only flip y if x is a 1.
In case D, only flip y if x is a 0. In cases C and D, whether qubit y flips or
not depends on qubit x’s value, so let’s call these the needy cases. “Umm, qubit x, sorry to bother you… What should I do to qubit y?” This is not true for cases This is not true for cases A and B, so let’s call them the laid-back cases. They’re the cases where the circuit goes: “Yeah, I don’ really care what you say qubit x, imma jus’ go do my own thing.” By the way, remember this exact terminology.
It’s important! …naw I’m kidding, these names are completely arbitrary! So, with that in mind, here is the circuit
I’m giving to you. By toying around with it, can you figure out if it’s a laid-back
case, or a needy case? Oh and by the way, I rigged the circuit to self-destruct if you
try to open it. So all you can do is pass qubits through it to see what happens. And,
it also self destructs after you use it once. So you only have one chance. Good luck! Okay, that was a little unfair. Let me show
you how to do it. What would happen if we pass in both qubits x and y as 0s? In case A, we would get two unaffected 0s
as expected. In case B, nothing would happen to qubit x,
and qubit y would flip vertically to a 1. In case C, qubit y would only flip if qubit
x is a 1. Qubit x is not a 1, so it’s unaffected. In case D, qubit y would only flip if qubit
x is a 0. Qubit x IS a 0, so it does flip. So, the result will only ever be 00, or 01.
If you actually did the experiment, and got 01, that means that qubit y would have flipped,
and you would know that this circuit is either case B or case D. But which is it? Would y
have flipped because it always flips, or would the circuit have been listening for instructions
from qubit x, and it happened to say to flip qubit y? Would something else have happened
if x was different? We could then then try passing in x as a 1 instead to see what happ… Whoops, the circuit gets destroyed after a single use, remember? And, we still don’t know if the circuit was a laid-back or a needy case. In fact, if you just pass 1s and 0s
into this circuit, you would always need to use it at least twice to get an answer. But
we can’t do that, because I’m a mean person and made it explode after 1 use. Let’s try thinking of something else. Now,
the problem we had before is that we didn’t know if qubit y flipped, because it always
flips, or it only did that because qubit x was a 0. We didn’t know if it would flip
if x was a 1. So, let’s see what would happen if we put qubit x into a superposition, so
that it’s a 1 and a 0 at the same time. That makes sense. With that, we can see what
would happen when qubit x is a 1 AND when qubit x is a 0 at the same time. To do that
let’s use the Hadamard gate. The Hadamard gate is useful because it can take qubits
into and out of super positions. Let’s use it to take a qubit 0, and turn it into a qubit
right. Qubit right, is half-way 0 and half-way 1. Or as many like to say, it’s 0 and 1
at the same time. If you’re doing the math, qubit right represents this quantum state. That should do it. We’re testing both scenarios
at once. Let’s see what would happen. In case A, nothing ever happens to either
qubit, so they are unaffected. In case B, qubit y flips no matter what qubit
x is. So it flips. In case C, qubit y flips only if qubit x is
a 1. But it’s both a 1 AND a 0. So does it flip? Does it both flip and not flip at
the same time? Something interesting happens here. Qubit y goes into a superposition, but
it isn’t tied back to its own opposing state. It’s tied back to the state of qubit x.
In other words, qubits x and y come out in an entangled state. They can no longer be
considered separate from each other. Now, measuring either of these qubits will cause
both to collapse to 1 or 0. In this particular case, they will always collapse to the same
value. Here’s the reasoning: If we found that qubit y flipped, it would because qubit
x was a 1, since that is the only scenario in which qubit y WILL flip in case C. Similarily,
if we found that qubit y didn’t flip, it would be because qubit x was a 0, since that
is the only case in which qubit y will NOT flip. In case D, qubit y flips only if qubit x is a 0. Again, qubit x is kind of both 1 and
0 at the same time. In this case, the qubits come out entangled again, but they will always
collapse to opposing values. If qubit y flips, it was only because qubit x was a 0. If qubit
y doesn’t flip, it was only because qubit x was a 1. At first, it looks like we may have solved
the problem. The qubits come out entangled in only the needy cases, and not the laid-back
cases. So all we need to do now is check whether or not the qubits are entangled, and we have
our answer! So what would happen if we tried it? What would happen if we put in qubit right
for x, and qubit 0 for y. Say we find that qubit y came out flipped. We would know that
we don’t have case A, because it never flips in that case. And let’s say qubit x comes
out a 1. Wait, was that because it was entangled and became a 1 when we measured qubit y (as
in case C), or was it because it randomly collapsed to a 1 from an entirely separate
superposition (as a possible case B). Like last time, we don’t know if we have a laid-back
case or a needy case, because there is simply no way to check if two individual qubits
are entangled or not. This is the problem I was talking about earlier. Sure, you can
do multiple calculations at the same time by putting qubits into superpositions. There
is information about two separate calculations in these qubits, but you lose that information
when you read them. Well it turns out, there is a very clever way of getting around this… Okay, enough stalling! This is the solution.
Set x to qubit right, and set y to qubit left, both in superpositions. If you’re doing
the math, qubit left represents this quantum state. In case A, nothing happens.
In case B, y always flips. But flipping qubit left top to bottom just results in qubit left
again, so oddly, nothing happens. In case C, the strangest thing occurs. The
classical description of the circuit says that qubit x never changes. But, somehow,
the circuit actually flips qubit x left to right, even though nothing is really supposed
to happen to it! The same thing happens in case D. As for how this happens, hold that thought
for just a sec. We’re really close to solving the problem. Now, one last obstacle remains. As a reminder,
we are trying to figure out whether the circuit is laid-back or needy. In the laid-back cases,
qubit x faces right. In the needy cases qubit x faces left. But we can’t measure this
qubit yet, because superpositions collapse randomly to a 1 or 0. All we need to do, is
pass it through the Hadamard gate once more. Now, qubit x only comes out as a 0 when the circuit is laid-back, and only comes out as a 1 when the circuit is needy. Problem solved. So let’s do that. Qubit x comes out
as a 1. Therefore, we know that this circuit is either case C or case D. But no matter
what, it was a needy case. This is called Deutsch’s algorithm. It depends on the fact that qubit y is specifically
facing left. This mechanism results in the ability for qubit x to flip left-to-right.
If qubit y was facing right instead of left, nothing ever happens to qubit x. It’s a
lot like how multiplying a number by 1 keeps you where you are, but multiplying by -1 flips your sign. Likewise, qubit y facing right keeps qubit x where it is, but qubit
y facing left flips qubit x left-to-right. But why is this? Left and right are both half-way
0 and half-way 1. Why should something so drastically different happen under each case?
Well, left and right do each collapse to 1 or 0 with the same probability. But mathematically,
they are as different from each other as 0 is different from 1. Qubit right is half-0-half-1
in-phase, and qubit left is half-0-half-1 out-of-phase. They are, in a sense, “opposites”.
For those following the math, [0,1] and [left,right] each form an orthonormal basis. So, while the in-phase state has no effect
on qubit x, the out-of-phase state has the ability to flip it. And if this still doesn’t make much sense, which it probably shouldn’t, just remember, as Richard Feynman once said:
“If you think you understand quantum mechanics, you don’t understand quantum mechanics.”
After all, the mathematics we are dealing with here involves transformations through
4-dimensional vector spaces, so it’s a little tricky to come up with good analogies. This
terminology zoo and this flipping piece of card are honestly all I got. And if any of you can, please research quantum
computing, go through the math, and find a better way to explain this. It really bugs
me when I get the math but I can’t explain the essence of it. What kind of channel is
this? Frame of Matrix Algebra? Quantum Mechanics… you have defeated me. Moving on… Since we’ve solved our problem,
here’s what the individual circuits actually look like on the inside, made out of elementary
quantum gates. And here are their classical analogs. So, what have we learned from all of this?
That there is no hope in understanding quantum weirdness? I don’t know the answer to that
question, but that isn’t the point of this video. The point is that the answers extracted
from quantum computations contain no more information than the answers extracted from
classical computations. With classical bits, running the circuit once, only ever narrowed
our answer down to 2 possible cases. Using a quantum algorithm did not change this, but
it WAS able to use quantum mechanics to cleverly twist the question we were asking to result
in information that we found more useful. Now, you may be wondering one more important
question. What was the point of this? If I already built this quantum circuit, I obviously
knew that I gave you a needy case. Why did I keep that fact secret and make you figure
it out? When would finding out something that was already knowable in the first place be
practical? Well, this is precisely what a quantum computer does best. Take this example. I mentioned in the last
video how quantum computers can break current computational security techniques. Today’s
internet security depends on the fact that is very hard to factor a large number which
is a product of two primes. It’s not so bad with a small 2-factor number like 15,
but it’s practically impossible for a classical computer to factor large ones. If someone
can factor this giant number, the security fails. How does a quantum computer break this?
It turns out that it’s possible to create a quantum circuit which implements some modular
arithmetic with this giant number. It’s a lot like the circuit I gave you. If we pass
in superpositions to it, we are able to obtain a result, which we can then use in some classical
arithmetic to quickly find the giant number’s prime factors. Classical security has been
broken. This is known as Shor’s algorithm. It’s been done as a proof of concept, but
it shouldn’t actually be a real threat anytime soon, because the physical implementation
of quantum computers is very difficult, and is currently being researched. But it does
put pressure on humanity to develop more robust security techniques. Maybe quantum security
techniques? In each example, we had a problem with an
answer that was computable, and was already a property of the very quantum circuit that
was already built. We do not need quantum computers to solve these problems in principle.
They just enable us to solve some problems, such as these, in less time. There’s another, more optimistic one called
Grover’s algorithm, which finds the location of an item in an unordered list. Classically,
because this list is randomly organized, you would need to check each item, one by one,
until you found the answer. Grover’s algorithm allows us to find it much faster. If we had,
say around 10 000 items, instead of performing up to 10 000 queries, you would only need
a measly 100. This opens up numerous possibilities in database theory. There are more quantum algorithms, but I think
we’ve already learned the essentials today. Quantum computers cannot do anything that
classical computers can’t already do in principle. Quantum computers cannot give you
multiple answers to the multiple computations it may be doing at once. What they can do,
is twist the questions we can ask it, to aim the answer down to something more relevant. Quantum computing is hard, which may be part
of the reason it’s still being researched. But be on the look out for computationally intensive software being partly running on quantum processors. They may be just around the corner. KABOOM! 😀

100 Replies to “You don’t know how Quantum Computers work!”

  1. D wave = the Devil & his minions, coming thru, injecting our phys.bodies, gaining our Souls- thru Waves. That simple. 😊

  2. I still don't understand quantum computing at all, but I feel I don't understand it at all with a bit more clarity after watching this video. At least it has a go at saying what a QC's inputs and output might actually be. All the other attempts seem to just really bang on and on about having a 1 and 0 at the same time – and I want to shout out YES, YES, YES I HAVE GOT THAT PART, PLEASE SHUT UP! So "good on you Frame of Essence" – although I feel that Universals suggestion of alcoholism does really tempt me.

  3. 9:58 – Why is the X qubit constrained to be a 1 if entangled with the Y qubit? I understand everything else up to this point.

  4. What if you put a superposition qubit into a normal computer? What if you put a lot of them? Will they randomize into 0 or 1 and make a random program of 0 and 1? (String, bit, bit)

  5. Some kind of voodoo magic, just cant grasp it….. hoo leee shit i just got it 😀 i was about to comment that you can caculate all thous things so why to go bother and "simulate" your math in real world, so arnt thous computers designed to do exactly the thing? They like physicaly put thous equasions and solve them instantly? Does this makes sence? I dont know anymore, freaking voodoo..

  6. When you’re translating it to ones and zeros could we maybe add positives and negative’s or maybe colors to better understand it I don’t know that’s just a thought not getting rid of the ones and zeros but maybe augmenting them to maybe there’s a ladder that’s in the translation so we have a better understanding I don’t know maybe I’m completely wrong and I’m trying to understand quantum computing it’s very hard and thank you for an answer and I appreciate you reading my really long question and have a nice day and happy Fourth of July

  7. if then if then if ……..
    if you have 8 bits with unknown state
    how many if's you will need to know its value?

  8. I can see things you can't,
    so I am a scientist but you.

    now this quantum bullshit was 3 years old.
    what's new in your quantum computing?
    any progress achieved there?

  9. Ooh! I think I kind of get it! Is the reason why the spin of qubit X can be flipped have to do with us interacting with it / measuring it? So it is actually impossible to build gates that depend on the states/spins of multiple qubits and not have them interact with each other and possibly changing? And it just so happens that in this case, with this gate, qubit X can be flipped right to left by Y when Y is left?

  10. I don't understand how the difficulty of computing things together when qbits are either up or down is solved by thinking that the qbits are either left or right… -_-

  11. In IBMQ, When scenario c is executed, it returns this: 58% chance of getting 00000010 and 42% chance of getting 00000000.

  12. This is by far the best explanation on YouTube of how quantum computers work. You should have MUCH more subscribers. Thank you 🙂

  13. Returning to this video 6 weeks into my quantum mechanics course at uni has been very interesting. 🙂

    It's made entanglement a little more clear

  14. Hey great video, but I don't understand how is quantum data fed into the quantum computer. For example in Grover's algorithm, it easy to understand what's happening in terms of phase and mean rotation, but how is it the data fed? This is a huge question to which I am not able to find answer. Please make an end-to-end video if possible.

  15. before this I thought quantum gates are stable. now they are in superposition and can collapse. WTF? does it mean that quantum computers have to construct circuit each time it runs an algorithm?

  16. i love this channel! <3
    one question: how would it feel like to write a program for this s***.
    i mean, we will always get one of the two things(0 or 1) as outBit so programs must look like the ones we write at the moment, tho the inBit would look… weird.

  17. Yep. we're doomed! Just matter of time before a smart person shows what a dumb rich villain /president can do with this sh**..

  18. Don't press "Read More"

    You have good luck in 2019, 2020, 2021, 2022, 2023, 2024 and 2025
    Like to Activate

  19. Thank you, this is awesome. I finally get some grasp on how it actually works, and not just that it "works exponentially faster"…

  20. THank U! much better explanation than most ive seen, I absolutely hate when people say the doing all computations at once thing because of this exact reason that you can only measure one thing! So you really need to use entanglement and interference to get the magic juice out… but heck you explained deutsch's algorithm in a youtube video! no skimping on the details here :^) 'Between' is also my favorite terminology to use for other points on the bloch sphere too, its so much more explanatory I think than just saying 'and'

  21. I just do not get it how do you get more or different of an answer from quantum computing. Can anyone help me?

  22. If measuring a qubit causes its superposition to collapse to one value or another, how do your quantum computer's gates measure qubit X in order to determine how qubit Y should be affected, without affecting qubit X's superposition and causing it to collapse prematurely prior to the Hadamard gate?

  23. hehhehehe I think you overcomplicated the video.
    Just a tip. For me personaly, when I explain the concept of QC the easiest thing is to just compare the outcomes of a classical walk and quantum walk algorithm with one particle and two particles (so 3 probability distributions). Showing that even a superposition (since it can be described by wave mechanics) really shows the difference in classical and quantum perspective. I think is even possible without math.
    But anyway, understanding QC can be tricky. It gets trickier when you want to understand other computational models and other ways of encoding information into continuous variable or even hybrid encoding. For an experimentalist it can be a bit tricky.
    Good job.

  24. It would seem you could make a qbit out of macroscopic stuff, except for the measurement. That measurement is where the magic happens.

  25. Here’s a thing you could do to make this more understandable. Show the unit circle diagram of each qubit as it passes through the algorithm. Then we can see what the quantum gates truly do to the qubits instead of stumbling over a weird explanation.

  26. All you need to know is this: clever algorithms make use of super position. REGULAR computing is highly complex: processor pipelines, ISA's, two-zillion languages, peripheries, memory store and retrieval, etc etc etc…Quantum Computing is even MORE complex, so relaying it neatly to the uneducated is nearly impossible. I have a degree in CS(ee), and this video confused me…..

  27. This was informative and entertaining. As an aside, since this comment comes from the future, 3+ years after the release of this video, we can proudly announce that there are commercially available quantum computing systems now (yay!)… but by "commercial" I mean people who can afford to spend $5-15 million dollars… and have specialized facilities to maintain the low temperature conditions necessary…. And a don't care about the electricity bill… and, to reiterate, you are the sort of person who calls a million dollars "spare change" (awww….). But hey, classical computers used to take up a whole room in the late 40's so… hope?

  28. Man! Best video on QC so far! I've watched IBM's videos and what not! This actually is the best intro. Why don't you make more videos in the series explaining the technical stuff in somewhat more depth?

  29. Finally something that i could understand, that paper flipping trick still amazes me even though i had enough amusement with my own example of it.

  30. Fun fact: The machine is actually Case C, because we know that it's Needy, and that it's either Case B or Case C.

  31. I think you could make the Bloch circle an actual term if you submitted it to the right channels. What those channels are, I don't know, but it's definitely an option.

  32. If x is true, then y is flipped. Or it could be that if x is false, then y is flipped. Sometimes the value of y is just flipped between true and false independent of x. Your job is to figure out what is going on. The solution is to make x be right and y be left. Even if a quantum computer is built, how many people will be able to write software for it?

  33. Don't give up haha, jk jk. Try explaining the movement (spin) of a qubit through time. I am by faaaaar not an expert but i have a ton of questions and have no clue who to ask. Could the 'messy' mechanics of quantum particle be simply explaind as particle movent trough time, but considering that 'time' is just another special dimension?

  34. MAYBE "you don't know how quantum computers work!" BECAUSE YOU WON'T EXPLAIN IN COMPARISON, HOW NONE QUANTUM COMPUTERS WORK. WAIT. IT WAS EXPLAINED (VIDEO POSITION; 0:41 / 15:48). THERE IS NO FLIPPING-FLIPPING. IT'S JUST THAT YOU WOULD HAVE A NEW TYPE OF SWITCHING COMPONENT FOR COMPUTERS, DEVOID OF USING TRANSISTORS. THE SWITCHING SYSTEM IS SO NEW, THAT IT'S BEEN THEORIZED THAT 'IT WILL BE CAPABLE OF ACCESSING 1 AND 0 AT THE SAME TIME'. IMAGINE IF YOU WILL WHAT IS SAID OF ‘PARTICLES BEING AT DIFFERENT PLACES AT THE SAME TIME’. HOW DO YOU SUPPOSE THAT'S POSSIBLE? IF IT'S TRUE AT ALL. SIMPLE, FREQUENCY. SERIOUSLY, I IGNORED THAT TRUTH FROM THE DAY I LEARNED OF IT.

    I SPENT YEARS WORKING ON IDEAS OF MY OWN AND HAVE SEEN FREQUENCIES AS HIGH AS 43GHZ @ 3dB—CONTINUOUS—. AND THE ONLY THING I EVER REALIZED THAT MAKES MORE SENSE THAN HIGH FREQUENCIES, IS PERMEABILITY. “43GHZ @ 3dB—CONTINUOUS—”, MEANS, A PARTICULAR SIMULATED “PERMEABILITY”, DROPS AS FREQUENCY GOES UP. TO SOLVE THAT PROBLEM, HIGHER AND HIGHER AMPLITUDES OF CURRENT OR VOLTAGES MAY BE USED TO MANIPULATE PARTICLES— IN THAT CASE, BY USAGE OF CURRENT, YOU END UP WITH HEAT. NO ONE KNOWS WHAT HAPPENS WITH THE USAGE OF HIGH VOLTAGE LOW CURRENT, THEY HAVE THEIR THEORIES AND HYPOTHESIS—. HENCE, HOW CAN SOMETHING BE MORE THAN ONE PLACES AT THE SAME TIME SO TO SPEAK, IF THEY CAN’T EVEN GET TO A NEW LOCATION?

    IN THE FIRST PLACE, NO ONE IN THOSE PROJECTS ARE TAKING THOSE TINY STEPS PEOPLE SPEAK OF. THEY WANT IT ALL FAST.

    TRY TO FIND OUT WHAT’S THE HIGHEST Hz MANAGED ALREADY? WHAT'S THE “PERMEABILITY— ‘GRAPHENE TECHNOLOGY’—? AND HOW MANY QUBITS WERE GENERATED AS WORKABLE ‘ENTANGLED’ CHARACTERS?

    DAMN, THIS VIDEO WAS SINCE 2015. I DON'T THINK IN THIS CASE IT MATTERS THOUGH.

  35. 9:50 I didn't understand why that happened, when you pass the qubit x through a hadamard gate, and it goes to 0 in laid-back, and 1 in needy cases. Shouldn't it be random and collapse to either 0 or 1, independent of the circuit ?

  36. does this means that future ourselves doesn't want us to learn quantum computing, because in future its becomes more weird.
    If you think weirdness is like entanglement.

  37. I try to understand what ‘s happened here but no luck. Basically, why quantum computer so fast is because one qbit represents four states(up down left right) while in a normal computer ,one bit can only represent two states(zero or one)right?? However there is no way to figure out left and right because it will change to up or down randomly if you read it? In this video, we can only figure out laid (case a and b )or needy(case c and d), but we still can’t figure out which case. so how quantum computer work? The thing is , it just fall back to “one bit” data , normal one bit represent zero or one, while qbit represent laidback or needy……

  38. well, very nice desciption, probably one maaningfull of 1000 around… so its also that qubit inputs are in fact 2D complex numbers (very hard to encode there, btw), so in fact analog signal where the outputs of quantum gates are always by-design collapsed digital results only? so in fact, it can be simulated also as analog computer ALU made of op-amps? (well, 1D real numbers? very aproximated explanation??)

  39. Honestly bro, you explained qbit the best, with first video ofcourse. Other video seems just repeating another explanation which lead to another confusion. At first I thought it just doesnt make sense to process a random rotating qbit and how qbit become 1 and 0 at the same time. Now i can sleep.

  40. Was all that applied in the Google machine? And if so, does all that implies that the particles used are entangled (you need to create them)? In order to predict the states, it sounds like they should be.

Leave a Reply

Your email address will not be published. Required fields are marked *