Am I getting this right? Quantum computing allows for all possible inputs to be entered as a single input, it then isolates all the inputs and performs the algorithm on each isolated input, then the output is mixed back into a single output of all possible answers. The "magic" is the algorithm is ran on each isolated input at the same time instead of in serial like a normal PC would do in a loop. So the more qubits you can support the more input/output states can be handled at the same time.
Yup, that’s exactly right! The tricky thing is getting something useful out of that big final state. Usually you can’t- but annoyingly all the information really has been calculated. There’s only a handful of special cases known when you can get some useful info out of the final state
Yes, the quantum computer works on the different basis states in parallel but keep in mind that the number of basis states in a superposition scales _exponentially_ with the number of qubits. This means that for n qubits, you can get a superposition of 2^n states which get treated separately by part of the algorithm until they are recombined. For example, for 10 qubits, you can get a superposition of 1024 states. So them being capable of working on states in parallel is like half of the speedup. Just working on things in parallel is something that a GPU can do already, but not exponentially so.
@@LookingGlassUniverseAm I correct that the class of Constraint System Problems like a sudoku puzzle, map coloring, and wave propagation tiling could be solved in a single step given a big enough set of qubits to represent the state?
@@LookingGlassUniverseThat's exactly what I thought about how quantum computers work. No one ever said that, but always that annoying fact of a qubit being 0 and 1 at the same time, from which I deduced that hypothesis but never confirmed.
Instantly subbed when you showed the parts where you were confused and working it out. We don't always see this part of the process, and it's so important for reminding us that it's okay to make mistakes. It was even better being able to see you reach a genuine moment of insight at 22:54. And this presentation was crystal clear--I learned a ton! A general note: please increase the volume; the ads were way louder than the video
This is what I love about physics. It's like "You can demonstrate fundamental truths about the universe just using a handful of items. You probably have this one in the back of a drawer somewhere, you might have this one under the sink, and this one costs THOUSANDS OF DOLLARS."
The universe is a sim imo. Does the physics of a video game matter to you? It's been programmed into the sim, things are likely exaggerated like they are in games.
precision repeatability that adequately isolates environmental variables (including installation and a maintenance plan, sturdy enough to survive years of undergrads) costs thousands of dollars. Demonstrating an effect, especially an allegory of an effect or at least the concept of the effect, can be incredibly inexpensive. Its the actual analytical hardware that gets pricey. Partly because of the small market, partly because of the captive market (gives Thor Labs the stink eye), but yeah, a lot of cool stuff can be done with effectively trash. Hell, there is even a trash laser (the nitrogen TEA laser I believe)!
As a quantum computing (experimentalist) Ph.D. student, it’s nice to hear that you second guess everything as well 🙃 my least favorite part of studying is how I’ll end up spending 2 hours second guessing everything on a topic I previously thought I mastered only to realize I was right from the beginning. Your video about light slowing down and the evident struggle of self-teaching has to be the most reassuring thing ever - so thank you for that. Us physicists live for the a-ha moments though, I suppose 🙂 loved the video and your art style, keep up the great work!
Well, imagine the swindler has put another black box in the first, unbeknownst to you, 90° out of phase, from it's prior use. Best that you find an alternate field of study before it's too late!
hey, how did you get into it, ive only just finished my bach in physics and a double degree with CS as well, but i have no idea how to move towards my long term dream
as an overly educated evolutionary biologist still traumatized by the academic experience of emails and their responses 😂, let’s try to please find less generic phrases such as “keep up the good work” in response to truly unique and ingenious efforts. Love you all 🎉.
@@NinjamagicsFind an advisor/mentor that is a good person that offers you freedom to explore and learn while also offering true mentorship and feedback. You have to decide the field and choose with discernment and after many days of nerding out on the subject field and literatures. After immersing in the literature , contact your chosen prospective interest advisors. Their response or lack there of will be extremely informative as to where you are at in your educational journey , where you fit, and whether or not they are a fit for you. Dive deep and have fun friend.
29:56 I THINK I'M JUST STARTING TO GRASP WHAT "CHOOSING ONE" OR "COLLAPSING" MEANS!!! Thank you, thank you, thank you. I don't fully understand quantum computing yet, but this was really helpful to even start. This was so abstract before this video, I can't believe you made a didactic tool so powerful.
Part 1 (where I make the quantum computer): th-cam.com/video/muoIG732fQA/w-d-xo.html If you want to do this experiment at home, you can! It's very simple. All you'll need is: - a weak red laser pointer (the type in cat toys are generally safe) - polarizing film or polarizing filter. If you have polaroid glasses or certain camera ND filters you may already have this. Otherwise it's available on amazon - half waveplate (the plastic thing) is this one: www.edmundoptics.co.uk/f/polymer-retarder-film/14827/ (λ/2 Retarder Film (WP280)) - You don't need calcite, but if you want to play with it, you can find it on etsy usually. Look for a sample that's exceptionally clear
At @11:20 you say one lambda and full wave plate. But that should be half lambda as here in the description, right? One lambda wouldn’t change the light, right?
Really, really enjoyed this video. I've seen this problem explained before, but usually it's done so fast that after a while I retain nothing. But with your slow pace + real life demo (and not just animation) I have the feeling it got engraved in my mind this time (and the HW answer is balanced). However, what I really want to know is why light slows down ?? 😉😅
@@justpaulo big answer made small.. the wavelength of light gets a kickback in phase when it combines with the em waves generate by the electron due to the incident of your light.
Some time ago I took a course on quantum computing, but I never expected to see such a nice illustration with simple materials. What's even more funny is that I'm also learning about optical mineralogy, where these principles are relevant as well. An unknown mineral could behave pretty much like the secret box and the way of finding out is rather similar.
I took a cours in quantom computing and your videos just blew my mind. Everything became crystal clear ! I’ve sent the link to many of my professers and they loved it to. Instant sub, thanks for taking time to make these brillant videos.
Christ alive, I’ve been wondering how quantum computers work since I was a kid and learned about them in the 00’s. This explanation/experiment has cleared up so much for me. And thank you for making me feel smart by answering the question in my head of “would it be possible to measure if it’s a 0 or -0 by splitting the beam and checking the interference”. The fact it’s explained with pipe cleaners and polarized filters makes me upset that we never did this at home 😅
Well that was an hour of useful information (watched both videos in one sitting). I’ve been searching for someone to explain in a human manner how quantum computing works at the 1/0 state of digital computers. No one has until these videos. I can see why they are more powerful than digital, and why it’s so difficult to compute with them. To me they are similar to an analog computer which was thought to be more powerful than digital but it was just harder to get to work properly. Cudos to you! And thank you!🙏
analog is not comparable to quantum at all. Analog is very much classical in its workings. There are no superposition calculations going on in analog systems. The potential of analog systems is speed. Using the physical properties of electricity to do the actual calculations - is very fast. The downside is flexibility. An analog circuit is only able to do a very specific calculation. It is not programmable in the same sense as a digital curcuit. ------------------------- This means that analog systems have great potential to speed up our digital computers, as a computation accelerator deployable in certain situations. Very much like a graphics card in its specific usecases. If the whole system is suppose to be analog, you will have very low flexibility after deployment. Kind of contrary to what makes computers extremely widely useful. Its not like I know a mathematical proof, but something tells me intuitively that building a turing complete system in complete analog is basically impossible. You will always depend on some digital circuits for programmability.
@@jonaswox wow you just wrote a whole book to prove my point. You said “the potential of analog systems is speed” that’s the exact same potential of quantum computers. You said “analog circuits is only able to do a very specific calculation” that is the exact same as saying quantum computer can perform one single algorithm. You said analog is not programmable in the same sense as a digital circuit. So are quantum computers based of what the videos show. So clearly both quantum and analog are comparable. And just so that there’s no confusion, in my statement I said they are similar, I’m very much aware of the fact that quantum checks the particle position to perform its task, and analog uses frequencies to perform its tasks. FYI, we already have analog processors to perform matrix multiplication for AI purposes. They are much faster than digital processors and use less power. We also have light processors for the same purposes as the analog processors. Both analog and light are much better suited for AI, but right now they’re just too expensive. This is what we’re going to need to solve the complex problems artificial intelligence is trying to solve. Quantum computing is probably going to smoke light and analog, but it’ll use more power to do so.
@@ElvisRandomVideos You are obviously not educated in the subject. No offense :) Im not proving your point at all :D Yes it seems like you have discovered the potential of embedded systems ;) I once was in a course on embedded systems, ... 20 years ago Dedicated circuits for specific calculations is not a new idea at all. And you always pay with flexibility when gaining speed.
I love the symmetry in this experiment. (3 possible ways to split those functions into pairs and for each pairing there is a measurement to discern which pair the function belongs to, but we still need 2 measurements to get the 2 bits needed to identify the function.)
Yeah! Welcome to information theory :) With a classical computer, you would never be able to get beyond that. There would be different ways to measure, potentially, but it's fundamentally impossible to get more information out of it than you put in, and with a classical computer, the information you put in can only have one bit. With quantum computing, it has to turn classical at the far end (the final measurement step), so there's still a fundamental limit (with one superposition input and one measurement, you still can't tell exactly which function it is), but there are now measurement forms that involve TWO inputs.
Very cool! Makes me want to get back to making quantum circuits. If you add a *polarizing beam splitter to the end of the circuit and 2 photoresistors to each mode (H or V) connected to an arduino, you can interface your circuit with your computer. I did that to build a QRNG a few years back that I used to generate mazes, was the most useful circuit I could make with a single beam splitter lol
I am sorry to inform you, but you can't directly read the quantum states of photons using an Arduino and photoresistors. Here's why: Photoresistors measure light intensity, not quantum states: Photoresistors work by changing their electrical resistance based on the amount of light hitting them. They essentially measure the number of photons (light intensity), not the specific quantum states those photons are in. Quantum measurement problem: Reading a quantum state is a delicate process. Observing a quantum system inherently affects its state. An Arduino and photoresistor setup doesn't have the sensitivity to measure individual photons without collapsing their superposition or entangling them in unwanted ways. While your QRNG (Quantum Random Number Generator) project using a polarizing beam splitter and photoresistors is clever, it leverages the probabilistic nature of photon polarization, not direct quantum state readout. The beam splitter sends photons into different paths based on their polarization (a quantum property). However, the photoresistors still only measure the resulting light intensity at each path, which is a classical outcome of the quantum process. This randomness in intensity is what you use for generating random numbers.
Finally a video that explains theiugh demonstration hiw quantum comouters work and what they do. Most videos repeat the same concepts about superposition, but don't explain exactly how the tech itself is set up to work.
Wave plates have axes of their own. They shift one component in that coordinate system relative to the other component in that coordinate system. Quarter wave plates shift by 90 degrees and half wave plates by 180 degrees. If the incoming polarization has equal x and y components, then quarter wave plates convert linear polarization into circular polarization and vice versa. Half wave plates allow us to rotate an incoming polarization by any amount.
Great narrative and explanation. I love that you showed your confusion and perseverance through doubting yourself - a key part of the learning process in my experience!
I'm so excited I found this channel, I love science but don't get to scratch that itch in my line of work. This feels like I'm right there doing the experiment, thank you
[I’m writing many comments. This is 1] I really like this video. I’ll preface by saying that I’ve taken a graduate level course in quantum computing and am very familiar with quantum mechanics. As a complete introduction for someone knowing nothing, I think you jumped into polarization of light and wave-plates a bit quickly by just taking the model for granted. I imagine people with no background there will struggle to follow. But at the same time you take things slowly and clearly. The visual aids are great. You manage to make it very hands on and yet functional. Of course, in making a serious quantum computer a big issue is to create and preserve entanglement which you don’t really need for a 1qbit version. But that’s exactly the cheat that makes this simple enough.
Thank you very much for this clear explanation, I finally understand thanks to you what we mean by "quantum computer", how we can make "calculations" with light. It’s magical! Bravo for your work!!!!
I love seeing your learning process in working with this! It helps me learn along with you! I think using a continuously shining light, and computing by changing aspects of the light is super clever! And... If I understand correctly, the final experiment was balanced.
The notation with 1 and 0 and + and - is really confusing. It would have been helpful to rewrite the functions when introducing the quantum version of the black box. For example 0: flipped/1: not flipped. When writing with the markers on your white board different color could be used too. (using the white board as the optical bench and for notes is genius btw!)
|+> is just the superposition of the usual 0,1 basis with a plus: |0> *+* |1> (and a global factor of 1/sqrt(2) that's not necessary to get the idea across) |-> is the same but combined with a minus (on the 1s side): |0> *-* |1> (and same factor of 1/sqrt(2) ) So the names for |+> and |-> are self explanatory
I think the background is a metafor for what it feels like to watch this as a ley person. I understood more of it than any other video about quantum computation though, so that’s good.
This is a really good illustration of observation in Quantum mechanics. We can only "test" for one of the states at a time, be it the |1>, the |0>, the |+> or the |->. What we get is defined by how we choose this basis of measurement, and none of these might actually be the states of the particle - the state of the particle just gives it some likelyhood to interact similarly, or oppositely of the state we chose to measure. I think this experiment would be augmented nicely if the light we get at the end was to be measured by a simple photodiode and voltmeter system, so we could see exactly when the strength of the light is halved and when we get a, say, 30-60 distribution. Also, as a random thought: we got two measurements in the end. Would that be "cheating" when the riddle states that we can only "use" the quantum computer _once?_
It blows my mind that ANY human can come up with a q-alhorithm .....like, it's crazy. Grover's, shor, quantum courier, etc etc. it almost feels like you have to know the answer before the algorithm 🤷 Machine learning is a super cool way to explore the space of q-states. Maybe folks will use it to explore possible q-alhorithms 🤷 I like you showing the messiness of the scientific process. Well done....more should show this! Love your explanation of duetsch-josza so far (the electron goes AGAINST the E-field though....minor pickyness. Use a positron) 👍👍👍😍😍 great supplement to Mike and ike
And... not a question, but how this experiment is presented (figuring out which part can work as which operation, getting confused sometimes, etc.) really makes it seems like you and the audience are doing science together... which I don't see a lot :)
it made me so vicariously filled with joy to see you get emotional for finally putting the science you spent so long studying theoretically into practice :)
Technically waveplate has fast/slow axis. So, in order to make the experiment more convincing, instead of conveniently constructing F3 & F4 mystery function with simply an “empty” box, you can actually aligning the waveplate fast axis exactly to |+>, and you will get -|+>. Still, it is absolutely a great video! Appreciate it a lot.
NGL, I saw that it took you 40 minutes to do the verification and it made me think of what Joel Spolsky said about PhD computer science people in the industry: basically if they aren't sure they get wrapped around the axle about being sure rather than just kicking out "something". Which is great in the scientific field, not so much commercially. That said, as a not PhD computer science human in the industry who tries to balance both, I appreciate your approach and love that you took the time. I wish I had the time, which is why I am thinking about doing my PhD.
I don't think it is necessarily good for the scientific field. People take a lot of time thinking about something rather than just trying something. If you have more action you can actually test whether your idea/ way of thinking was correct or incorrect. Then you can change what you have learned during the process if it was incorrect. You may even stumble across something interesting during the process. There's a balance to be had even in the scientific field, what's the point in thinking about stuff and wasting time when you can just test it? Ofc if depends on resources, but from a data scientist perspective I find some academic pondering debilitating for some projects.
@@liambailey5630 Unfortunately, at the Ph.D. level or at the level where you extend the boundaries of science, the goal is to prove or disprove something, and fully document your process and thoughts, so that your peers can understand, reproduce, extend, or prove it wrong. Outside of that level, at the engineering or application level, the goal to "fail quickly" is indeed much more productive. In some fields, that's good, while in others... for example medicine, I'd prefer that drug companies, etc. use the complete and transparent scientific method with peer review... rather than the "fail quickly" method that doesn't always explore completely and deeply.
@@hanksimon1023 Just because you provide quick action does not mean that the methodology is not documented or that its unethical. You can test ideas quickly without pondering for months. Some drugs were discovered by accident not by thinking. Obviously, testing on subjects rather than testing ideas in a lab is different. I agree that there needs to be a balance though.
@@liambailey5630 Thalidomide was tested inadequately. And, the Pfizer CEO refused to release early data on the COVID vaccine, saying that the data were company proprietary... This is a not so subtle way of saying that his profits [and confirmation bias] were more important than peer review.
Editing suggestion: you seem to have a small but usable library of graphic assets to play with. For times when you need to insert a voice-over from the editing room, create a reusable graphic assets from those smaller assets you currently have to quickly drop in and avoid the black screen. Example: the Mad Hatter standing in the center of the frame with his bow spinning on the tree and field background. Or Alice with thoughts bubble and in the thought bubble is a 3 dots who's opacity is fading in and out in a wave. HAPPY EDITTING!
Congrats for getting it to fully work. I think a view visuals could help making this easier to grasp - from looking at your setup I can't really see which filter is ofiented in which way. What does 45° even referemce too? Is the table surceface 0°?
I think the physical representations of your functions might be flipped around. Shouldn’t the function that does nothing (0 -> 0 and 1 -> 1) actually be nothing (the empty box scenario)? Overall, shouldn’t your results be this: if the polarization doesn’t change from input to output then it’s a balanced function, and if it changes by 90 degrees then it’s a constant function?
Flipping the EM wave only introduces a Pi phase shift into original quantum state, you need a photonic Pauli-X gate to rotate the polarization by 90 degree, which can be easily achieved by two properly aligned mirrors, or a properly aligned 1/2 lambda waveplate. She claims herself a PhD student without realizing her wrong explanation cannot even convince herself.
The "flipping the light upside down" is just shifting the phase of the light by 180 degrees... right? If this is the case, you can tell that the light is the "negative" of the initial light by combining the "negative" and the initial light together (they will cancel). Overall, maybe the computation can be seen by making a reference light (same as initial) and see how the output light interferes with the reference
that is not the first time i kinda understood quantum computers but i think this time i am going to remember, nobody ever did anything that detailed and yet informal way enough for me to stay that long watching
If a system gives a constant output for any input, then you cannot guess what the input was given the output. Guessing the input means that you want to reverse the process(in your head)to know the initial state. Any process described by the Schrödinger equation is a reversible process so any quantum system has to be reversible.
Even if you could calculate anything related to prime numbers - that’s such a specific task, such that it would be impossible to really get the benefits from it. As a data scientist, I prefer using older computers (~ 5 years old) over newer / faster ones, because it helps me to understand how my programs will perform on most people’s devices. And I save money. Quantum computers can be very expensive I hear. I think it’ll definitely benefit cryptography, like you said. It’s cool that you can make one, and understand the pros and cons.
Super cool to see animations; glad to see sponsors funding some of this work! The passion and perspiration is irreplaceable tho... PS: Nobody said there would be homework! 😮 I think the top secret box is a balanced function because only the first cuts out
Hi. I'm not AT ALL 'Super Math Guy' so excuse the 'input', PLEASE; but, "PS: Nobody said there would be homework! 😮" cracked me UP!! (I'll have to take /n times to rewatch this - and others.) Appreciate you folks! 👍 [Barry Setterfield and Halton Arp and WG Tiff and - by reference - JW Selentic, Bernand Haisch, Hal Puthoff, Timothy Boyer, Luis de la Pena (with out 'the math') are my 'stimulus'.
Great explanation and well delivered - even showing that PhD's can get get entangled when thinking about how a new material impacts the output.. I get the core concept you are putting across is the superposition created by introducing horizontal and vertificate polarisation into the same computation. But isn't the final "test" actually two measurements? First at 45 deg; then the second at 135 deg? Hence not aligning with the "single measurement" requirement?
We don't get the answer to the final question?! How cruel! I love these past few videos (and your past ones for that matter) because they made me understand what quantum computing is and what it isn't, as well as how it generally works. Amazing work!
What's happened at minute 21:54? When you wrote on that polymer.. what a diabolical material is that?? lol I mean it appears as if the ink passes through the polymer and writes directly onto the board, but as soon as you remove the polymer the writing disappears from the board and one can understand that it is on the polymer instead. So strange... but anyway you did a super great job with this video. Keep sharing knowledge! :)
Lol it looked weird but I think it was just the shadow of the polymer being cast on itself from the left. It looked as if it was bent up more than it actually was
I'm lost at 22:15. You say that F3 takes 0 to 0 and 1 to 0. So if you put nothing in the box, 0 will go to 0 (i.e. no change) and 1 will go to 0 (that seems like a change). What am I missing?
It is just an unfortunate and confusing selection of using zeroes and ones to refer to inputs and results ( to flip or not to flip) of each function. It is better to name inputs as V and H, so outputs of F3 is the same , V and H, because you don't flip nothing. To see the difference: F4 wants "ones" for both, so it would be : Inputs V and H, outputs -V and -H . ( Even though I don't see the quantumness here (yet?))
Hmmm seems like sometimes what we use as |1> and |0> for the input is different that what we use as |1> and |0> for the output. Can we really do this? especially if we want the output of one "quantum box" to be fed to the input of another quantum box (maybe not in this early experiment)
The change in f tables (f1 was originally the identity?) was hard to follow but I guess that finally we just have rotations of multiples of 90 degrees... f1=+ 90, f2= - 90, f3=0, f4= - 180, and so we can distinguish between { f1, f2 } and { f3, f4} by preparing | 0 > +| 1 >. Is that about right? Very enjoyable series!
The part I found really hard to follow throughout the video is how you're encoding the result. It would have been good to have a table of how you encode the input bit and the output bit before you built the computer. Or alternatively show the working computer before you show how you built it. This is a major part of understanding and you barely gloss over it at the start. So while you were building the functions it wasn't clear in my head what each function did in terms of encoding; you wrote it in terms of 1 and 0, not in terms of the final encoding, and both things are necessary to understand.
I always complain when people say "factor a prime number" instead of "factor a composite number" ... but it's actually a real, hard problem that should better be called "proving a number is prime", which is very useful. If there are only 2 factors, the number itself and 1.
I think there was some flipping of the definitions of f1, f2, f3 and f4 during the video. For the original definitions at the beginning of the video I think these should be the correct implementations |0> = horizontal polarization |1> = vertical polarization |+> = up and right = +horizontal +vertical at 45 degrees |-> = down and right = +horizontal -vertical at 45 degrees F1 - Do Nothing laser -> 0 delay -> output = no change F2 - Always Swap laser -> half waveplate aligned horizontally -> half waveplate aligned vertically -> output = -horizontal -vertical (pi delay = lambda/2 delay) Note: This wasn't the implementation shown in the video because I don't think it was shown correctly in the video due to the f2 definition changing but correct me if I'm wrong F3 - Swap if |1> (vertical polarization) laser -> half waveplate aligned vertically -> output = horizontal -vertical (pi delay to vertical polarization) F4 - Swap if |0> (horizontal polarization) laser -> half waveplate aligned horizontally -> output = -horizontal + vertical (pi delay to horizontal polarization) Then to solve the problem laser -> 45 degrees polarizer (up right) -> now in the |+> state -> f1/f2/f3/f4 -> 45 degrees polarizer (up right) -> light = f1/f2, no light = f3/f4
In the 1st part of your video, You´re changing the phase, not flipping the polarization, so if you change the phase of the polarized beam and add it to a non inverted phase beam you should be able to detect the change, since opposite phases should cancel out. at 16:00 now you´re flipping the polarization, that should be easier to detect, right?
Question: does it really matter that we use quantum objects or is it only a matter of sending two pieces of information at the same time (horizontal + vertical polarization)? In a classical computer we measure only the voltage. What if we measured the voltage and the current? It's just an example, this might be any classical object that just has more than one property that are sort of independent on each other. Btw, awesome video and great effort!
You said (4:30) that it is too hard of a problem to find what is the exact function he uses (f1,2,3,4) so we're only gonna find if it is balanced or constant. But what if you use a beam splitter before the secret function and combine it with the output? you input |+> = |0> + |1>: if it flips the horizontal direction, after the computation you are left with - |0> + |1> which will combine with |+> to output |1> if it flips the vertical direction, after the computation you are left with |0> - |1> which will combine with |+> to output |0> if it flips both, you will be left with -|0> - |1> which will combine with |+> to be no light if it flips nothing then you are left with |+> You then measure the combined light and check if it is strongest in the vertical, horizontal, zero everywhere or 45 degrees What do you think about this? (you do need to measure multiple angles but I hope it is allowed)
Yes, but to make multiple measurements you need to run the quantum algorithm multiple times, and then it's no faster than a classical computer. And in terms of the back story, it violates the leprechaun's "one run" rule.
Thanks for this very interesting videos! Few notes regarding the optics: I'm not sure what wave plates you were using, was it half wave plate? If so, it simply rotates the linear polarization and you can replace it solution of chiral material, sugar in water for example. The concentration or the optical path will determine the rotation. Full wave plate only does nothing for a specific color it was designed for. Not sure you've matched the laser color and the full wave plate. Polarizer used for photography like the one I think u use in front of the laser, are usually directional since they are made of two layers: linear Polarizer and quarter wave plate after it. I'm not sure what Polarizer you've used and what direction was in use. Thanks again for the great, interesting videos! 😊❤
Grate chanal: I'm sure most here know that but just for general info. a wave plate or lambda plade rotates the polarisation; a lambda 1/4 plate rotates the polarisation a 90° as a lambda 2/4 plade rotates it 180° and so on. I use them when working on advanced interferometry and holography.
Hi, great video. I'm an ignorant (dropped out from high school) so I don't understand physics but I do know a little bit of IT and computer science. I keep hearing this "quantum computer" a lot but idk much about it as it seems confusing. Your projects seems to be simple, as far I'm seeing it some input laser with a polarizing filter and another polarizing filter on the middle and board to stop the light so you can see the output. That's all what I understood from the video as it is long and since I'm ignorant I really can't follow up what you saying and when I try to skip and skip to the moment of some experiment where you actually do something then i just don't understand the result because it somehow related the middle explanation things that I can't get. So let me get this straight, can you name what you made? A 1 bit full adder? or what? Could you please explain in the comment in a simplified way :) Also, what would make a 1 bit full adder made out of optical filters different than one made out from transistors? speed? so speed makes it quantum? could you please clarify in a simple manner for non student. And thanks for sharing!
33:08 Without going back into the video... I have to remember what you put... Mkay. F1, F2 were balanced, meaning the output changes. F3 and F4 were constant, meaning it doesn't. It's not F4 because +45° cut out the light and -45° kept it. Nor is it F3 because both 45s should have cut out the light. And then since -45° showed a 0, or no change in state, then you put in F2? ... ... and now I'm thinking I'm incorrect... lol. Also... where might one find a cute, funny, creative and intelligent quantum physicist woman? ...asking for a friend... lol XD. Well done on the video, lol. Definitely informative!
Very cool stuff! Inspiring! At the end, in the final measurement, you turn the filter two different ways. That seems like cheating, e.g., giving two bits of input, but it also seems unnecessary. Was that just to give the audience some idea of the relative levels of light? I.e., if someone already knew what's bright and what's not in the output, could they leave the final filter stationary and still know the answer?
That’s right! You don’t need to turn the filter- you can just do it at one of those two angles and infer what the other result would be by using the total brightness
a way of detecting if flipping wave works would be to check the result of interference with original light - successful flipping would result in decstructive intereference.
Is there ANY currently theorized path to quantum computers being generally faster than 'normal' binary computing? by that I mean, replacing current desktop computers and being better at running 'normal' code? Maybe a number of qubits where the exponential scaling of possible states leads to some phase transition into general compute practicality? And/or the possibility of some yet to be discovered quantum algorithm that makes them practical for general computations? Or is it basically true that quantum computers will always only have special use cases?
Crazy how much stuff starts to make make sense the more u learn about electricity ) like the charge of atoms and electrons being the reason for its bias
It can count syntax statement total characters and separate statement by parts, and sorting arrays to those parts in a syntax statement to do square root as parallel group count. On surface level. In analog, oscillating RLC diagrams. So if a string of parked cars on city road, between one city block and two traffic signals, then if a row of cars at a red light accumulate until light goes green, that batch of cars make the length of trip from red light and pass the length of block to the last signal, in that instance the first green light has no car at its intersection so cars further back were at constant speed, those cars zoom pass the light as of it had never been red to begin with, then if the mph is 30, then the car zoomed passed the light and at 1 minute, after for 0.5 miles, the square root in syntax of length of parked cars if each car was a character byte per 1 car parked. In an analog RLC circuit. The correct answer is actually 6(5) minus 0.01 = 29.99 mph, at is 1 , so half a mile is equal to 1 minute at 30 mph, if cars zoom the green light, is equivalent to 1 cent, or Linkin park.
Really interesting demo. Quick question - in classical computing we have different options for hardware implementation (eg. you can build a turing complete mechanical computer without electricity, you can build a turing complete computer from relays, or these days you can build a turing complete computer from semiconductor) ... in your video you're demonstrating an implementation that uses light and polarizing lenses/filters as the gates, but I've also seen some stuff saying that we can build quantum computers based on electron spin (I think this is why they need to be at 0 Kelvin, so thermal energy doesn't affect the spin or something?) ... question is, is there a preferred method for implementing the hardware of a quatum computer? Are light and electron spin the only two known methods right now, for implementing a quantum computer? Thanks again for your video, I've never seen this done with light before and it was super fun to watch. You're awesome! EDIT: balanced, but I still cant tell which specific one of the four functions it is..i may have missed something that helps us narrow down exactly which of f1,f2,f3,or f4 it is...
In principle, one can make a quantum computer using the phase of the electromagnetic field (the keyword is Gaussian information). It doesn't require cryogenics but has challenges of its own. In turn, spin-based quantum computers are rather a minority. I don't think anyone makes polarization-based quantum computer. These are hard to make: besides cryogenics, they require difficult controls.
Would this be similar to offering a 2.5V signal to a TTL circuit to force it to reveal more about its structure than could be obtained by simply offering 0V or 5V? (By cleverly using some glitch in the circuit that activates only by offering 2.5V) Essentially you created a 3rd type of input 0V, 5V and 2.5V. but that is cheating. You are only allowed to input 0V or 5V. (O or 1)
It would be cool to set up the experiment with an lcd screen (older one that has linearly polarized light) with a blank image as a light source and a human eye (Haidinger's brush) as a detector to make a Human-*quantum computer :D It's likely that the waveplates will spil the fun but maybe some hint of the haidinger's brush will still be visible. note: slowly wiggling head from side to side helps me to to see the haidinger's brush.
Hey there I think this is just a part of how you can define what quantum is about. There is also that Cambridge research with the quantum teleportation back in time then back in present with a molecule or something I can read remember it’s right now exactly but that’s sort of another example, right? Larger skill I think this can be used. Infinite way is actually if you can find the right sequence. …
But the light is not 'slowed down' in the filter, it's actually phase-shiftet by half an period, so the wave seems 'flipped'. Am I explaining this right? The only 'slowing down' is happening right in the filter with the refraction of light. 3B1B did a great video why light is just slowed down in a denser medium.
11:27 Quarter wave plate? This brings up memories from way back when I used a circular polarizer on ma Praktica VLC2 SLR, that had a semi-transparent mirror and noticed the very specific blue/brown color tint when using a polarizer on it, as the semi-transparent mirror used polarization to reflect light to the viewfinder and the metering sensor for the exposure. The memories this brings back. This technology feels (and is) "ancient". 🙂 I apologize for spamming, but I am drunk. ;-)
your old thumnails with original drawing are really good. i was happy to see your content isnt ai generated garbage. i truly think if you avoid using the same thumbnails as those spam bot channels you will go far albeit may take a lil bit longer but will be organic growth that sustains itself. keep up the hard work your doing it great, pat yourself on the back
Sounds like quantum computing would be useful for procedural generation in Video Games due to it being able to compute all possible inputs as one input I wonder if you have any idea how it could be used as a tool to aid character/environmental design within the game engine
But, as far as I understood, you only need a continuous-wave laser that renders a coherent state (together with polarizers, waveplates and some interferometry) to make this calculation, right? There is nothing quantum here, just classical optics. It looks that many problems that people claim to be solved with quantum stuff can be equally solved using classical light, like implementing quantum walks. Please let me know if I'm wrong, it is something I would like to understand better. The video was amazing btw and I learned some cool stuff!
You are right: this is a classical computing device. To see where quantum computers start to shine, one must look at three-qubit realizations, for example, that same Deutsch-Jozsa algorithm for functions on two-bit inputs. For functions on one bit, there's a single input solution, for example, f(x) XOR f(NOT x), which complicates comparisons.
This blew my mind, the fact that you are implementing a quantum algorithm (alas the simplest, Deutsch's) with a laser pointer and some polarizers and filters either fixed with play-dough or just hold free hand... Kudos! What is the intrinsic difficulty in scaling this scheme to more qubits to get true linear optical quantum computers?
Correction: it takes seconds for a classical computer to factor any 30-digit (decimal) integer. 100-digit in only mildly difficult. It's standard to use 616-digit (2048-bit) integers for practical purposes, like authenticating a website.
Mithuna, I noticed in the MIT Lectures they cover the Deutsch-Jozsa algorithm, and it looks different from what you did here there. I saw it again and at least the problem is the same for you and them, but I dont understand why in the MIT Lectures they begin with two particle entangled states and mention no cloning. I get the gist of your approach, create 4 unitary transforms which basically flips some of the basis vectors to encode the functions, but I dont know why the MIT one seems so much more difficult--whats the payoff of entangling there. Thanks, good thought provoking video!
Great question! This comes down to how you assume the challenger gives you the black box. If you assume they gave you a version of the box that works the same as a classical computer then it takes an input x and a blank bit (we’ll say it’s 0) and it outputs x and f(x). Then you need to do the entanglement trick to get it into the more useful form we assume in this video. In the end, you end up with a computer that takes |x,-1>-> -1^(f(x))|x, -1> Since this does nothing to the second qubit at all, let’s just get rid of it. Now the black box computer is in a form that’s much more useful to us
So as I see it, if you can only give a 0 or 1 as input, you can only tell if the function is either (f1 or f3) or (f2 or f4) but if you can use a quantum computer you can only tell if the function is either (f1 or f2) or (f3 or f4)? So why would you choose quantum over classical as it seems like they reduce the output state equally.
Am I getting this right? Quantum computing allows for all possible inputs to be entered as a single input, it then isolates all the inputs and performs the algorithm on each isolated input, then the output is mixed back into a single output of all possible answers. The "magic" is the algorithm is ran on each isolated input at the same time instead of in serial like a normal PC would do in a loop. So the more qubits you can support the more input/output states can be handled at the same time.
Yup, that’s exactly right! The tricky thing is getting something useful out of that big final state. Usually you can’t- but annoyingly all the information really has been calculated. There’s only a handful of special cases known when you can get some useful info out of the final state
Yes, the quantum computer works on the different basis states in parallel but keep in mind that the number of basis states in a superposition scales _exponentially_ with the number of qubits. This means that for n qubits, you can get a superposition of 2^n states which get treated separately by part of the algorithm until they are recombined. For example, for 10 qubits, you can get a superposition of 1024 states.
So them being capable of working on states in parallel is like half of the speedup. Just working on things in parallel is something that a GPU can do already, but not exponentially so.
So like quantum multithreading?
@@LookingGlassUniverseAm I correct that the class of Constraint System Problems like a sudoku puzzle, map coloring, and wave propagation tiling could be solved in a single step given a big enough set of qubits to represent the state?
@@LookingGlassUniverseThat's exactly what I thought about how quantum computers work. No one ever said that, but always that annoying fact of a qubit being 0 and 1 at the same time, from which I deduced that hypothesis but never confirmed.
Instantly subbed when you showed the parts where you were confused and working it out. We don't always see this part of the process, and it's so important for reminding us that it's okay to make mistakes. It was even better being able to see you reach a genuine moment of insight at 22:54. And this presentation was crystal clear--I learned a ton! A general note: please increase the volume; the ads were way louder than the video
This is what I love about physics. It's like "You can demonstrate fundamental truths about the universe just using a handful of items. You probably have this one in the back of a drawer somewhere, you might have this one under the sink, and this one costs THOUSANDS OF DOLLARS."
the universe loves shoving things about itself in our faces until the topic shifts to one of its insecurities
The universe is a sim imo. Does the physics of a video game matter to you? It's been programmed into the sim, things are likely exaggerated like they are in games.
precision repeatability that adequately isolates environmental variables (including installation and a maintenance plan, sturdy enough to survive years of undergrads) costs thousands of dollars. Demonstrating an effect, especially an allegory of an effect or at least the concept of the effect, can be incredibly inexpensive. Its the actual analytical hardware that gets pricey. Partly because of the small market, partly because of the captive market (gives Thor Labs the stink eye), but yeah, a lot of cool stuff can be done with effectively trash. Hell, there is even a trash laser (the nitrogen TEA laser I believe)!
As a quantum computing (experimentalist) Ph.D. student, it’s nice to hear that you second guess everything as well 🙃 my least favorite part of studying is how I’ll end up spending 2 hours second guessing everything on a topic I previously thought I mastered only to realize I was right from the beginning. Your video about light slowing down and the evident struggle of self-teaching has to be the most reassuring thing ever - so thank you for that. Us physicists live for the a-ha moments though, I suppose 🙂 loved the video and your art style, keep up the great work!
life of a scientist. Everytime you get the smallest insecurity you have to investigate :D But this is ultimately what makes us better.
Well, imagine the swindler has put another black box in the first, unbeknownst to you, 90° out of phase, from it's prior use. Best that you find an alternate field of study before it's too late!
hey, how did you get into it, ive only just finished my bach in physics and a double degree with CS as well, but i have no idea how to move towards my long term dream
as an overly educated evolutionary biologist still traumatized by the academic experience of emails and their responses 😂, let’s try to please find less generic phrases such as “keep up the good work” in response to truly unique and ingenious efforts. Love you all 🎉.
@@NinjamagicsFind an advisor/mentor that is a good person that offers you freedom to explore and learn while also offering true mentorship and feedback. You have to decide the field and choose with discernment and after many days of nerding out on the subject field and literatures. After immersing in the literature , contact your chosen prospective interest advisors. Their response or lack there of will be extremely informative as to where you are at in your educational journey , where you fit, and whether or not they are a fit for you. Dive deep and have fun friend.
29:56 I THINK I'M JUST STARTING TO GRASP WHAT "CHOOSING ONE" OR "COLLAPSING" MEANS!!! Thank you, thank you, thank you. I don't fully understand quantum computing yet, but this was really helpful to even start. This was so abstract before this video, I can't believe you made a didactic tool so powerful.
It's crazy... This was rather scattered and yet gave me much better insight than most other explanations 🤣🤷♂️
Part 1 (where I make the quantum computer): th-cam.com/video/muoIG732fQA/w-d-xo.html
If you want to do this experiment at home, you can! It's very simple.
All you'll need is:
- a weak red laser pointer (the type in cat toys are generally safe)
- polarizing film or polarizing filter. If you have polaroid glasses or certain camera ND filters you may already have this. Otherwise it's available on amazon
- half waveplate (the plastic thing) is this one: www.edmundoptics.co.uk/f/polymer-retarder-film/14827/ (λ/2 Retarder Film (WP280))
- You don't need calcite, but if you want to play with it, you can find it on etsy usually. Look for a sample that's exceptionally clear
At @11:20 you say one lambda and full wave plate. But that should be half lambda as here in the description, right? One lambda wouldn’t change the light, right?
Very clear calcite may be sold under the varietal name "iceland spar"
Really, really enjoyed this video. I've seen this problem explained before, but usually it's done so fast that after a while I retain nothing. But with your slow pace + real life demo (and not just animation) I have the feeling it got engraved in my mind this time (and the HW answer is balanced).
However, what I really want to know is why light slows down ?? 😉😅
CONSTANT?
@@justpaulo big answer made small.. the wavelength of light gets a kickback in phase when it combines with the em waves generate by the electron due to the incident of your light.
I watched the first video, and now here i am. Usually, i skip a lot when watching TH-cam, but your explanation beat my adhd
Some time ago I took a course on quantum computing, but I never expected to see such a nice illustration with simple materials. What's even more funny is that I'm also learning about optical mineralogy, where these principles are relevant as well. An unknown mineral could behave pretty much like the secret box and the way of finding out is rather similar.
I took a cours in quantom computing and your videos just blew my mind. Everything became crystal clear ! I’ve sent the link to many of my professers and they loved it to. Instant sub, thanks for taking time to make these brillant videos.
Christ alive, I’ve been wondering how quantum computers work since I was a kid and learned about them in the 00’s. This explanation/experiment has cleared up so much for me. And thank you for making me feel smart by answering the question in my head of “would it be possible to measure if it’s a 0 or -0 by splitting the beam and checking the interference”.
The fact it’s explained with pipe cleaners and polarized filters makes me upset that we never did this at home 😅
This was honestly extreamly helpful in understanding the role of quantum computing. Thanks for uploading this!
Well that was an hour of useful information (watched both videos in one sitting). I’ve been searching for someone to explain in a human manner how quantum computing works at the 1/0 state of digital computers. No one has until these videos. I can see why they are more powerful than digital, and why it’s so difficult to compute with them. To me they are similar to an analog computer which was thought to be more powerful than digital but it was just harder to get to work properly. Cudos to you! And thank you!🙏
analog is not comparable to quantum at all.
Analog is very much classical in its workings. There are no superposition calculations going on in analog systems.
The potential of analog systems is speed. Using the physical properties of electricity to do the actual calculations - is very fast.
The downside is flexibility. An analog circuit is only able to do a very specific calculation. It is not programmable in the same sense as a digital curcuit.
-------------------------
This means that analog systems have great potential to speed up our digital computers, as a computation accelerator deployable in certain situations. Very much like a graphics card in its specific usecases.
If the whole system is suppose to be analog, you will have very low flexibility after deployment. Kind of contrary to what makes computers extremely widely useful.
Its not like I know a mathematical proof, but something tells me intuitively that building a turing complete system in complete analog is basically impossible. You will always depend on some digital circuits for programmability.
@@jonaswox wow you just wrote a whole book to prove my point.
You said “the potential of analog systems is speed” that’s the exact same potential of quantum computers.
You said “analog circuits is only able to do a very specific calculation” that is the exact same as saying quantum computer can perform one single algorithm.
You said analog is not programmable in the same sense as a digital circuit. So are quantum computers based of what the videos show.
So clearly both quantum and analog are comparable. And just so that there’s no confusion, in my statement I said they are similar, I’m very much aware of the fact that quantum checks the particle position to perform its task, and analog uses frequencies to perform its tasks.
FYI, we already have analog processors to perform matrix multiplication for AI purposes. They are much faster than digital processors and use less power. We also have light processors for the same purposes as the analog processors. Both analog and light are much better suited for AI, but right now they’re just too expensive. This is what we’re going to need to solve the complex problems artificial intelligence is trying to solve. Quantum computing is probably going to smoke light and analog, but it’ll use more power to do so.
@@ElvisRandomVideos You are obviously not educated in the subject. No offense :) Im not proving your point at all :D Yes it seems like you have discovered the potential of embedded systems ;) I once was in a course on embedded systems, ... 20 years ago
Dedicated circuits for specific calculations is not a new idea at all. And you always pay with flexibility when gaining speed.
I love the symmetry in this experiment. (3 possible ways to split those functions into pairs and for each pairing there is a measurement to discern which pair the function belongs to, but we still need 2 measurements to get the 2 bits needed to identify the function.)
Yeah! Welcome to information theory :) With a classical computer, you would never be able to get beyond that. There would be different ways to measure, potentially, but it's fundamentally impossible to get more information out of it than you put in, and with a classical computer, the information you put in can only have one bit. With quantum computing, it has to turn classical at the far end (the final measurement step), so there's still a fundamental limit (with one superposition input and one measurement, you still can't tell exactly which function it is), but there are now measurement forms that involve TWO inputs.
Very cool! Makes me want to get back to making quantum circuits.
If you add a *polarizing beam splitter to the end of the circuit and 2 photoresistors to each mode (H or V) connected to an arduino, you can interface your circuit with your computer.
I did that to build a QRNG a few years back that I used to generate mazes, was the most useful circuit I could make with a single beam splitter lol
I am sorry to inform you, but you can't directly read the quantum states of photons using an Arduino and photoresistors.
Here's why:
Photoresistors measure light intensity, not quantum states: Photoresistors work by changing their electrical resistance based on the amount of light hitting them. They essentially measure the number of photons (light intensity), not the specific quantum states those photons are in.
Quantum measurement problem: Reading a quantum state is a delicate process. Observing a quantum system inherently affects its state. An Arduino and photoresistor setup doesn't have the sensitivity to measure individual photons without collapsing their superposition or entangling them in unwanted ways.
While your QRNG (Quantum Random Number Generator) project using a polarizing beam splitter and photoresistors is clever, it leverages the probabilistic nature of photon polarization, not direct quantum state readout.
The beam splitter sends photons into different paths based on their polarization (a quantum property). However, the photoresistors still only measure the resulting light intensity at each path, which is a classical outcome of the quantum process. This randomness in intensity is what you use for generating random numbers.
0:02 You are the only youtuber who says, "You don't have to watch my previous video " 😂😂
Finally a video that explains theiugh demonstration hiw quantum comouters work and what they do. Most videos repeat the same concepts about superposition, but don't explain exactly how the tech itself is set up to work.
Fantastic video. I've heard these concepts before, but a simple and elegant demonstration like this is really valuable. Great job.
Your videos are so good, have to binge watch them all now
Very happy to see this channel continue to grow and spread quantum awareness!
Wave plates have axes of their own. They shift one component in that coordinate system relative to the other component in that coordinate system. Quarter wave plates shift by 90 degrees and half wave plates by 180 degrees. If the incoming polarization has equal x and y components, then quarter wave plates convert linear polarization into circular polarization and vice versa. Half wave plates allow us to rotate an incoming polarization by any amount.
Great narrative and explanation. I love that you showed your confusion and perseverance through doubting yourself - a key part of the learning process in my experience!
I'm so excited I found this channel, I love science but don't get to scratch that itch in my line of work. This feels like I'm right there doing the experiment, thank you
[I’m writing many comments. This is 1] I really like this video. I’ll preface by saying that I’ve taken a graduate level course in quantum computing and am very familiar with quantum mechanics. As a complete introduction for someone knowing nothing, I think you jumped into polarization of light and wave-plates a bit quickly by just taking the model for granted. I imagine people with no background there will struggle to follow.
But at the same time you take things slowly and clearly. The visual aids are great.
You manage to make it very hands on and yet functional. Of course, in making a serious quantum computer a big issue is to create and preserve entanglement which you don’t really need for a 1qbit version. But that’s exactly the cheat that makes this simple enough.
I have been wanting a video like this for years lol. thank you so much!
Thank you very much for this clear explanation, I finally understand thanks to you what we mean by "quantum computer", how we can make "calculations" with light. It’s magical! Bravo for your work!!!!
I love seeing your learning process in working with this! It helps me learn along with you!
I think using a continuously shining light, and computing by changing aspects of the light is super clever!
And... If I understand correctly, the final experiment was balanced.
The notation with 1 and 0 and + and - is really confusing. It would have been helpful to rewrite the functions when introducing the quantum version of the black box. For example 0: flipped/1: not flipped. When writing with the markers on your white board different color could be used too.
(using the white board as the optical bench and for notes is genius btw!)
|+> is just the superposition of the usual 0,1 basis with a plus: |0> *+* |1> (and a global factor of 1/sqrt(2) that's not necessary to get the idea across)
|-> is the same but combined with a minus (on the 1s side): |0> *-* |1> (and same factor of 1/sqrt(2) )
So the names for |+> and |-> are self explanatory
@@Tomyb15 huh? zero plus minus doesnt equal plus.
@@authenticallysuperficial9874 my mistake. I meant to write |1> there.
@@Tomyb15 I know, but for the purposes of this explanation video they are not ideal.
I think the background is a metafor for what it feels like to watch this as a ley person.
I understood more of it than any other video about quantum computation though, so that’s good.
The quality of teaching here is extraordinary
This is a really good illustration of observation in Quantum mechanics.
We can only "test" for one of the states at a time, be it the |1>, the |0>, the |+> or the |->. What we get is defined by how we choose this basis of measurement, and none of these might actually be the states of the particle - the state of the particle just gives it some likelyhood to interact similarly, or oppositely of the state we chose to measure.
I think this experiment would be augmented nicely if the light we get at the end was to be measured by a simple photodiode and voltmeter system, so we could see exactly when the strength of the light is halved and when we get a, say, 30-60 distribution.
Also, as a random thought: we got two measurements in the end.
Would that be "cheating" when the riddle states that we can only "use" the quantum computer _once?_
It blows my mind that ANY human can come up with a q-alhorithm .....like, it's crazy. Grover's, shor, quantum courier, etc etc. it almost feels like you have to know the answer before the algorithm 🤷
Machine learning is a super cool way to explore the space of q-states. Maybe folks will use it to explore possible q-alhorithms 🤷
I like you showing the messiness of the scientific process. Well done....more should show this!
Love your explanation of duetsch-josza so far (the electron goes AGAINST the E-field though....minor pickyness. Use a positron) 👍👍👍😍😍 great supplement to Mike and ike
So true! They feel like magic when everything perfectly cancels in just the right way at the end!
Eh, it accelerates rather than "going", so its displacement will be 180° out of phase with the negative of the EM wave.
Sorry, but what perfectly canceled in the just the right way??@@LookingGlassUniverse
26:05 Yes, beam splitter. 🙂 Was waiting too long for the term "beam splitter".
And... not a question, but how this experiment is presented (figuring out which part can work as which operation, getting confused sometimes, etc.) really makes it seems like you and the audience are doing science together... which I don't see a lot :)
it made me so vicariously filled with joy to see you get emotional for finally putting the science you spent so long studying theoretically into practice :)
Technically waveplate has fast/slow axis. So, in order to make the experiment more convincing, instead of conveniently constructing F3 & F4 mystery function with simply an “empty” box, you can actually aligning the waveplate fast axis exactly to |+>, and you will get -|+>. Still, it is absolutely a great video! Appreciate it a lot.
9:58 Correct, in the opposite direction first, but it remains in the same polarization. Just phase-shifted by 180°.
I loved everything about this video. Thank you so much for putting all this effort into the experiment and explaining what's going on
NGL, I saw that it took you 40 minutes to do the verification and it made me think of what Joel Spolsky said about PhD computer science people in the industry: basically if they aren't sure they get wrapped around the axle about being sure rather than just kicking out "something". Which is great in the scientific field, not so much commercially. That said, as a not PhD computer science human in the industry who tries to balance both, I appreciate your approach and love that you took the time. I wish I had the time, which is why I am thinking about doing my PhD.
I don't think it is necessarily good for the scientific field. People take a lot of time thinking about something rather than just trying something. If you have more action you can actually test whether your idea/ way of thinking was correct or incorrect. Then you can change what you have learned during the process if it was incorrect. You may even stumble across something interesting during the process. There's a balance to be had even in the scientific field, what's the point in thinking about stuff and wasting time when you can just test it? Ofc if depends on resources, but from a data scientist perspective I find some academic pondering debilitating for some projects.
@@liambailey5630 Unfortunately, at the Ph.D. level or at the level where you extend the boundaries of science, the goal is to prove or disprove something, and fully document your process and thoughts, so that your peers can understand, reproduce, extend, or prove it wrong. Outside of that level, at the engineering or application level, the goal to "fail quickly" is indeed much more productive. In some fields, that's good, while in others... for example medicine, I'd prefer that drug companies, etc. use the complete and transparent scientific method with peer review... rather than the "fail quickly" method that doesn't always explore completely and deeply.
@@hanksimon1023 Just because you provide quick action does not mean that the methodology is not documented or that its unethical. You can test ideas quickly without pondering for months. Some drugs were discovered by accident not by thinking. Obviously, testing on subjects rather than testing ideas in a lab is different. I agree that there needs to be a balance though.
@@hanksimon1023 Science is experimental, and failing (i.e. proving something is incorrect) should be considered a good thing.
@@liambailey5630 Thalidomide was tested inadequately. And, the Pfizer CEO refused to release early data on the COVID vaccine, saying that the data were company proprietary... This is a not so subtle way of saying that his profits [and confirmation bias] were more important than peer review.
Fun video! Advice- cast a shadow on the laser paper so it can be seen better. It's a bit washed out and hard to see.
Sorry about that!!
I like how you explain, it makes me think and thus learn instead of a superficial way which I then forget when the lesson is over.
Editing suggestion: you seem to have a small but usable library of graphic assets to play with. For times when you need to insert a voice-over from the editing room, create a reusable graphic assets from those smaller assets you currently have to quickly drop in and avoid the black screen. Example: the Mad Hatter standing in the center of the frame with his bow spinning on the tree and field background. Or Alice with thoughts bubble and in the thought bubble is a 3 dots who's opacity is fading in and out in a wave. HAPPY EDITTING!
I love how exited you are!
I'm surprised that you had a quantum computing phd all this time and we didn't know about it. That's a huge asset for you!
Congrats for getting it to fully work. I think a view visuals could help making this easier to grasp - from looking at your setup I can't really see which filter is ofiented in which way. What does 45° even referemce too? Is the table surceface 0°?
You really are blessed with that teacher mentality.
You have the ability to simplify things thats easy to understand for a novice like me !
I think the physical representations of your functions might be flipped around. Shouldn’t the function that does nothing (0 -> 0 and 1 -> 1) actually be nothing (the empty box scenario)? Overall, shouldn’t your results be this: if the polarization doesn’t change from input to output then it’s a balanced function, and if it changes by 90 degrees then it’s a constant function?
She messed up spin 1/2 qubit with photon qubit, she thought light polarization flipped means 0 -> 1, which is not true.
Flipping the EM wave only introduces a Pi phase shift into original quantum state, you need a photonic Pauli-X gate to rotate the polarization by 90 degree, which can be easily achieved by two properly aligned mirrors, or a properly aligned 1/2 lambda waveplate. She claims herself a PhD student without realizing her wrong explanation cannot even convince herself.
As far as I understand, the output of 1 is represented by negating the state (08:57). So the do nothing function is |0⟩ -> |0⟩ and |1⟩ -> -|1⟩.
The "flipping the light upside down" is just shifting the phase of the light by 180 degrees... right? If this is the case, you can tell that the light is the "negative" of the initial light by combining the "negative" and the initial light together (they will cancel). Overall, maybe the computation can be seen by making a reference light (same as initial) and see how the output light interferes with the reference
Interferometer time! :D
that is not the first time i kinda understood quantum computers but i think this time i am going to remember, nobody ever did anything that detailed and yet informal way enough for me to stay that long watching
Reversibility does not sound boring! I want to hear more about that
If a system gives a constant output for any input, then you cannot guess what the input was given the output. Guessing the input means that you want to reverse the process(in your head)to know the initial state. Any process described by the Schrödinger equation is a reversible process so any quantum system has to be reversible.
Feynman's Lectures on Computation describes the connections between reversible computing and quantum computing nicely.
This gives context for the thermodynamics in computing, and a brief mention of quantum computing too: th-cam.com/video/XY-mbr-aAZE/w-d-xo.html
Even if you could calculate anything related to prime numbers - that’s such a specific task, such that it would be impossible to really get the benefits from it.
As a data scientist, I prefer using older computers (~ 5 years old) over newer / faster ones, because it helps me to understand how my programs will perform on most people’s devices. And I save money.
Quantum computers can be very expensive I hear. I think it’ll definitely benefit cryptography, like you said. It’s cool that you can make one, and understand the pros and cons.
300k strong, congratulations!
Thank you!!
This explanation was super awesome! Thank you so much :D
You just earned yourself a follow.
Fascinating stuff.
Super cool to see animations; glad to see sponsors funding some of this work! The passion and perspiration is irreplaceable tho...
PS: Nobody said there would be homework! 😮 I think the top secret box is a balanced function because only the first cuts out
Hi. I'm not AT ALL 'Super Math Guy' so excuse the 'input', PLEASE; but, "PS: Nobody said there would be homework! 😮" cracked me UP!! (I'll have to take /n times to rewatch this - and others.) Appreciate you folks! 👍 [Barry Setterfield and Halton Arp and WG Tiff and - by reference - JW Selentic, Bernand Haisch, Hal Puthoff, Timothy Boyer, Luis de la Pena (with out 'the math') are my 'stimulus'.
Great explanation and well delivered - even showing that PhD's can get get entangled when thinking about how a new material impacts the output..
I get the core concept you are putting across is the superposition created by introducing horizontal and vertificate polarisation into the same computation.
But isn't the final "test" actually two measurements? First at 45 deg; then the second at 135 deg? Hence not aligning with the "single measurement" requirement?
We don't get the answer to the final question?! How cruel!
I love these past few videos (and your past ones for that matter) because they made me understand what quantum computing is and what it isn't, as well as how it generally works. Amazing work!
What's happened at minute 21:54? When you wrote on that polymer.. what a diabolical material is that?? lol I mean it appears as if the ink passes through the polymer and writes directly onto the board, but as soon as you remove the polymer the writing disappears from the board and one can understand that it is on the polymer instead.
So strange... but anyway you did a super great job with this video. Keep sharing knowledge! :)
Lol it looked weird but I think it was just the shadow of the polymer being cast on itself from the left. It looked as if it was bent up more than it actually was
Thank you so much for your video, and for the delightful drawings that accompany it!
I'm lost at 22:15. You say that F3 takes 0 to 0 and 1 to 0. So if you put nothing in the box, 0 will go to 0 (i.e. no change) and 1 will go to 0 (that seems like a change). What am I missing?
It is just an unfortunate and confusing selection of using zeroes and ones to refer to inputs and results ( to flip or not to flip) of each function. It is better to name inputs as V and H, so outputs of F3 is the same , V and H, because you don't flip nothing. To see the difference: F4 wants "ones" for both, so it would be : Inputs V and H, outputs -V and -H . ( Even though I don't see the quantumness here (yet?))
Hmmm seems like sometimes what we use as |1> and |0> for the input is different that what we use as |1> and |0> for the output. Can we really do this? especially if we want the output of one "quantum box" to be fed to the input of another quantum box (maybe not in this early experiment)
The change in f tables (f1 was originally the identity?) was hard to follow but I guess that finally we just have rotations of multiples of 90 degrees... f1=+ 90, f2= - 90, f3=0, f4= - 180, and so we can distinguish between { f1, f2 } and { f3, f4} by preparing | 0 > +| 1 >. Is that about right? Very enjoyable series!
The part I found really hard to follow throughout the video is how you're encoding the result. It would have been good to have a table of how you encode the input bit and the output bit before you built the computer. Or alternatively show the working computer before you show how you built it. This is a major part of understanding and you barely gloss over it at the start. So while you were building the functions it wasn't clear in my head what each function did in terms of encoding; you wrote it in terms of 1 and 0, not in terms of the final encoding, and both things are necessary to understand.
this is both easy and hard to understand at the same time... I love it
I always complain when people say "factor a prime number" instead of "factor a composite number" ... but it's actually a real, hard problem that should better be called "proving a number is prime", which is very useful. If there are only 2 factors, the number itself and 1.
I think there was some flipping of the definitions of f1, f2, f3 and f4 during the video. For the original definitions at the beginning of the video I think these should be the correct implementations
|0> = horizontal polarization
|1> = vertical polarization
|+> = up and right = +horizontal +vertical at 45 degrees
|-> = down and right = +horizontal -vertical at 45 degrees
F1 - Do Nothing
laser -> 0 delay -> output = no change
F2 - Always Swap
laser -> half waveplate aligned horizontally -> half waveplate aligned vertically -> output = -horizontal -vertical (pi delay = lambda/2 delay)
Note: This wasn't the implementation shown in the video because I don't think it was shown correctly in the video due to the f2 definition changing but correct me if I'm wrong
F3 - Swap if |1> (vertical polarization)
laser -> half waveplate aligned vertically -> output = horizontal -vertical (pi delay to vertical polarization)
F4 - Swap if |0> (horizontal polarization)
laser -> half waveplate aligned horizontally -> output = -horizontal + vertical (pi delay to horizontal polarization)
Then to solve the problem
laser -> 45 degrees polarizer (up right) -> now in the |+> state -> f1/f2/f3/f4 -> 45 degrees polarizer (up right) -> light = f1/f2, no light = f3/f4
In the 1st part of your video, You´re changing the phase, not flipping the polarization, so if you change the phase of the polarized beam and add it to a non inverted phase beam you should be able to detect the change, since opposite phases should cancel out. at 16:00 now you´re flipping the polarization, that should be easier to detect, right?
Question: does it really matter that we use quantum objects or is it only a matter of sending two pieces of information at the same time (horizontal + vertical polarization)? In a classical computer we measure only the voltage. What if we measured the voltage and the current? It's just an example, this might be any classical object that just has more than one property that are sort of independent on each other.
Btw, awesome video and great effort!
You said (4:30) that it is too hard of a problem to find what is the exact function he uses (f1,2,3,4) so we're only gonna find if it is balanced or constant. But what if you use a beam splitter before the secret function and combine it with the output?
you input |+> = |0> + |1>:
if it flips the horizontal direction, after the computation you are left with - |0> + |1> which will combine with |+> to output |1>
if it flips the vertical direction, after the computation you are left with |0> - |1> which will combine with |+> to output |0>
if it flips both, you will be left with -|0> - |1> which will combine with |+> to be no light
if it flips nothing then you are left with |+>
You then measure the combined light and check if it is strongest in the vertical, horizontal, zero everywhere or 45 degrees
What do you think about this? (you do need to measure multiple angles but I hope it is allowed)
Yes, but to make multiple measurements you need to run the quantum algorithm multiple times, and then it's no faster than a classical computer. And in terms of the back story, it violates the leprechaun's "one run" rule.
Thanks for this very interesting videos! Few notes regarding the optics: I'm not sure what wave plates you were using, was it half wave plate? If so, it simply rotates the linear polarization and you can replace it solution of chiral material, sugar in water for example. The concentration or the optical path will determine the rotation.
Full wave plate only does nothing for a specific color it was designed for. Not sure you've matched the laser color and the full wave plate.
Polarizer used for photography like the one I think u use in front of the laser, are usually directional since they are made of two layers: linear Polarizer and quarter wave plate after it. I'm not sure what Polarizer you've used and what direction was in use.
Thanks again for the great, interesting videos! 😊❤
Brilliant Miss. I love your use of shadows.
Love your videos, glad you're back..
Grate chanal: I'm sure most here know that but just for general info. a wave plate or lambda plade rotates the polarisation; a lambda 1/4 plate rotates the polarisation a 90° as a lambda 2/4 plade rotates it 180° and so on.
I use them when working on advanced interferometry and holography.
Glad to see I understand how quantum computers work. I wasn't sure before I watched this video!
Hi, great video. I'm an ignorant (dropped out from high school) so I don't understand physics but I do know a little bit of IT and computer science. I keep hearing this "quantum computer" a lot but idk much about it as it seems confusing. Your projects seems to be simple, as far I'm seeing it some input laser with a polarizing filter and another polarizing filter on the middle and board to stop the light so you can see the output. That's all what I understood from the video as it is long and since I'm ignorant I really can't follow up what you saying and when I try to skip and skip to the moment of some experiment where you actually do something then i just don't understand the result because it somehow related the middle explanation things that I can't get. So let me get this straight, can you name what you made? A 1 bit full adder? or what? Could you please explain in the comment in a simplified way :) Also, what would make a 1 bit full adder made out of optical filters different than one made out from transistors? speed? so speed makes it quantum? could you please clarify in a simple manner for non student. And thanks for sharing!
14:21 So your little polarized strip represents the absolute value of the light? If we were to put it into numerical terms?
33:08 Without going back into the video... I have to remember what you put... Mkay. F1, F2 were balanced, meaning the output changes. F3 and F4 were constant, meaning it doesn't. It's not F4 because +45° cut out the light and -45° kept it. Nor is it F3 because both 45s should have cut out the light. And then since -45° showed a 0, or no change in state, then you put in F2? ... ... and now I'm thinking I'm incorrect... lol.
Also... where might one find a cute, funny, creative and intelligent quantum physicist woman? ...asking for a friend... lol XD. Well done on the video, lol. Definitely informative!
Very cool stuff! Inspiring! At the end, in the final measurement, you turn the filter two different ways. That seems like cheating, e.g., giving two bits of input, but it also seems unnecessary. Was that just to give the audience some idea of the relative levels of light? I.e., if someone already knew what's bright and what's not in the output, could they leave the final filter stationary and still know the answer?
That’s right! You don’t need to turn the filter- you can just do it at one of those two angles and infer what the other result would be by using the total brightness
I wondered about the same thing, glad you asked! I wouldn't have noticed the difference in brightness 😅
a way of detecting if flipping wave works would be to check the result of interference with original light - successful flipping would result in decstructive intereference.
Is there ANY currently theorized path to quantum computers being generally faster than 'normal' binary computing? by that I mean, replacing current desktop computers and being better at running 'normal' code? Maybe a number of qubits where the exponential scaling of possible states leads to some phase transition into general compute practicality? And/or the possibility of some yet to be discovered quantum algorithm that makes them practical for general computations? Or is it basically true that quantum computers will always only have special use cases?
Could you please do this excellent video again but in 5 minutes...? I would like a clever summary from you. Pleeeeease
Crazy how much stuff starts to make make sense the more u learn about electricity ) like the charge of atoms and electrons being the reason for its bias
It can count syntax statement total characters and separate statement by parts, and sorting arrays to those parts in a syntax statement to do square root as parallel group count. On surface level. In analog, oscillating RLC diagrams. So if a string of parked cars on city road, between one city block and two traffic signals, then if a row of cars at a red light accumulate until light goes green, that batch of cars make the length of trip from red light and pass the length of block to the last signal, in that instance the first green light has no car at its intersection so cars further back were at constant speed, those cars zoom pass the light as of it had never been red to begin with, then if the mph is 30, then the car zoomed passed the light and at 1 minute, after for 0.5 miles, the square root in syntax of length of parked cars if each car was a character byte per 1 car parked. In an analog RLC circuit. The correct answer is actually 6(5) minus 0.01 = 29.99 mph, at is 1 , so half a mile is equal to 1 minute at 30 mph, if cars zoom the green light, is equivalent to 1 cent, or Linkin park.
The answer for the final riddle is constant right?
Really interesting demo. Quick question - in classical computing we have different options for hardware implementation (eg. you can build a turing complete mechanical computer without electricity, you can build a turing complete computer from relays, or these days you can build a turing complete computer from semiconductor) ... in your video you're demonstrating an implementation that uses light and polarizing lenses/filters as the gates, but I've also seen some stuff saying that we can build quantum computers based on electron spin (I think this is why they need to be at 0 Kelvin, so thermal energy doesn't affect the spin or something?) ... question is, is there a preferred method for implementing the hardware of a quatum computer? Are light and electron spin the only two known methods right now, for implementing a quantum computer? Thanks again for your video, I've never seen this done with light before and it was super fun to watch. You're awesome! EDIT: balanced, but I still cant tell which specific one of the four functions it is..i may have missed something that helps us narrow down exactly which of f1,f2,f3,or f4 it is...
In principle, one can make a quantum computer using the phase of the electromagnetic field (the keyword is Gaussian information). It doesn't require cryogenics but has challenges of its own. In turn, spin-based quantum computers are rather a minority. I don't think anyone makes polarization-based quantum computer. These are hard to make: besides cryogenics, they require difficult controls.
Would this be similar to offering a 2.5V signal to a TTL circuit to force it to reveal more about its structure than could be obtained by simply offering 0V or 5V? (By cleverly using some glitch in the circuit that activates only by offering 2.5V)
Essentially you created a 3rd type of input 0V, 5V and 2.5V. but that is cheating. You are only allowed to input 0V or 5V. (O or 1)
It would be cool to set up the experiment with an lcd screen (older one that has linearly polarized light) with a blank image as a light source and a human eye (Haidinger's brush) as a detector to make a Human-*quantum computer :D It's likely that the waveplates will spil the fun but maybe some hint of the haidinger's brush will still be visible.
note: slowly wiggling head from side to side helps me to to see the haidinger's brush.
Hey there I think this is just a part of how you can define what quantum is about. There is also that Cambridge research with the quantum teleportation back in time then back in present with a molecule or something I can read remember it’s right now exactly but that’s sort of another example, right? Larger skill I think this can be used. Infinite way is actually if you can find the right sequence. …
But the light is not 'slowed down' in the filter, it's actually phase-shiftet by half an period, so the wave seems 'flipped'.
Am I explaining this right?
The only 'slowing down' is happening right in the filter with the refraction of light. 3B1B did a great video why light is just slowed down in a denser medium.
Upside down light and minus light? What is the correct terminology? 180° out of phase? and how the light is polarized?
11:27 Quarter wave plate? This brings up memories from way back when I used a circular polarizer on ma Praktica VLC2 SLR, that had a semi-transparent mirror and noticed the very specific blue/brown color tint when using a polarizer on it, as the semi-transparent mirror used polarization to reflect light to the viewfinder and the metering sensor for the exposure. The memories this brings back. This technology feels (and is) "ancient". 🙂 I apologize for spamming, but I am drunk. ;-)
Very cool demonstration, except for one problem - I never could see the light pass through to the screen, due to the overall illumination!
your old thumnails with original drawing are really good. i was happy to see your content isnt ai generated garbage. i truly think if you avoid using the same thumbnails as those spam bot channels you will go far albeit may take a lil bit longer but will be organic growth that sustains itself. keep up the hard work your doing it great, pat yourself on the back
Sounds like quantum computing would be useful for procedural generation in Video Games due to it being able to compute all possible inputs as one input
I wonder if you have any idea how it could be used as a tool to aid character/environmental design within the game engine
But, as far as I understood, you only need a continuous-wave laser that renders a coherent state (together with polarizers, waveplates and some interferometry) to make this calculation, right? There is nothing quantum here, just classical optics. It looks that many problems that people claim to be solved with quantum stuff can be equally solved using classical light, like implementing quantum walks. Please let me know if I'm wrong, it is something I would like to understand better. The video was amazing btw and I learned some cool stuff!
You are right: this is a classical computing device. To see where quantum computers start to shine, one must look at three-qubit realizations, for example, that same Deutsch-Jozsa algorithm for functions on two-bit inputs. For functions on one bit, there's a single input solution, for example, f(x) XOR f(NOT x), which complicates comparisons.
Thanks for you answer! I'll think about that.
Excellent video! Can you please provide a parts list and recommend what laser to use that you think is safe for eyes. Thanks
This blew my mind, the fact that you are implementing a quantum algorithm (alas the simplest, Deutsch's) with a laser pointer and some polarizers and filters either fixed with play-dough or just hold free hand... Kudos! What is the intrinsic difficulty in scaling this scheme to more qubits to get true linear optical quantum computers?
I dunno, but I suspect that the setup and testing would grow exponentially for each added qubit?
Since you were able to affect the output, it is not constant, meaning balanced. The answer is balanced.
what about use a 90° rotator to take the front/rear one and put it in down/up to analyze it more easily ?
Correction: it takes seconds for a classical computer to factor any 30-digit (decimal) integer. 100-digit in only mildly difficult. It's standard to use 616-digit (2048-bit) integers for practical purposes, like authenticating a website.
Mithuna, I noticed in the MIT Lectures they cover the Deutsch-Jozsa algorithm, and it looks different from what you did here there. I saw it again and at least the problem is the same for you and them, but I dont understand why in the MIT Lectures they begin with two particle entangled states and mention no cloning. I get the gist of your approach, create 4 unitary transforms which basically flips some of the basis vectors to encode the functions, but I dont know why the MIT one seems so much more difficult--whats the payoff of entangling there. Thanks, good thought provoking video!
Great question!
This comes down to how you assume the challenger gives you the black box. If you assume they gave you a version of the box that works the same as a classical computer then it takes an input x and a blank bit (we’ll say it’s 0) and it outputs x and f(x). Then you need to do the entanglement trick to get it into the more useful form we assume in this video. In the end, you end up with a computer that takes |x,-1>-> -1^(f(x))|x, -1>
Since this does nothing to the second qubit at all, let’s just get rid of it. Now the black box computer is in a form that’s much more useful to us
So as I see it, if you can only give a 0 or 1 as input, you can only tell if the function is either (f1 or f3) or (f2 or f4) but if you can use a quantum computer you can only tell if the function is either (f1 or f2) or (f3 or f4)? So why would you choose quantum over classical as it seems like they reduce the output state equally.