The presentation style is like 3blue1brown, but this is for Quantum computing. I just checked all your videos, and I loved them. This is quality content!
I think it would be fascinating to see you code on a quantum computer. It would be something fresh and different from what's already being done. I've been studying quantum physics on arXiv, along with computer science, and I find it such an exciting area. Recently, I've also started exploring quantitative biology-it’s incredible how these fields intersect!
Got it! In the meantime while I work on this - check out some of my other videos. In my series on quantum algorithms I do code a quantum computer so if you want to check that out it may be of interest.
@@Lukas-Lab yea but still like it would be kinda sick seeing like textbooks in your avg youtube format yk. Somehow its easier than reading, even if its just a slideshow of random ahh images
If this algorithm is going to make billions, I want a piece of it, I’ve worked as a medicinal chemist for years, but I’m not using my skillset anymore to make money for biotech, I most certainly am collaborating with other disciplines & doing my own thing. Never again.
Nice video Lukas, but did you know that VQE is generally regarded to be an inefficient algorithm for quantum chemistry nowadays? It doesn't scale well because the variational optimization relies on gradients. Those gradients are exponentially supressed with the size of the problem at hand. It is basically the same problem they have in training neural networks, just much worse in quantum computing.
Yeah, thanks for sharing this it’s good information to have in the comments section! I tried to address this at 2:10 when I said that there are some significant problems with VQE and linked to issues In the description. At the end of the day - VQE probably isn’t the killer application, but quantum chemistry is going to be a strong application in general and I wanted an algorithm that someone can easily go and code which has broad support and a lot of documentation.
@@Neuroszima it's difficult to say. Most ground state energy estimation algorithms suffer from query complexity that is difficult to properly quantify. The reason is that they need good approximate ground states as inputs, and preparing those good ground states may be prohibitively difficult. I think that for quantum chemistry applications we still don't have any that will yield exponential quantum advantage. Even quantum phase estimation may turn out not to be useful for industrial applications, the gap from research applications to profit is much larger than most people realise in quantum computing. But you can check out techniques like statistical phase estimation, or greens function based methods. They are interesting and potentially could run on modest quantum devices.
I'm curious, how much money would it cost to actually run this on a quantum computer? IBM quantum charges something like 100 dollars per minute of usage
I take it the 3 qubits stand for the 3 hydrogen atoms. How is that justified and would it also work for 3 larger atoms, say carbon atoms for example? Thx
For larger atoms you need more qubits to simulate, you can think of the qubits as relating to the number of electrons in the simulation, not the number of atoms. Since hydrogen has 1 electron it maps nicely in that case.
si hay una forma de comunicarce contigo, me interesaria creo que tienes mucho talento y pienso que se puede poner al servicio de la evolucion de nuestra evolucion, si puedes dejame una forma de comunicarme contigo para mostrarte alguna cosas que creo que te facinaran ver.
@Lukas-Lab yeah I was forced to use it on my mit applied data science course. But when I trained neural networks, training took 20 minutes per net, where on my PC (I have ryzen 5 3600) it took like 5x faster on average.
Here before this becomes standard curriculum in 5 years
🫡
@@Lukas-Lab
No. U.
🫡
insanely underrated, this is amazing!
I hope you get popular someday. :D
Thanks so much!!
The presentation style is like 3blue1brown, but this is for Quantum computing. I just checked all your videos, and I loved them. This is quality content!
Thanks! Being compared to 3b1b is the ultimate praise lol
I think it would be fascinating to see you code on a quantum computer. It would be something fresh and different from what's already being done. I've been studying quantum physics on arXiv, along with computer science, and I find it such an exciting area. Recently, I've also started exploring quantitative biology-it’s incredible how these fields intersect!
Got it! In the meantime while I work on this - check out some of my other videos. In my series on quantum algorithms I do code a quantum computer so if you want to check that out it may be of interest.
@@Lukas-Lab Great!
I’m glad to have met you!
This guy explains stuff really well!
Thanks!
Lowkey I wish grad level youtube videos go mainstream soon.
So much more enjoyable than going through shitty papers
Thanks! Although to be honest - this is not nearly as detailed as any paper on the topic would be. So I’d still recommend reading the good papers.
@@Lukas-Lab yea but still like it would be kinda sick seeing like textbooks in your avg youtube format yk. Somehow its easier than reading, even if its just a slideshow of random ahh images
If this algorithm is going to make billions, I want a piece of it, I’ve worked as a medicinal chemist for years, but I’m not using my skillset anymore to make money for biotech, I most certainly am collaborating with other disciplines & doing my own thing. Never again.
great video!
Thanks!
Boosting for the algo 🔥
OUTSTANDING
lets goo i will finally get myself a personal quantum computer when you hit 100M subscribers and you make a giveaway
I thought people didn’t watch until the end 😭😭
Nice video Lukas, but did you know that VQE is generally regarded to be an inefficient algorithm for quantum chemistry nowadays? It doesn't scale well because the variational optimization relies on gradients. Those gradients are exponentially supressed with the size of the problem at hand. It is basically the same problem they have in training neural networks, just much worse in quantum computing.
Yeah, thanks for sharing this it’s good information to have in the comments section!
I tried to address this at 2:10 when I said that there are some significant problems with VQE and linked to issues In the description.
At the end of the day - VQE probably isn’t the killer application, but quantum chemistry is going to be a strong application in general and I wanted an algorithm that someone can easily go and code which has broad support and a lot of documentation.
Oh, I'm sorry I think I didn't catch that. Good explanation!
@@drokles what do you recommend as a go-to solution nowadays then? I mean the algorithm
@@Neuroszima it's difficult to say. Most ground state energy estimation algorithms suffer from query complexity that is difficult to properly quantify. The reason is that they need good approximate ground states as inputs, and preparing those good ground states may be prohibitively difficult. I think that for quantum chemistry applications we still don't have any that will yield exponential quantum advantage. Even quantum phase estimation may turn out not to be useful for industrial applications, the gap from research applications to profit is much larger than most people realise in quantum computing. But you can check out techniques like statistical phase estimation, or greens function based methods. They are interesting and potentially could run on modest quantum devices.
@@drokles great thanks
I'm curious, how much money would it cost to actually run this on a quantum computer? IBM quantum charges something like 100 dollars per minute of usage
I take it the 3 qubits stand for the 3 hydrogen atoms. How is that justified and would it also work for 3 larger atoms, say carbon atoms for example? Thx
For larger atoms you need more qubits to simulate, you can think of the qubits as relating to the number of electrons in the simulation, not the number of atoms. Since hydrogen has 1 electron it maps nicely in that case.
si hay una forma de comunicarce contigo, me interesaria creo que tienes mucho talento y pienso que se puede poner al servicio de la evolucion de nuestra evolucion, si puedes dejame una forma de comunicarme contigo para mostrarte alguna cosas que creo que te facinaran ver.
Very well explained !!! Run on the real QC
🫡 alright the people have spoken, I’ll see what I can do
@@Lukas-Lab On IBM Quantum systems you get 10 mins of QPU usage every month, maybe try to the latest 156Qubit Heron devices
Yeah unfortunately this would probably take much longer than that
@@Lukas-Lab No worries, take your time. Quality content takes time.
I guess I know what I'm going to be studying now😂
Feel free to follow along with the notebook in the description :)
Thumbnail of this video looked like something else🤔
I love you❤
If big pharma gonna pour their millions into quantum tech then I don't mind it lol
Nice work, even though i presonally dislike colab
Thanks! I choose colab because it’s super accessible so it lets the most people try it, but I don’t really use it much for real projects.
@Lukas-Lab yeah I was forced to use it on my mit applied data science course. But when I trained neural networks, training took 20 minutes per net, where on my PC (I have ryzen 5 3600) it took like 5x faster on average.
Science biiiiiiiiiiiiiiiiiiiiiiiiiiiiiliard ball
Comment for engagement
Ah yes, becuse pharma definitely needs more money