/sci/ - Science & Math

>>11820815
Learning qantum computing is easier than talking to your mom

GET THE FUCK OFF MY BOARD GET OUT GET OUT GET OUT YOU HAVE TO GO BACK GET THE FUCK OFF MY FUCKING BOARD YOU DISGUSTING PARASITE NIGGERS OUT OUT OUT >>>/out/

>>11820819
Kek

>>11820815
Fun field, lots of ways to tackle the problems.
The local gapped Hamiltonian problem is a part of quantum information that seems to involve heavily both physics theory and CS theory (and of course the requisite functional analysis).
What part of quantum computing do you like, anon?

>>11820819
HAHAHAHAHAHAHA FUNNY

>tfw no bf to teach me quantum computing
I-I know math up to ODE's so that should be sufficient. R-right?

>>11820825
I'm a popsci brainlet who has no idea what you are talking about.

>>11820885
People like you makes me want to leave this forum and never come back. I'd take 100 polfags over people like you

>>11820896
Look at this
https://arxiv.org/abs/1401.3916
This is a primer that takes you through some of the field

>>11820885
“Knowing” math up to ODEs likely means you know how to shut up and calculate up to basic ODE forms.
To get into quantum computing / information theory, you need complexity theory, linear and abstract algebra, functional analysis, classic information theory, (some, depends on your interests) quantum theory at the *very least.* Most of this is proof based, not calculation.

https://youtu.be/Eak_ogYMprk

>>11820815
What quantum gates/logical operations/operators are actually technically possible today?

Imagine the possibilities of watching porn on a quantum computer. The girl can have vagina and penis at the same time in superposition

>>11821032
Suddenly I care about quantum computing.

>>11820819
Fpbp

>>11820819
Based

What are the implications of QC in graphic / physics computations?
MMO world when?

>>11820924
On it.

>>11821469
Go back to /vg/

Im going to implement a quantum algorithm in python for my bsc thesis and then heuristically check if it can be generalized to qutrits.
Any resources/tips for doing this in python?
>>11820825
This sounds interesting, do you have a primer for topological quantum computing? Im in physics

>>11820924
>some quantum theory
what did he mean by this /sci/ ? :^)

>Quantum Computing
>>>/x/

>>11823564
I mean, yeah, some. If you’re interested in entanglement and information, you need a lot more than if you wanted to study just algorithms, where all you really have to understand how complex probability amplitudes work in the context of a running implementation.
I’m of the opinion that the most interesting physics and most interesting CS work in this field requires heavy knowledge of both, but not everyone wants to know that much.
There’s waaaay more work in quantum computation than “let’s build it.”

>>11823546
You should be able to find resources if you look up
https://www.cs.umd.edu/~xwu/mini_lib.html
http://www.mit.edu/~aram/advice/quantum.html

>>11823651
I don’t understand why you would care about any aspect of this venture that wouldn’t include the maximal focus on the quantum theory. The rest seems to pale in comparison in intrigue desu.

>>11823666
Because not everything interesting about quantum computers invokes a direct question about quantum mechanics underlying the operation of the machine. For example, being interested in the combinatorics in classical algorithms, complexity hierarchy, unique games conjecture, etc., isn’t the same as being interested in clocks, flip flops, and semi conduction / gaps and doping in the operation of the machine.
Again, I personally like subfields that invoke both heavy physics and heavy CS, like quantum Hamiltonian complexity, but some people aren’t interested in the physics, and honestly as long as they do meaningful work, that’s fine.

>>11820924
baffled you didn't mention a complete understanding of Bartlett functions and the theory of triple integrals.

>>11821032

>>11825188
jesus, how old is she, like 24? and look at the state of her face, even with makeup she looks like a 50 year witch, why do females age so badly?

>>11823574
What?

>talking about quantum computing
>talking about it
>let's talk about quantum computing

no

>>11820815
Sure. What do you want to talk about?

>>11820885
If you know linear algebra you know enough for the basics, assuming you're willing to accept some theorems on the basis of faith. If you don't know linear algebra, diffeq knowledge will not help you at all.
Obviously, if you want to get into cutting edge research, you need to know about many more topics across math, physics, and cs.

>>11820966
It's not hard to do an arbitrary 1- or 2-qubit gate with fairly high fidelity. The difficulty is stringing a large number of them together while maintaining coherence.

>>11821469
Besides incredibly fringe interests (typically made for sensationalism rather than productive research), none.

>>11823546
I don't know what you mean by "check if it can be generalized to qutrits." You can trivially embed a qutrit into a 4D qudit, which itself is just a system of 2 qubits. What's your goal with such a project?

>>11823666
I think you just need to expand your horizons more. If you think the perspectives of other fields in quantum info is uninteresting, I bet you just don't know enough about those fields. Not an insult, just an observation.

>>11826707
>if you want to get into cutting edge research, you need to know about many more topics across math, physics, and cs.
Such as?

>>11826942
For starters, functional analysis and operator algebras are a no-brainer. Even if you never touch an infinite-dimensional Hilbert space, I'm of the opinion that you should still understand the subject to a rigorous degree, at least if you want to do theory. Anyone worth their salt working in quantum should know the Stone-von Neumann theorem, for instance, even if you never work with continuous variable quantum computation.

Lie theory is also essential. This has some obvious overlaps with the above but is good to emphasize on its own. After all, quantum mechanics is just unitary operators generated by some Hamiltonian(s). Also, spin representations are important.

Besides math, well this post is already getting too long, but I'll just list some buzzword topics:
>obviously, complexity theory
>game theory -> quantum game theory (this can lead you down a deep rabbit hole about hidden variables, contextuality, cohomology, ...)
>statistical physics, Ising model + its generalizations, why we care about partition functions, gapped/gapless models, thermal states
>condensed matter topics in general; fermi/bose-hubbard models; anything topological

This is highly noncomprehensive... honestly I'm of the opinion that, if you have, say, a bachelor's level understanding of math, physics, and cs, then you should be set to learn whatever else you need as it arises. At a certain point you just have to pick it up as you go, you know? (Well I personally can't claim the cs aspect, so maybe don't take me too seriously.)

>>11827079
Not the guy you’re responding to, but I’m almost finished with my undergrad double majoring in math and CS. I’ve taken up to measure / differential forms, Galois theory, algebraic topology, gonna be either taking grad functional or self studying before applying to grad schools. On the CS side (at least what’s relevant) I have 2 semesters of grad complexity. What would you suggest learning on the physics side? I’ve taken an intro quantum mechanics course, and I started self studying thermodynamics since there were apparently some really interesting connections between CS and physics there - actually, I started getting more excited for physics when reading about the novel connections with computation.
I’m really interested in learning more about (and possibly writing in) quantum Hamiltonian complexity, since it seems like a really cool intersection of physics and CS theory. Would I be able to do this from a CS PhD program, and is there anything else you would recommend learning?

>>11820815
>D-does anyone here want to talk about quantum computing?
>D-does
Not anymore.

I'm interested in applications of quantum computing to engineering problems. Can't find any resources on that.

Anybody got some cool stuff?

>>11826707
Im not sure on this, but my guess is that just because you can embed it, doesn't mean you can get the same information out of the system with an algorithm on qubit systems.
Its an approximation algorithm that outputs a bound for the highest energy eigenvalue of an N qubit system which can be described by a 2 local Hamiltonian, which is similiar to a Max cut problem. It proceeds by taking random measurements and outputs bloch states.
The math breaks down there for qutrits as there are vectors in the bloch sphere that dont correspond to valid states.

>>11827612
This. Nothing more sickening on this planet than a man larping as a two dimensional girl.

This is why America is burning right now.

>>11827544
In terms of physics background, I'd say you'll want to be comfortable with ideas from condensed-matter/many-body physics. Definitely want to be comfortable with 2nd quantization, ladder operator algebras, that sort of stuff. Basic models like transverse-field Ising, Heisenberg, Hubbard are probably good places to start.
Stoquastic Hamiltonians / "the sign problem" are all the rage nowadays, so it probably can't hurt to learn about some classical simulation algorithms like path integral Monte Carlo. And speaking of classical algorithms, DMRG is a pretty standard thing to know about. (Also renormalization in general.)

I suspect a CS program would be fine, if not preferable, especially for a focus on complexity. You just have to make sure your applying to the right schools with the right professors for your interests. That said, definitely be open to topics you might not have considered before.

Anything else... well I think your background is more than sufficient. Like I said before, you'll pick up things fast as you need them. If you want a head start, maybe get familiar with tensor network language and matrix product state representations, those are pretty ubiquitous tools.

>>11827720
You certainly do get the same information out. You just take the 4 level system and throw out one of the levels. Whether that's practical on realistic hardware is an entirely different question, but it's certainly doable with classical simulation where you're just multiplying matrices (albeit inefficient).
As for the problem itself... well it's really late for me so maybe I'll come back to this in the morning, but it seems that the issue is just that the geometry of qutrits is complicated (iirc 3d projective Hilbert space is not a 3-sphere). But that shouldn't stop you from just applying some random 3x3 unitary matrix and computing expectation values.

>>11827883
For qutrits its an 8sphere.
I looked at my description of the problem again and i was regarded and tired.
The algorithm takes in a N level qubit 2-local Hamiltonian and outputs a product state phi , s.t. the eigenvalue of H wrt phi provides an upper bound to the maximum eigenvalue of H.
Generalizing this to qutrits is not trivial and i dont think my professor really expects me to solve this in my thesis, as he hasnt either.

>>11825269
An average human ages like 6 years in a PhD, she dropped theirs but the damage is already done.

>>11825269
An average human ages like 12 years when doing a PhD, she dropped hers but the damage is already done.

>>11827845
>I suspect a CS program would be fine, if not preferable, especially for a focus on complexity.
I see I see. I’ve started to get interested in the fundamental physics the more I’ve studied this subfield - would I be able to write more and more about physics if I started in the subfield that uses complexity to classify local, gapped Hamiltonian systems? The reading I’ve done says that such endeavors have both clear and meaningful progress on both CS and fundamental physics

>>11827920
I see. It seems your problem is not really an algorithmic one IMO, but a more analytic one. At least, that's the more interesting aspect of it to me, trying to obtain some nontrivial bounds based on the parametrization of qutrits.

As for the algorithm itself (I wouldn't call it a quantum algorithm really, since everything with product states can be done efficiently classically), if you just want to run some heuristics, seems straightforward.
>generate initial state on [math]N[/math] qutrits, [math]\bigotimes_{i=1}^N U^{(0)}_i |0\rangle[/math]
>compute [math]\langle H \rangle[/math]
>do something like gradient ascent or Nelder-Mead, constrained to the [math]U_i^{(1)}[/math]'s being unitary
As you probably already know, the main issue is how you decide the initial state, which radically affects the convergence rate (and accuracy, i.e. not getting stuck in local minima). It's going to depend entirely on the specific instance of the Hamiltonian, so I doubt there's a "good" universal ansatz. However, this does seem basically like Hartree-Fock, albeit in an entirely different setting (qubits/trits instead of fermions, generic 2-local Hamiltonians instead of the electronic Hamiltonian). Of course, quantum algorithms like VQE and QAOA have already been trying to figure out good initial conditions too (QAOA is really just VQE but with a specific structure to the ansatz and applied to diagonal Hamiltonians only).
I think an interesting thing to observe is that the variational approach will give you a lower bound to the eigenvalue, while analytic methods typically provide upper bounds. Might be interesting to see how these two estimates compare, and you could even compare against exact diagonalization for a couple of small values of [math]N[/math].

>>11828613
If you're in quantum info, you're going to do physics, there's no avoiding that. Classifying Hamiltonians is literally what the condensed matter people do, regardless of whether it's through the lens of complexity, symmetries, homology, or hell even critical exponents.

>>11828730
I see. So you’re saying the title of my PhD wouldn’t gate me from certain journals or lines of work, yes? I say this because whenever I look up resources in quantum computation and information, I see a lot of physics researchers, or at least more physics than CS. But then it gets muddied when I notice that some of the seemingly big quantum information books out there are by CS researchers.

>>11828730
>>11828750
Oh, forgot to say, thanks again for all the info, it means a lot that you went out of the trouble to do so anon

>>11828750
No, there really isn't any distinction for theoretical work. It's mainly a difference in the work culture and technical interests surrounding you, if you so choose to follow them. Big groups like Berkeley and UMD bring together physics and cs people regularly and so I imagine the department-specific culture is even less ingrained there.

As you say, plenty of big names are in CS departments. Off the top of my head: Umesh Vazirani, Thomas Vidick, Andrew Childs, Scott Aaronson are all in (and come from) CS. There's naturally more people coming from physics because of how the field is and what kind of interests it piques, not because of who's hiring.
>>11828785
No problem. I'm just doing my best to procrastinate as hard as possible right now anyways.

>>11820815
No! Nobody wants to talk with you, you lower lifeform animu! >:O

>>11828918
Ah that makes sense! I was getting worried there, thinking the field was “soft gated” by math and physics degrees, but my issues are probably that the people with the interest in the physics...would naturally come from math-phys and physics research groups more than CS groups.
Any recommendations of schools to apply for? UMD, Berkeley, university of Texas-Austin, university of Illinois - champagne Urbana, and university of Chicago are all on my radar, but they’re top schools and I’m feeling major anxiety applying

>>11828697
Thanks a lot for your perspective, although most of this is over my head right now. Im going to study and look up everything you said.
Indeed, it seems like the focus is on the parameterization of the qutrit and to get bounds out of it, as my professor really pushed the bloch sphere.
What do you do?

>>11827621
Anyone? :-(

>>11827612
>>11827735

>>11829525
You can't mêmé.

>>11821032

>>11820815
May be useful in protein folding, but any security issues proposed by quantum computing can be done on binary if you flip the equation right way.

Like there is no need e.g. for prime factorization in RSA cracking.

>>11829499
Sorry anon, I dont know much on quantum computing or anything quantum mechanics so my funky article i found might be retarded to you, but here you go.

https://singularityhub.com/2017/06/25/6-things-quantum-computers-will-be-incredibly-useful-for/

>>11820815
Every person should have to read link related before they get to spout off their opinions on the worth and value of quantum computing.

https://quantumalgorithmzoo.org/

>>11820819
As a NEET, this is unironically true.

>>11829499
Found this book while getting resources: "ivan djordevic - quantum information processing ans Quantum Error correction. an engineering approach" you should be able to get it on libgen. It reaches the subject so maybe not what youre searching for, but it might have something regarding your question in the beginning or something, you should look into it.