March 2020 - ITP-quantumseminare

tomorrow's group meeting @ 15:00 in zoom
by Joe Renes 30 Mar '20

30 Mar '20

Hi all, Tomorrow we'll continue with our zoom group seminar with a talk by Janka Möller, who did an externally supervised thesis on "Deep Reinforcement Learning Applied to Byzantine Agreement". See below for the abstract. The link to the meeting is https://ethz.zoom.us/j/362994444 Best, -joe Abstract: We use Deep Reinforcement Learning to simulate worst-case behaviour of failing nodes for reaching Byzantine Agreement. In particular we study the King algorithm for the synchronous communication model and the Ben-Or algorithm for the asynchronous communication model. We find that the byzantine agent successfully learns strategies leading to maximal run time in various settings of both algorithms and we study the resulting actions in detail. We conclude that Deep Reinforcement Learning is a useful tool to simulate worst-case behaviour in the Byzantine Agreement Problem.

1 0

meeting this week 15:00 Tuesday
by Joe Renes 23 Mar '20

23 Mar '20

Hi all, This week we will have a group seminar meeting, at the usual time Tuesday at 3pm. Here is the zoom link: https://ethz.zoom.us/j/362994444, and it should also show in the calendar entry. This week Alberto Rolandi will tell us about his semester project with Henrik. Details below and in the calendar. Best, -joe Title: Extensivity of Rényi entropies in translational invariant matrix product states Abstract: It has recently been shown that if a system’s energy eigenstates have extensive entanglement then the system will equilibrate. This criterion appears not to be applicable to systems having eigenstates with low entanglement, such as many-body localized systems which are well described by matrix product states (that follow an area law for entanglement). We were able to show that, despite the area law, generic translation invariant matrix product states have extensive entanglement (in disconnected sub-systems). Further, under an additional assumption, we were able to give a quantitative lower bound on the entropy density in the thermodynamic limit.

1 0

meetings cancelled
by Joe Renes 16 Mar '20

16 Mar '20

Hi all, Our large group meeting is cancelled until further notice. We may be able to resume virtually, but let's wait until our schedules stabilize a bit. Best, -joe

1 0

tomorrow's group meeting E41.1 @ 3pm
by Joe Renes 09 Mar '20

09 Mar '20

Hi all, Tomorrow we are back in E41.1 at 3pm. We'll hear from Artemiy Burov, who will present the results of his master thesis that he did under the supervision of Johannes Bausch in Cambridge. The title and abstract are below. Best, -joe Title: The Complexity of Low-Dimensional Translationally-Invariant Spin Lattices Abstract: In this presentation I will give an overview of the proof of hardness of the Translationally-Invariant Local Hamiltonian Problem for the case of 2D systems using the technique of embedding a Turing Machine into a spin lattice.

1 0

tomorrow's group meeting: 3pm in HIT F31.2
by Joe Renes 02 Mar '20

02 Mar '20

Hi all, Tomorrow we are in a different room, F31.2, as E41.1 is in use by someone else. We have a visitor talk by Chris Cade of CWI and QuSoft in Amsterdam. His title and abstract are below. Best, -joe Title: The one clean qubit model without entanglement is classically simulable Abstract: Entanglement has been shown to be necessary for pure state quantum computation to have an advantage over classical computation. However, it remains open whether entanglement is necessary for the speedups achieved by quantum computers that use mixed states. The one clean qubit model is a model of quantum computation, inspired by NMR computing, in which the input state is very close to a maximally mixed state. Despite the apparent weakness of this model, it has been shown to be able to solve some problems that appear to be intractable classically, and there are a number of complexity-theoretic results that give strong evidence that the model is indeed hard to classically simulate in general. It is fairly well known that the amount of entanglement present in these computations is severely limited, prompting very credible suggestions that entanglement is not necessary for (mixed-state) quantum computers to achieve significant speedups over classical ones, with alternative proposals often focussing on other (non-classical) correlations that are present in mixed-state computation. In this work we show that the presence of entanglement is indeed necessary for any kind of computational speedup in the one clean qubit model, and that without it, the model becomes classically simulable. Joint work with Mithuna Yoganathan.

1 0