Re: [Qirg] Question on the Network Model for Quantum Networks

[Wojciech Kozlowski <[e-mail]>]

On 28/03/2019 07:38, Gelard Patrick wrote:
>
> -----Message d'origine-----
> De : Qirg <[e-mail]> De la part de Marcello Caleffi
> Envoyé : mercredi 27 mars 2019 21:28
> À : Patrick Gelard <[e-mail]>
> Cc : Wojciech Kozlowski <[e-mail]>; [e-mail]
> Objet : Re: [Qirg] Question on the Network Model for Quantum Networks
>
> Hi Patrick,
>
>> Il giorno 27 mar 2019, alle ore 20:40, Patrick Gelard <[e-mail]> ha scritto:
>>
>> Hi,
>>
>> Software-Defined Networking (SDN) could also be use to orchestrate and automate/program the reservation of classical networking resources. And therefore by analogy SDN could be used to program the entanglement routing.
>>
>>>> I would say that quantum networking aims at teleporting qubits between remote nodes
>> it seems to me that quantum networking can be used for other quantum services which don't use teleportation  (QKD based on entanglement, super dense coding, ...) !!!
>>
> Indeed.
>
> But, m2c, I seriously doubt super dense coding could ever be the killer application for a network paradigm shift such as the one required by entanglement distribution.. It seems much easier (and less expensive) to simply double the bandwidth..
While super dense coding may not be the killer application, I am more in 
favour of the view that the network only provides the entanglement and 
teleportation is one of the services/applications on top of it - just 
like TCP is a reliable delivery service built on top of the IP 
best-effort datagram service.

This distinction comes from the fact that most applications currently 
proposed use entangled states as their building block and as Patrick 
pointed out they may not teleport any quantum state. You could implement 
all applications using teleportation, but this might lead to situations 
where you create a Bell Pair in order to teleport a qubit that is part 
of a Bell Pair since what the application needs is a Bell Pair.

Some applications will need teleportation, but for some it will be 
unnecessary overhead, especially in the case of anything QKD based as 
Patrick pointed out.
> QKD, well, again m2c, QKD can (might? not sure whether there is a definitive answer in terms of practical key rate) benefit from entanglement -- depending on the scheme — whereas distributed quantum computing (or blind computing) seems quite strongly dependent on entanglement and they both may be killer application since no entangled-less network can provide such services
>
> Anyhow, these are my personal feelings .. What about? Don’t you believe we should generalize quantum networks to it?
>
> [Patrick] In classical Network we have unicast, multicast and aycast IP communication service, thus why not have different modes of connectivity in quantum networks that can increase, for example, distributed computing power. Building  To be able to build interconnection network graphs of multipartite entanglement would be relevant in my opinion.  Now what are the physical challenges to achieve multipartite entanglement ?
>   
Multipartite entanglement is definitely a service that is desirable - 
there are proposed applications that build on top of GHZ states and some 
that use W states (both are multipartite). However, to me, what is not 
clear is where it should be done. One could generate these states at the 
physical layer and then distribute them via teleportation. Another 
option is to build these states up from Bell Pairs. This may be less 
efficient, but it may also not be - this is not clear yet. In this case 
one would have to answer whether it's the network stack that does this, 
or it's a service layer on top of the network stack (layer 4 like 
TCP/UDP) or it should be done by the application itself.

One of the members of this list, Alexander Pirker, has done some work on 
building quantum networks using multipartite states. You can read more 
about it here: https://arxiv.org/abs/1810.03556.

>> /Patrick
>>
>
>
>> Le 27/03/2019 à 18:24, Marcello Caleffi a écrit :
>>>> Il giorno 27 mar 2019, alle ore 14:32, Wojciech Kozlowski <[e-mail]> ha scritto:
>>>>
>>>> On 27/03/2019 14:08, Stephen Botzko wrote:
>>>>> This whole area is new to me, so it's quite likely my question has already been thought about.
>>>>>
>>>>> My understanding is that the goal of quantum networking is to create (and possibly manage) entangled qubits.  Each entangled qubit is an object that exists in precisely two locations at any point in time, and has other properties (in particular fidelity).
>>>>>
>>>> The goal of quantum networking is to create, distribute, and manage *Bell Pairs* which consist of two qubits split across two locations. It's not one qubit in two places.
>>> I would say that quantum networking aims at teleporting qubits between remote nodes (which may or may not — then entanglement swapping seems a suitable candidate — be directly connected through quantum links).
>>>
>>> Teleporting needs entanglement.
>>>
>>> Bell pairs are the simplest form of entanglement. But they are not the only form of entanglement.
>>>
>>> I believe that sooner or later we should start looking into multipartite entanglement as done by the physics community in the last ten years or so.
>>>
>>>>> I am wondering why we are trying to map classical networking concepts like "quantum connections" and packet store+forward onto this problem domain.  Has anyone looked at using a different framework (for instance content defined networking)?  It might be a more natural starting point.
>>>>>
>>>> The reason for starting with the standard framework is that it is well understood, there is a wealth of available literature on the topic, and also there are many experts who understand the subject. A lot of people have also noticed similarities between certain classical protocols like RSVP and the problem of distributing Bell Pairs so it seems natural to pursue this direction.
>>>>
>>>> However, I wouldn't say that this is the only way to go or that some other framework might not be more optimal. In fact, Rod, one of the group's chair, has for example proposed to use a recursive architecture (https://www.nii.ac.jp/pi/n8/8_65.pdf). I haven't heard anyone suggest content defined networking for quantum yet, but if one were to come up I'm sure the group would also be interested in such a proposal.
>>>>
>>>>> Also, I am thinking that it might be useful to construct a data model that shows how entangled qubits are created, moved, destroyed, etc. This could also show any physics constraints (for instance, cases where bits need to be transported on the quantum link). The tutorial shows some of this, but I think it is incomplete.
>>>>>
>>>> That is a good point, and might be worth adding to the "architectural principles" work which will be going on in the near future.
>>>>
>>>>> Stephen
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> Qirg mailing list
>>>>>
>>>>> [e-mail]
>>>>> https://www.irtf.org/mailman/listinfo/qirg
>>>> _______________________________________________
>>>> Qirg mailing list
>>>> [e-mail]
>>>> https://www.irtf.org/mailman/listinfo/qirg
>>>
>>>
>>> _______________________________________________
>>> Qirg mailing list
>>>
>>> [e-mail]
>>> https://www.irtf.org/mailman/listinfo/qirg
>> 	Garanti sans virus. www.avast.com
> _______________________________________________
> Qirg mailing list
> [e-mail]
> https://www.irtf.org/mailman/listinfo/qirg