Device-Independent Quantum Information Processing represents a new paradigm for quantum information processing: the goal is to design protocols to solve relevant information tasks without relying on any assumption on the devices used in the protocol. For instance, protocols for device-independent key distribution aim at establishing a secret key between two honest users whose security is independent of the devices used in the distribution. Contrary to standard quantum information protocols, which are based on entanglement, the main resource for device-independent quantum information processing is quantum non-locality. Apart from the conceptual interest, device-independent protocols offer important advantages from an implementation point of view: being device-independent, the realizations of these protocols, though technologically challenging, are more robust against device imperfections. Current and near-future technology offer promising perspectives for the implementation of device-independent protocols.

The project DIQIP explores all these fascinating possibilities. Its main objectives are (i) obtaining a better characterization of non-local quantum correlations from an information perspective, (ii) improve existing and derive new application of this resource for device-independent quantum information processing and (iii) design feasible implementations of device-independent protocols. We plan to tackle these questions with an inter-disciplinary approach combining concepts and tools from theoretical and experimental physics, computer science and information theory.

The project has already led to many significant results. Among those, we highlight in what follows a few that are representative of the work done so far. First, in relation to the study of non-local correlations, a joint publication by partners in the consortium proved that models that postulate the existence of hidden influences to explain quantum correlations are incompatible with the no-signaling principle. This result represents a significant strengthening of the seminal no-go theorem on hidden-variable theories. Second, the first example of genuine hidden non-locality was obtained by members in the consortium. Hidden non-locality is an intriguing phenomenon displayed by some entangled states. The correlations obtained when measuring these states have a classical explanation, but this explanation is no longer valid when the state is subjected to a sequence of measurements. Evidence for this phenomenon had been obtained in the past but the strict proof of its existence was obtained within DIQP during 2013. We also proposed the first intrinsically multipartite principle for quantum correlations. A significant effort has been devoted in the last years to derived quantum correlations from information principles. It was shown that multipartite

Principles are necessary for this project to succeed. A joint publication of DIQIP partners provided the first example of such a principle.

Moving to device-independent protocols, it was shown that randomness amplification, a task that is known to be impossible in classical information theory, becomes possible when exploiting the non-local correlations obtained from entangled states. Randomness amplification is an information task in which the goal is to produce random numbers of arbitrarily good quality from a source of imperfect random bits. It was shown that randomness can be amplified in the quantum regime, under solely the assumption of validity of the no-signaling principle. First, a protocol for randomness amplification was provided, showing that initial bits of high, yet not perfect, randomness could be amplified. Second, a protocol attaining full randomness amplification, that is, such that any source of bits of arbitrary yet non-vanishing randomness can be distilled into a new source of bits of arbitrarily good randomness, was derived. Moving to security proofs of device-independent protocols, a systematic method to quantify the intrinsic randomness generated in non-locality experiment was provided. The question is very natural: given some observed non-local quantum correlation, how much randomness do they contain? The obtained method is simple but extremely powerful, as it provides the amount of randomness that can be optimally certified by some observed non-local correlations. It finds application in any protocol based on randomness certified by Bell non-locality, such as the existing protocols for randomness generation or quantum key distribution. Finally, novel architectures were proposed for the experimental implementation of device-independent quantum key distribution. For instance, schemes based on Bell tests performed locally, or on the heralded absorption of photons were analysed and shown to be promising candidates for the realization of these protocols.

Finally, and despite being a fully theoretical proposal, the project also led to experimental realizations of the theoretical ideas developed by the consortium. For instance, some of the project partners collaborated in the first experimental demonstration of the concept of dimension witnesses. This concept allows one to experimentally bound the dimension of an unknown quantum system. It also allows for the certification of the quantum nature of a system, if its dimension is assumed. This certification is at the basis of semi-device-independent quantum information protocols. Second, we also collaborated in the experimental demonstration of the concept of device-independent entanglement witnesses. Here, the goal is to certify that a multipartite system is entangled without making any assumption about its internal working, such as the Hilbert space dimension. A recent experiment demonstrated all these issues. We also worked on the experimental demonstration of a recent no-go theorem obtained by partners in the consortium for models in which the quantum state represents mere information about an underlying physical state of the system.

Finally, some project partners collaborated with the group of Prof. Kwiat in a new demonstration of a detection-loophole-free Bell test. Photons that are not detected are the main problem for the experimental implementation of device-independent protocols. Having a Bell test that is free from the detection loophole represents a significant improvement and opens the way to the experimental implementation of device-independent protocols. In this context, the new experiment is a significant step for the realization of schemes of device-independent randomness generation with acceptable rates.

All these results provide an overview of the three main types of questions addressed in the project: (i) understand the main differences between the classical and quantum formalism, especially at the level of quantum correlations, (ii) exploit these differences to design new or improve existing quantum information protocols and (iii) study or collaborate in the experimental implementation of these ideas. Work for the remaining of the project will proceed along these lines. Our long-term objective is that the project will establish a solid ground basis for a device-independent approach to quantum information applications and foundations of quantum physics.

Project partners

• Institute of Photonic Sciences (ICFO) - Spain
• GAP-Optique - Switzerland
• ETH (GAP-ETH) - Switzerland
• Laboratoire de recherche en informatique (LRI) - France
• Université Libre de Bruxelles (ULB) - Belgium
• Laboratoire d’Information Quantique - Belgium
• University of Bristol (UoB) - United Kingdom
• University of London Royal Holloway (RH)- United Kingdom
• LIAFA - France
• Université Paris Diderot - Paris 7 - France