The boson-sampling problem has triggered a lot of interest in the scientific community because of its potential of demonstrating the computational power of quantum interference without the need of non-linear processes. However, the intractability of such a problem with any classical device relies on the realization of single photons approximately identical in their spectra. In this paper, we discuss the physics of boson sampling with non-identical single-photon sources, which is strongly relevant in view of scalable experimental realizations and triggers fascinating questions in the complexity theory.
The boson-sampling problem has triggered a lot of interest in the scientific community because of its potential of demonstrating the computational power of quantum interference without the need of non-linear processes. However, the intractability of such a problem with any classical device relies on the realization of single photons approximately identical in their spectra. In this paper, we discuss the physics of boson sampling with non-identical single-photon sources, which is strongly relevant in view of scalable experimental realizations and triggers fascinating questions in the complexity theory.
Multiphoton interference is at the very heart of quantum foundations and applications in quantum sensing and information processing. In particular, boson-sampling experiments have the potential to demonstrate quantum computational supremacy while relying on only multiphoton interference in linear optical interferometers. However, even when photonic losses are negligible, scalable experiments are challenged by the rapid decrease of the probability of the success of current schemes with probabilistic sources for a large number of single photons in each experimental sample. Remarkably, we show a boson sampling scheme in which the probability of success increases instead of decreasing with the number of input photons, eventually approaching a unit value even with nondeterministic sources. It is achieved by sampling measurements not only at the interferometer output but also over all the possible numbers of photons at each input port and, different from scattershot boson sampling, allowing a random number of occupied input ports from one experimental run to another. Therefore, these results provide an exciting route toward future demonstrations of quantum computational supremacy with scalable experimental resources.
The non-classical interference undergone by multiple photons in passive, linear optical networks gives rise to a surprising computational power of such quantum systems. This is prominently highlighted by boson sampling, a scheme for quantum computation which relies solely on such multiphoton interference and omits the usage of more sophisticated techniques such as entangling gates or feed-forward control. Despite these limitations, it still allows to efficiently solve a class of computational problems which is intractable for classical computers. In this thesis, several open challenges on the way to an experimental realization of boson sampling are addressed. To that end, the idealized model for boson sampling is extended to account, for example, for inner degrees of freedom of the single-photon pulses or the probabilistic nature of single-photon sources. It is shown that the difficulties arising from these real-world effects can be overcome by suitably adapting existing boson sampling schemes. Additionally, the extended model for such interference is applied to reveal its intricate dependence on input state, evolution inside the interferometer, and the detection process. Finally, a novel description of SPDC in terms of temporal Bogoliubov transformations is provided, revealing new, intuitive insights into the temporal structure of photon correlations for such sources.
Boson sampling is a computational problem that has recently been proposed as a candidate to obtain an unequivocal quantum computational advantage. The problem consists in sampling from the output distribution of indistinguishable bosons in a linear interferometer. There is strong evidence that such an experiment is hard to classically simulate, but it is naturally solved by dedicated photonic quantum hardware, comprising single photons, linear evolution, and photodetection. This prospect has stimulated much effort resulting in the experimental implementation of progressively larger devices. We review recent advances in photonic boson sampling, describing both the technological improvements achieved and the future challenges. We also discuss recent proposals and implementations of variants of the original problem, theoretical issues occurring when imperfections are considered, and advances in the development of suitable techniques for validation of boson sampling experiments. We conclude by discussing the future application of photonic boson sampling devices beyond the original theoretical scope.
A boson-sampling device is a quantum machine expected to perform tasks intractable for a classical computer, yet requiring minimal nonclassical resources as compared to full-scale quantum computers. Photonic implementations to date employed sources based on inefficient processes that only simulate heralded single-photon statistics when strongly reducing emission probabilities. Boson sampling with only single-photon input has thus never been realized. Here, we report on a boson-sampling device operated with a bright solid-state source of single-photon Fock states with high photon-number purity: the emission from an efficient and deterministic quantum dot-micropillar system is demultiplexed into three partially indistinguishable single photons, with a single-photon purity 1-g^{(2)}(0) of 0.990±0.001, interfering in a linear optics network. Our demultiplexed source is between 1 and 2 orders of magnitude more efficient than current heralded multiphoton sources based on spontaneous parametric down-conversion, allowing us to complete the boson-sampling experiment faster than previous equivalent implementations.
The boson sampling problem has brought a lot of attention in the quantum information field because it is not efficiently solvable with a classical computer; nonetheless it can be implementable with linear optical interferometers with single-boson sources. Recently, we introduced a more general problem, the multi-boson correlation sampling problem, which allows to take advantage of the muti-mode spectral distribution of the bosonic sources together with time-correlated measurements in order to achieve a sampling not only over the output channels of the interferometer but also over the joint detection times. This problem was analyzed for both single-photon sources and thermal sources. In this work, we demonstrate that it is possible to use single qubit bosonic sources in order to sample not only in the described "space-time" degree of freedom but also over all the possible exponential number of multi-qubit output states.
Abstract Boson Sampling is a task that is conjectured to be computationally hard for a classical computer, but which can be efficiently solved by linear-optical interferometers with Fock state inputs. Significant advances have been reported in the last few years, with demonstrations of small- and medium-scale devices, as well as implementations of variants such as Gaussian Boson Sampling. Besides the relevance of this class of computational models in the quest for unambiguous experimental demonstrations of quantum advantage, recent results have also proposed the first applications for hybrid quantum computing. Here, we introduce the adoption of non-linear photon–photon interactions in the Boson Sampling framework, and analyze the enhancement in complexity via an explicit linear-optical simulation scheme. By extending the computational expressivity of Boson Sampling, the introduction of non-linearities promises to disclose novel functionalities for this class of quantum devices. Hence, our results are expected to lead to new applications of near-term, restricted photonic quantum computers.