Ghost imaging is a counter-intuitive phenomenon-first realized in quantum optics-that enables the image of a two-dimensional object (mask) to be reconstructed using the spatio-temporal properties of a beam of particles with which it never interacts. Typically, two beams of correlated photons are used: one passes through the mask to a single-pixel (bucket) detector while the spatial profile of the other is measured by a high-resolution (multi-pixel) detector. The second beam never interacts with the mask. Neither detector can reconstruct the mask independently, but temporal cross-correlation between the two beams can be used to recover a 'ghost' image. Here we report the realization of ghost imaging using massive particles instead of photons. In our experiment, the two beams are formed by correlated pairs of ultracold, metastable helium atoms, which originate from s-wave scattering of two colliding Bose-Einstein condensates. We use higher-order Kapitza-Dirac scattering to generate a large number of correlated atom pairs, enabling the creation of a clear ghost image with submillimetre resolution. Future extensions of our technique could lead to the realization of ghost interference, and enable tests of Einstein-Podolsky-Rosen entanglement and Bell's inequalities with atoms.

Bell correlations are a foundational demonstration of how quantum entanglement contradicts the classical notion of local realism. Rigorous validation of quantum nonlocality have only been achieved between solid-state electron spins, internal states of trapped atoms, and photon polarisations, all weakly coupling to gravity. Bell tests with freely propagating massive particles, which could provide insights into the link between gravity and quantum mechanics, have proven to be much more challenging to realise. Here we use a collision between two Bose-Einstein condensates to generate spin entangled pairs of ultracold helium atoms, and measure their spin correlations along uniformly rotated bases. We show that correlations in the pairs agree with the theoretical prediction of a Bell triplet state, and observe a quantum mechanical witness of Bell correlations with $$6\sigma$$ 6 σ significance. Extensions to this scheme could find promising applications in quantum metrology, as well as for investigating the interplay between quantum mechanics and gravity.

Ghost imaging is a technique -first realized in quantum optics [1,2] -in which the image emerges from cross-correlation between particles in two separate beams. One beam passes through the object to a bucket (single-pixel) detector, while the second beam's spatial profile is measured by a high resolution (multi-pixel) detector but never interacts with the object. Neither detector can reconstruct the image independently. However, until now ghost imaging has only been demonstrated with photons.Here we report the first realisation of ghost imaging of an object using massive particles. In our experiment, the two beams are formed by correlated pairs of ultracold metastable helium atoms [3], originating from two colliding Bose-Einstein condensates (BECs) via s-wave scattering [4,5]. We use the higher-order Kapitza-Dirac effect [6] to generate the large number of correlated atom pairs required, enabling the creation of a ghost image with good visibility and sub-millimetre resolution. Future extensions could include ghost interference as well as tests of EPR entantlement [7] and Bell's inequalities [8].

The control and manipulation of quantum systems without excitation is challenging, due to the complexities in fully modeling such systems accurately and the difficulties in controlling these inherently fragile systems experimentally. For example, while protocols to decompress Bose-Einstein condensates (BEC) faster than the adiabatic timescale (without excitation or loss) have been well developed theoretically, experimental implementations of these protocols have yet to reach speeds faster than the adiabatic timescale. In this work, we experimentally demonstrate an alternative approach based on a machine learning algorithm which makes progress towards this goal. The algorithm is given control of the coupled decompression and transport of a metastable helium condensate, with its performance determined after each experimental iteration by measuring the excitations of the resultant BEC. After each iteration the algorithm adjusts its internal model of the system to create an improved control output for the next iteration. Given sufficient control over the decompression, the algorithm converges to a novel solution that sets the current speed record in relation to the adiabatic timescale, beating out other experimental realizations based on theoretical approaches. This method presents a feasible approach for implementing fast state preparations or transformations in other quantum systems, without requiring a solution to a theoretical model of the system. Implications for fundamental physics and cooling are discussed. Significance Engineering the fast evolution of a quantum system between states is a key problem to be solved in the development of quantum technologies, such as quantum computing. We experimentally demonstrate a general approach using a Machine Learning algorithm that develops a model of the system, based on previous performance, to create further educated guesses on how to improve. Applied to a system similar to moving a cup of liquid between two locations (while blindfolded) the algorithm reaches a speed faster than previous approaches, dealing well with the complex dynamics and experimental imperfections present with its empirical approach. The resulting fast dynamics open the door to understanding how quantum mechanical systems reach equilibrium while the method provides a new tool to taming complex quantum systems.

Abstract:We have developed and characterised a stable, narrow linewidth external-cavity laser (ECL) tunable over 100 nm around 1080 nm, using a single-angled-facet gain chip. We propose the ECL as a low-cost, high-performance alternative to fibre and diode lasers in this wavelength range and demonstrate its capability through the spectroscopy of metastable helium. Within the coarse tuning range, the wavelength can be continuously tuned over 30 pm (7.8 GHz) without mode-hopping and modulated with bandwidths up to 3 kHz (piezo) and 37(3) kHz (current). The spectral linewidth of the free-running ECL was measured to be 22(2) kHz (Gaussian) and 4.2(3) kHz (Lorentzian) over 22.5 ms, while a long-term frequency stability better than 40(20) kHz over 11 hours was observed when locked to an atomic reference. 175-210 (2012). 18. List of parts and instruments. We used an Innolume GM-1060-150-PM-250 gain module, a Thorlabs C240TME-1064 mounted aspheric lens, a Thorlabs GR13-1210 blazed diffraction grating, a Thorlabs KMSS/M kinematic mirror mount, a Thorlabs PA4FKW piezo chip, European Thermodynamics APH-127-10-25-S TEC modules, an Epcos S861 thermistor, and an AFW Technologies PISO-83-2-C-7-2-FB polarization maintaining in-fibre isolator in the construction of the laser. The laser was controlled using a custom-built current controller, ILX Lightwave LDT-5100 temperature controllers, and a PiezoDrive PDu-150CL piezo driver. The laser was frequency stabilised using a Brimrose TEM-250-50-10-2FP fibre-coupled AOM, an SRS SR510 lock-in amplifier, and a custom-built PI controller.