We apply an online optimization process based on machine learning to the production of Bose-Einstein condensates (BEC). BEC is typically created with an exponential evaporation ramp that is optimal for ergodic dynamics with two-body s-wave interactions and no other loss rates, but likely sub-optimal for real experiments. Through repeated machine-controlled scientific experimentation and observations our ‘learner’ discovers an optimal evaporation ramp for BEC production. In contrast to previous work, our learner uses a Gaussian process to develop a statistical model of the relationship between the parameters it controls and the quality of the BEC produced. We demonstrate that the Gaussian process machine learner is able to discover a ramp that produces high quality BECs in 10 times fewer iterations than a previously used online optimization technique. Furthermore, we show the internal model developed can be used to determine which parameters are essential in BEC creation and which are unimportant, providing insight into the optimization process of the system.

A Bose-Einstein condensate is used as an atomic source for a high precision sensor. A 5 × 10 6 atom F=1 spinor condensate of 87 Rb is released into free fall for up to 750 ms and probed with a T = 130 ms Mach-Zehnder atom interferometer based on Bragg transitions. The Bragg interferometer simultaneously addresses the three magnetic states, |m f = 1, 0, −1 , facilitating a simultaneous measurement of the acceleration due to gravity with a 1000 run precision of ∆g/g= 1.45 × 10 −9 and the magnetic field gradient to a precision 120 pT/m.Acquiring accurate and precise data on magnetic and gravity fields is critical to progress in mineral discovery [1,2] Like their classical counterparts, sensors based on cold atoms measure the trajectory of the test particles [19]. Unlike classical particles, atoms offer internal degrees of freedom, allowing for the possibility of additional simultaneous measurements including time, magnetic fields and magnetic field gradients. Although these advantages are intrinsic to all atomic sources, ultra-cold BoseEinstein condensates (BEC) offer additional benefits over thermal atoms. An intrinsic feature of a BEC is a spatial coherence equivalent to the size of the cloud which is generally 100's of µm while thermal sources have spatial coherence length on the order of the de Broglie wavelength,mk B T (∼ 0.1µm). This spatial coherence has been shown to provide robustness to systematics which result in loss of fringe contrast such as cloud mismatch at the final beam splitter pulse [20]. The BEC then allows a sensor to be operated unshielded in varying environments where background field gradients and curvatures are non-negligible.This letter introduces a new type of sensor which simultaneously measures gravity and magnetic field gradients to high precision. In this lab based sensor, an optically trapped cloud of 87 Rb atoms is cooled to condensation and projected into an F = 1 spin superposition, then passed through a vertical light pulse Mach-Zehnder interferometer based on Bragg transitions [21,22]. The spin superposition in combination with the large spatial coherence of the BEC allows simultaneous precision measurement of gravity and absolute magnetic field gradient in an unsheilded device [23]. A 2 × 10 6 atom condensate is used in the interferometer [24] with no loss in contrast over all interferometer times.The experimental schematic is shown in Figure 1. A hot sample of 87 Rb atoms is created and precooled in a 2D magneto-optical trap (MOT). These precooled atoms are transferred to an aluminum ultra-high vacuum cell via a high impedance gas flow line and blue detuned push beam. In 6 s, 5 × 10 9 atoms are collected in a 3D MOT where a standard compression and polarization gradient cooling sequence is applied achieving a 20 µK temperature. The atoms are then loaded into a hybrid magnetic quadropole and crossed optical dipole trap. An initial stage of evaporation is completed using a microwave knife over 4.5 s leaving 4×10 7 atoms at 4 µK and a phase space density of 1 × 10 −4 . The magne...

We observe the breakup dynamics of an elongated cloud of condensed 85 Rb atoms placed in an optical waveguide. The number of localized spatial components observed in the breakup is compared with the number of solitons predicted by a plane-wave stability analysis of the nonpolynomial nonlinear Schrödinger equation, an effective one-dimensional approximation of the Gross-Pitaevskii equation for cigar-shaped condensates. It is shown that the numbers predicted from the fastest growing sidebands are consistent with the experimental data, suggesting that modulational instability is the key underlying physical mechanism driving the breakup.