1. Quantum phase magnification
by O. Hosten, R. Krishnakumar, N. J. Engelsen, and M. A. Kasevich
Science
Volume 352(6293):
June 24, 2016
Published by AAAS

2. Fig. 1 Conceptual description of quantum phase magnification.
Conceptual description of quantum phase magnification. (A) Illustration of the magnification protocol on the Bloch sphere. The Wigner quasi-probability distributions are shown for two separated initial CSSs (left) and after the states are magnified through collective interactions (right). Here this is shown with N = 900 atoms and a magnification of M = 3 for pictorial clarity. Experimentally we use up to N = 5 × 105 and M = 100, permitting us to concentrate on a planar patch of the Bloch sphere. (B and C) Effect of the (shearing) interaction used for mapping Jz onto Jy for a pair of different initial states with separations S and S′ = S/2; each panel shows three different magnification factors. Note that a π/2 rotation about the Jx axis needs to follow to complete the protocol. CSSs (B) and 6-dB squeezed states (C) together illustrate the requirement of larger magnifications to separate two initially squeezed states. (D) A small rotation θ about the Jx axis is added before the shearing step, eliminating the requirement of larger magnifications for squeezed states by giving rise to a refocusing of the Jy noise. At an optimal magnification (here M = 3), the noise-refocusing scheme maps the initial Jz onto Jy, preserving the SNR associated with the two initial states.
O. Hosten et al. Science 2016;352:
Published by AAAS

3. Fig. 2 Experimental setup implementing phase magnification.
Experimental setup implementing phase magnification. (A) 87Rb atoms are trapped inside a high-finesse cavity (length 10.7 cm) using a 1560-nm cavity mode as a one-dimensional optical lattice. A 780-nm mode is used to generate collective atomic interactions and to probe the cavity resonance frequency (Jz measurements) by recording the phase of a reflected probe pulse (~10 pW, 200 μs). Microwaves are for atomic state rotations. A charge-coupled device (CCD) imaging system measures the population difference between the hyperfine states after releasing the atoms from the lattice and spatially separating the states. (B) Because of the commensurate frequency relationship between the trapping laser and the interaction/probe laser, all atoms are uniformly coupled to the 780-nm mode. (C) The 780-nm mode couples the two hyperfine clock states separated by ωHF to the excited manifold with opposite detunings. Thus, the two states pull the intracavity index of refraction in opposite directions, leading to a cavity frequency shift proportional to Jz. (D) The mechanism leading to the collective atomic interactions ( Hamiltonian) that enables the magnification process: linking of the intracavity power to Jz, producing a Jz-dependent ac-Stark shift. The frequencies of the interaction beam νint and probe beam νprobe are indicated.
O. Hosten et al. Science 2016;352:
Published by AAAS

4. Fig. 3 Characterization of the basic magnification process with CSSs.
Characterization of the basic magnification process with CSSs. (A) Sample distributions (400 samples each) comparing the cavity-based measurements of Jz with fluorescence imaging–based measurements after a magnification of M = 45. The two distributions in each plot correspond to different initial states with 〈Jz〉 = ±200 prepared using 2 × 105 atoms. (B and C) Magnification of the separation between the two distributions as a function of accumulated ac-Stark shift phase ϕAC imparted on the atoms at fixed cavity-light detuning of 36 kHz (B) or as a function of cavity-light detuning δ0 at fixed ϕAC = 0.6 rad (C). Solid lines are fits to the data as a function of ϕAC and δ0, respectively, in Eq. 1. Fitted curves agree with theoretical curves (not shown) to within 10%. (D) SNR associated with the two distributions as a function of the magnification parameter, normalized to that obtained by the cavity measurements (normalized SNR). Magnification is varied by changing ϕAC. The solid line is a fit of the form M/(α2 + M2)1/2; the fit parameter α contains information primarily about fluorescence detection noise. In (B) to (D), error bars and shaded regions denote the 68% statistical confidence interval for data and fits, respectively.
O. Hosten et al. Science 2016;352:
Published by AAAS

5. Fig. 4 Magnification process with noise refocusing using 8-dB squeezed spin states.
Magnification process with noise refocusing using 8-dB squeezed spin states. (A) Post-magnification Jz noise in units of CSS noise for different amounts of prior rotation θ about the Jx axis (see Fig. 1D). Solid lines are a global fit to the entire data set with two free parameters: dθ/dt (the rate of change in θ with microwave pulse time) and the Jz noise of the initial squeezed states. Obtained values are within 15% of the calculated values. The inset shows the distribution of two separated 8-dB squeezed initial states (5 × 105 atoms) as identified by cavity measurements [to be compared with the M = 30 distribution in (B)]. The dashed line shows M × (Jz noise contribution from the initial squeezed states); the dotted line shows M × (CSS noise). Error bars denote the 68% statistical confidence interval. (B) The distributions after the magnification protocol at the indicated M values for θ = 29 mrad. The normalized SNR becomes 0.96 ± 0.06 at M ≈ 1/|θ| (middle histogram), where the dashed line is tangent to the θ = 29 mrad noise curve in (A). For M values to either side of 1/|θ|, the two distributions blur into each other (top and bottom histograms).
O. Hosten et al. Science 2016;352:
Published by AAAS