Source-linked AI summary
A high-accuracy algorithm for designing arbitrary holographic atom traps
M. Pasienski, B. DeMarco
TL;DR
The paper addresses the difficulty of computing holograms for arbitrary optical intensity profiles, which matters for high-quality cold-atom traps. It presents the mixed-region amplitude freedom algorithm and finds substantially better accuracy and roughness than common alternatives, while identifying physical implementation as the remaining challenge.
Problem
Arbitrary CGHs cannot generally be computed directly from a desired intensity profile, limiting high-quality arbitrary optical profiles for cold-atom applications.
Method
The paper realizes the mixed-region amplitude freedom (MRAF), an iterative Fourier-transform algorithm that controls intensity in a bounded two-dimensional focal-plane region.
Results
MRAF typically improves accuracy by one order of magnitude and roughness by two orders of magnitude over GS and AA for continuous target profiles.
Takeaways & Limitations
MRAF provides a relatively low-complexity, rapidly convergent route to smooth, arbitrarily shaped two-dimensional optical dipole traps with percent-level errors.
Takeaways & Limitations
Achieving MRAF’s high accuracy experimentally is challenging because SLMs can have quantized, finite-resolution, non-uniform, and nonlinear phase responses.
Abstract
from arXiv · showhide
We report the realization of a new iterative Fourier-transform algorithm for creating holograms that can diffract light into an arbitrary two-dimensional intensity profile. We show that the predicted intensity distributions are smooth with a fractional error from the target distribution at the percent level. We demonstrate that this new algorithm outperforms the most frequently used alternatives typically by one and two orders of magnitude in accuracy and roughness, respectively. The techniques described in this paper outline a path to creating arbitrary holographic atom traps in which the only remaining hurdle is physical implementation.
1. Introduction
Computer-generated holograms control focal-plane intensity without attenuating the input amplitude, but arbitrary profiles are difficult to compute directly. The MRAF algorithm addresses this challenge for two-dimensional atom-trap profiles, improving accuracy and roughness while remaining subject to focal-plane and physical-implementation constraints.
- Motivation: Computer-generated holograms use a pixelated phase raster to control focal-plane intensity while preserving the optical-field amplitude.The raster is implemented with an SLM or similar device and illuminated by a monochromatic beam.
- Motivation: Arbitrary CGHs are difficult to calculate directly, so IFTAs iteratively compare predicted and desired focal-plane intensities while modifying the kinoform.GS and AA are identified as the most frequently used IFTA variants.
- Contribution: The MRAF algorithm typically improves accuracy by one order of magnitude and roughness by two orders of magnitude over GS and AA for continuous target profiles.It controls intensity within a bounded two-dimensional subset of the focal plane and reaches percent-level accuracy, typically at a threefold efficiency cost.
- Scope: MRAF creates arbitrary two-dimensional optical traps, but confinement to the focal plane requires an additional tightly focused sheet of light.The method’s single-plane intensity control limits the trap geometry to two dimensions.
- Method setup: The computational input matrix must extend beyond the focusing optics’ clear aperture and include zero-intensity points to resolve the output plane fully.The physical size of the matrix representing the input plane is d.
2. Algorithm
The MRAF algorithm iteratively designs a CGH by propagating fields between input and focal planes, while controlling amplitude freedom and power distribution across signal and noise regions. Its initial phase strategy avoids optical vortices, and the method reaches few-percent intensity error within tens of iterations.
- IFTA operation: An IFTA designs a CGH that converts an input field into a target focal-plane intensity by exploiting unconstrained output-plane phase.The iterative procedure propagates fields with Fourier transforms, applies target-intensity constraints, and uses the backward-propagated phase for the next iteration.
- IFTA operation: Each iteration propagates the current field, combines it with the target profile using mixing parameters, and stops when the figure of merit no longer improves.The final input-plane phase becomes the kinoform transferred to the physical device.
- MRAF algorithm: MRAF introduces amplitude freedom only in a restricted noise region while preserving phase freedom throughout the output plane.A single mixing parameter controls power distribution between the signal region, which overlaps the atom-interaction area, and the noise region.
- MRAF algorithm: The signal-region constraint improves target matching, while amplitude freedom in the noise region reduces CGH efficiency and makes that region less controlled.The algorithm maintains constraints on target-profile power and total output power while concentrating accuracy where atoms interact with the light.
- Initial phase: A suitable initial phase is needed because optical vortices cannot be eliminated once present and can cause stagnation in the iterative design.The initialization seeks output power overlapping the target envelope while avoiding phase singularities and zero-intensity points.
- Initial phase: Quadratic phase combined with linear and conical gradients avoids vortices for MRAF and respectively adjusts field size, centroid position, and ring formation.Quadratic phase roughly matches the output-field envelope to the target; linear gradients address shifted targets, while conical gradients create rings.
- Performance: A few percent error in the predicted intensity profile is achieved within tens of iterations.The result follows from combining the propagated field with the target intensity during MRAF iterations.
3. Results
The MRAF algorithm was evaluated on six target intensity profiles designed for potential ultra-cold-atom applications and compared with GS and AA using accuracy and roughness measures. Across these tests, MRAF produced smooth profiles with few-percent average error, substantial pixel-level accuracy, and lower efficiency than the alternatives.
- Test profiles: Six target intensity profiles represented geometries proposed for ultra-cold-atom experiments, including multiply connected, star-shaped, square, and connected-reservoir patterns.The profiles were selected for potential applications such as multiply-connected gases and vortex generation in Bose-Einstein condensates.
- Evaluation metrics: Accuracy was measured as the r.m.s. fractional error from the target over a measure region normalized to equal power, while roughness quantified small-scale deviations.The measure region excludes zero-intensity target pixels; roughness is based on curvature of the difference between predicted and target profiles.
- MRAF results: MRAF converged in fewer than 100 iterations for every target profile after optimizing mixing and initial phase parameters.The optimized parameters included mixing values selected by minimizing the accuracy metric η over ranges of phase-profile and mixing parameters.
- MRAF results: A factor of 9 improvement in accuracy and a factor of 190 improvement in roughness were obtained on average versus GS and AA, with average errors at the few-percent level.The comparison used the measure regions defined for the six target profiles and found comparatively smooth MRAF outputs despite roughness not being optimized directly.
- MRAF results: For target (a), 95% of signal-region pixels had less than 3% error, whereas GS and AA each had at least 45% of pixels exceeding 10% error.This histogram-based comparison tests spatially distributed accuracy rather than only the average measure-region error.
- Trade-offs: MRAF efficiency was approximately 2–3 times lower than GS and AA, but a comparable holographic profile remained more efficient than an intensity-mask implementation.For one related profile, the reported transmission was approximately 29% for MRAF versus 3% for the intensity mask.
4. Conclusion
The MRAF algorithm produces smooth, percent-level-error holographic trap profiles and converges rapidly, but experimental implementation remains constrained by non-ideal CGH response and related technical limitations.
- Within tens of iterations, MRAF converges to comparatively smooth outputs with errors at the percent level for six target profiles.
- The mixing parameter minimizing error approximately coincides with a minimum in roughness for MRAF.
- Experimental realization assumes aberration-free optics, the paraxial approximation, single polarization, and ideal CGH response.
- Non-uniform and nonlinear phase response in some SLMs makes achieving MRAF’s percent-level accuracy experimentally challenging.