Finding the Dark Hole with the Lights On: A New Approach to Focal Plane Wavefront Sensing

TL;DR

This paper introduces a novel focal plane wavefront sensing method that improves dark hole creation in exoplanet imaging by overcoming shot-noise limitations and enabling more accurate DM configurations.

Contribution

It proposes regression procedures for FPWS that do not depend on small DM steps, using new representations (EGF and EAE) for better control and estimation.

Findings

Enables FPWS observations with DMs far from optimal configurations.

Improves accuracy of DM configuration predictions.

Simultaneously estimates planetary images with new methods.

Abstract

In direct imaging of exoplanets from space, achieving the required dynamic range (i.e., planet-to-star contrast in brightness) currently relies on coronagraphic technology combined with active control of one or more deformable mirrors (DMs) to create a dark region in the image plane, sometimes called a "dark hole." While many algorithms have been proposed for this purpose, all of them employ focal plane wavefront sensing (FPWS) in order to calculate the optimal DM configuration to create the desired dark hole. All current algorithms are limited by their own success in that, as the dark hole is achieved, the FPWS procedure becomes shot-noise limited due to he low intensity in the dark hole. This article proposes a FPWS procedure that allows determination of the optimal DM configuration without relying on information obtained when the DM is near the optimal configuration. This article gives regression procedures for FPWS that do not assume the DM step size is small, which should allow two important improvements to the control loop: 1) performing informative FPWS observations with DM configurations that are sufficiently distant from the optimal dark hole configuration to mitigate shot-noise limitations, and 2) more accurately predicting the DM configuration that will achieve the desired objective in the dark hole control loop. In order to treat this more challenging FPWS problem, two different representations are presented. The first of these, is called the empirical Green's function (EGF), is easy to implement, and has a block-diagonal matrix structure that is well-suited to parallel processing. The other representation, based on an explicit aberration expansion (EAE) requires the regression to estimate a smaller number of parameters than the EGF, but leads to a dense matrix structure. The EGF and EAE methods both simultaneously estimate the planetary image.