Consequences of non-Gaussian instrumental noise in perturbed nulling interferometers

Felix A. Dannert, Philipp A. Huber, Thomas Birbacher, Romain Laugier, Markus J. Bonse, Emily O. Garvin, Adrian M. Glauser, Veronika Oehl, Sascha P. Quanz · 2025

Read Consequences of non-Gaussian instrumental noise in perturbed nulling interferometers by Felix A. Dannert, Philipp A. Huber, Thomas Birbacher, Romain Laugier, Markus J. Bonse, Emily O. Garvin, Adrian M. Glauser, Veronika Oehl, Sascha P. Quanz (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-20653.txt.

Introduction

Nulling interferometers in space could image Earth-like planets directly in the mid-infrared, but the instruments wobble, and that wobble adds systematic noise. The nine authors revisit models of that noise and confirm by simulation that the noise in the differential output is not Gaussian. Standard detection tests assume Gaussian noise, so they overstate confidence in detection thresholds.

For the first time, the team derives the true noise distribution for a dual Bracewell nulling interferometer: it follows repeated convolutions of Bessel functions. With the right distribution they build a sharper hypothesis test and a semi-analytical prediction of detection performance. The efficient model is public code, meant to feed science yield estimates and mission design for concepts such as the Large Interferometer For Exoplanets.

Catalog record

CREATORFelix A. Dannert, Philipp A. Huber, Thomas Birbacher, Romain Laugier, Markus J. Bonse, Emily O. Garvin, Adrian M. Glauser, Veronika Oehl, Sascha P. Quanz
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE134 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2506-20653

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Consequences of non-Gaussian instrumental noise in perturbed nulling interferometers Felix A. Dannert, Philipp A. Huber, Thomas Birbacher, Romain Laugier, Markus J. Bonse, Emily O. Garvin, Adrian M. Glauser, Veronika Oehl, Sascha P. Quanz Abstract With the astrophysics community working towards the first observations and characterizations of Earth-like exoplanets, interest in space-based nulling interferometry has been renewed. This technique promises unique scientific and technical advantages by enabling direct mid-infrared observations. However, concept studies of nulling interferometers often overlook the impact of systematic noise caused by instrument perturbations. Earlier research introduced analytical and numerical models to address instrumental noise and, building on these results, we reproduce key simulations and report that the noise in the differential output of nulling interferometers follows a non-Gaussian distribution. The presence of non-Gaussian noise challenges the validity of classical hypothesis tests in detection performance estimates, as their reliance on Gaussian assumptions leads to overconfidence in detection thresholds. For the first time, we derive the true noise distribution of the differential output of a dual Bracewell nulling interferometer, demonstrating that it follows iterative convolutions of Bessel functions. Understanding this noise distribution enables a refined formulation of hypothesis testing in nulling interferometry, leading to a semi-analytical prediction of detection performance. This computationally efficient instrument model, implemented in a publicly available codebase, is designed for integration into science yield predictions for nulling interferometry mission concepts. It will play a key role in refining key mission parameters for the Large Interferometer For Exoplanets (LIFE). I Introduction Since its conception by Bracewell 1978 , the exoplanet community has made great progress in demonstrating space-based mid-infrared nulling interferometry as a viable technique to detect and characterize Earth-like exoplanets ( Lawson et al. 2007 ; Cockell et al. 2008 ; Quanz et al. 2021a ) . Their scientific potential is primarily demonstrated through simulations and widely accepted performance metrics like sensitivity and contrast ( Kammerer & Quanz 2017 ; Dandumont et al. 2020 ; Quanz et al. 2022 ; Konrad et al. 2022 ; Kammerer et al. 2022 ; Loicq et al. 2024 , e.g.) . However, this approach is in tension with the fact that high-contrast observations are highly susceptible to instrument perturbations ( Lay 2004 ; Douglas et al. 2018 ; Nemati et al. 2020 ; Nemati et al. 2023 ) . In nulling interferometry, vibrations and thermal expansions of the instrument can change the phase response by introducing additional optical path, while pointing and high-order wavefront errors can influence the amplitude response (see Table 1 in Lay 2006 for an overview and Martin et al. 2003 ; Martin & Booth 2010 ; Martin et al. 2012 ; Gheorghe et al. 2020 ; Ranganathan et al. 2024 ). Such perturbations degrade the interferometric null and introduce systematic noise that can be mistaken for an exoplanet detection, an effect often not explicitly accounted for by the standard performance metrics – such as sensitivity and contrast – that underpin most simulation-based studies of nulling interferometry. Most scientific performance evaluations fail to provide estimates by how much the described measurements could be impacted by these effects. This is partially remedied by important initial estimates for the consequences of systematic noise made by Serabyn 2000 and Lay 2004 in the context of the proposed nulling interferometers of the early 2000s. They address the much-discussed issue of systematic noise by assuming that phase, amplitude and polarization of the incident light are perturbed following a specific statistical distribution. Subsequently, they employ an analytical method for predicting the effect of perturbations on exoplanet detection limits. To this day, their methods are fundamental in the design of nulling interferometers ( Defrère et al. 2010 ; Dannert et al. 2022 ) . But, crucially, Serabyn 2000 and Lay 2004 did not consider how the instrument transforms these input perturbations and gives rise to distinct statistical noise distributions in the output of the nulling measurement. The importance of knowing this noise distribution is underlined in studies of single-aperture high-contrast imaging applications. Here, the distribution of speckles in images is consistently reported to be non-Gaussian ( Soummer & Aime 2004 ; Soummer et al. 2007 ; Pairet et al. 2019 ) , with bright speckles being overabundant compared to Gaussian noise. Only in recent years, the community started reporting on how fundamentally this can affect the science performance of the instrument, particularly by increasing the probability of false positive detections ( Bonse et al. 2023 ) . First concrete evidence of nulling interferometers suffering from similar limitations was delivered by Mennesson et al. 2010 and Hanot et al. 2011 using the Palomar fiber nuller. Indeed, they report a non-Gaussian distribution of the noise, but more importantly, they demonstrate how knowledge of this distribution can be used to increase instrument performance. Martinache & Ireland 2018 later predicted that certain interferometer setups called Kernel nullers are fundamentally self-calibrating and exhibit improved noise behavior. The consideration of systematic instrumental noise, including its true distribution, has fundamental implications for the design of future nulling interferometers, the prediction of their performance, the derivation of mission requirements and the conceptualization of novel science cases that are enabled by nulling. Hence, systematic noise should be considered early in the design process of the emerging Large Interferometer for Exoplanets (LIFE) . First steps in this direction are taken by Dannert et al. 2022 ; Laugier et al. 2023 ; Huber et al. 2024 and Huber et al. (2025, in preparation). This way, even if fundamental design changes become necessary, they can be implemented at comparably low cost. If systematic effects are identified late during production or even operation, mitigating their effects can become prohibitively expensive. Additionally, instrument design can benefit from an early analysis of systematic noise. Instead of budgeting for unknown instrumental effects, the exact sources of error can be pinpointed. Often, one can profit from this identification of the dominant noise source by relaxing the constraints on other sources of noise ( Dannert et al. 2022 ) . By outlining the method for simulating a perturbed nulling interferometer in Section II , the aim is to help scientists to assess the impact of systematic noise on their measurements. Section III formalizes the detection of exoplanet signals in nulling data through hypothesis testing on matched filters. Analyzing the impact of the filters in temporal frequency space, a fully analytic solution of the test statistic is derived. This solution can be seamlessly included in existing exoplanet detection yield models ( Dannert et al. 2022 ; Savransky & Garrett 2015 ; Morgan et al. 2023 ; Morgan et al. 2024 ; Stark et al. 2024 , see, e.g.,) . Section IV argues that systematic noise produced by nulling interferometers cannot be Gaussian-distributed. A semi-analytical method for deriving the true distribution of the noise is presented. This allows for the construction of a lookup table, enabling the correct interpretation of the non-Gaussian noise in hypothesis testing. Implications of this method for the LIFE project and beyond are discussed in Sections V and VI . All of the above steps are distilled into the publicly available Python package InLIFEsim 1 1 1 github.com/fdannert/InLIFEsim ( Dannert 2025a ) , built to support scientists and instrument designers in incorporating instrumental noise into their simulations. While this

Read the complete text