Fringe tracking performance monitoring: FINITO at VLTI

TL;DR

This paper evaluates the performance of fringe tracking at VLTI using real-time data, proposing metrics to optimize exposure times and correlating performance with atmospheric conditions, including initial neural network modeling efforts.

Contribution

It introduces new performance metrics for fringe tracking, correlates them with environmental data, and explores neural network models for performance prediction.

Findings

Optimal exposure times depend on seeing and coherence time.

Fringe tracking performance correlates with atmospheric conditions.

Preliminary neural network models show promise in predicting performance.

Abstract

Since April 2011, realtime fringe tracking data are recorded simultaneously with data from the VLTI/AMBER interferometric beam combiner. Not only this offers possibilities to post-process AMBER reduced data to obtain more accurate interferometric quantities, it also allows to estimate the performance of the fringe tracking a function of the conditions of seeing, coherence time, flux, etc. First we propose to define fringe tracking performance metrics in the AMBER context, in particular as a function of AMBER's integration time. The main idea is to determine the optimal exposure time for AMBER: short exposures are dominated by readout noise and fringes in long exposures are completely smeared out. Then we present this performance metrics correlated with Paranal local ASM (Ambient Site Monitor) measurements, such as seeing, coherence time or wind speed for example. Finally, we also present some preliminary results of attempts to model and predict fringe tracking performances, using Artificial Neural Networks.