Parallelization of the SIR code for the investigation of small-scale features in the solar photosphere

TL;DR

This paper presents a parallelized version of the SIR code, significantly improving its speed and usability for inverting large high-resolution spectropolarimetric data sets of the solar photosphere.

Contribution

The authors developed and standardized a parallel version of the SIR code, enabling faster analysis of extensive solar data sets compared to previous versions.

Findings

Substantial speedup achieved with parallel SIR code

Enhanced usability and standardized input/output formats

Feasibility of inverting large high-resolution data sets efficiently

Abstract

Magnetic fields are one of the most important drivers of the highly dynamic processes that occur in the lower solar atmosphere. They span a broad range of sizes, from large- and intermediate-scale structures such as sunspots, pores and magnetic knots, down to the smallest magnetic elements observable with current telescopes. On small scales, magnetic flux tubes are often visible as Magnetic Bright Points (MBPs). Apart from simple $V/I$ magnetograms, the most common method to deduce their magnetic properties is the inversion of spectropolarimetric data. Here we employ the SIR code for that purpose. SIR is a well-established tool that can derive not only the magnetic field vector and other atmospheric parameters (e.g., temperature, line-of-sight velocity), but also their stratifications with height, effectively producing 3-dimensional models of the lower solar atmosphere. In order to enhance the runtime performance and the usability of SIR we parallelized the existing code and standardized the input and output formats. This and other improvements make it feasible to invert extensive high-resolution data sets within a reasonable amount of computing time. An evaluation of the speedup of the parallel SIR code shows a substantial improvement in runtime.