# Identifying the 3FHL catalog: I. Archival Swift Observations and Source Classification

## TL;DR

This study identifies and classifies previously unassociated sources in the 3FHL gamma-ray catalog using multi-wavelength data and machine learning, revealing most are likely BL Lac objects.

## Contribution

It combines archival Swift-XRT data, multi-wavelength catalog associations, and machine learning to classify unassociated gamma-ray sources for the first time.

## Key findings

- Majority of sources are likely BL Lac objects.
- Redshift range from 0.277 to 2.1.
- Machine learning suggests most are BL Lacs.

## Abstract

We present the results of an identification campaign of unassociated sources from the Fermi Large Area Telescope 3FHL catalog. Out of 200 unidentified sources, we selected 110 sources for which archival Swift-XRT observations were available, 52 of which were found to have exactly one X-ray counterpart within the 3FHL 95\% positional uncertainty. In this work, we report the X-ray, optical, IR, and radio properties of these 52 sources using positional associations with objects in various catalogs. The Wide-field Infrared Survey Explorer color-color plot for sources suggests that most of these belong to the blazar class family. The redshift measurements for these objects range from z=0.277 to z=2.1. Additionally, under the assumption that the majority of these sources are blazars, three machine-learning algorithms are employed to classify the sample into flat spectrum radio quasars or BL Lacertae objects. These suggest that the majority of the previously unassociated sources are BL Lac objects, in agreement with the fact the BL Lac objects represent by far the most numerous population detected above 10 GeV in 3FHL.