Highlights from H.E.S.S.

. In the 15 years since its construction, the H.E.S.S. gamma-ray observatory has allowed the study of the Very High Energy gamma-ray sky at resolutions and sensitivities which were never before possible. During this period H.E.S.S. has discovered a rich zoo of both galactic and extra galactic source classes, made measurements of the galactic cosmic ray spectrum and placed limits on fundamental physical processes. A summary of the latest H.E.S.S. results for these source classes has been presented at the conference, describing the most interesting new observations and their physical interpre-tation. Additionally the latest upgrades and improvements to the H.E.S.S. hardware and data analyses, and the future science prospects for the experiment have been described.


Introduction
The H.E.S.S. array, inaugurated 15 years ago, revolutionized our view on the Very High Energy (VHE) gamma-ray sky with more than 100 sources discovered. A second phase of the experiment started in 2012, with the installation of the fifth, largest telescope (28 m diameter), allowing the threshold to be lowered from ∼ 100 GeV down to ∼ 20 GeV and bridge the gap with space instruments such as Fermi-LAT. To date, H.E.S.S. is the only hybrid 1 in the field, thus paving the way for the next generation Cherenkov Telescope Array. In the last years, tremendous efforts went into the design and optimization of analysis software able to exploit simultaneously monoscopic & stereoscopic events in a consistent way. At the same time, the field of VHE γ-ray astronomy is experiencing a kind of phase-transition, where single discoveries tend to have less importance and where population studies and detailed analysis of sources of specific interest are opening new fields. In this context, a new paradigm of simulation (Section 2.3) is now emerging and opening new science capabilities, with unprecedented resolution and understanding of the instrument.
The last years have seen the long awaited emergence of multi-wavelength, multi-messenger astronomy, advocated for since decades, with the discovery of gravitational waves from a binary neutron star merger [1,2] and more recently, with the association of a very high energy neutrino observed by IceCube with an active galactic nuclei [3]. H.E.S.S., thanks to its very fast reaction time and slewing speed, is actively involved in this new time-domain astronomy.
After • the H.E.S.S. Galactic Plane survey [5], cornerstone of the publication and the major project of the collaboration over the last 10 years; • the first population studies, on Pulsar Wind Nebula (PWN) [6] and Supernova Remnants (SNR) [7]; • detailed analysis on several sources, and in particular first evidence of emission beyond the shell of a SNR [8]; • a comprehensive view of the Galactic Center region, including diffuse emission [9]; • upper limits on several specific objects and/or potential new source classes.

H.E.S.S. Galactic Plane Survey
The H.E.S.S. Galactic plane survey (HGPS) [5] is a major long-term project, corresponding to ∼ 2700 hours of high-quality observations acquired with the H.E.S.S.-I telescope array from 2004 to 2013.
Results have already been published on a small (∼ 10%) fraction of the current data set [10,11].  [12] and the VERITAS Cygnus survey [13,14] regions are illustrated in blue and green, respectively. From [5]. The region of the Milky Way covered by the HGPS (Galactic longitude between −110 and 65 degrees and Galactic latitude |b| < 3.5 degrees) is depicted as a white rectangle in Fig. 1 and compared to the HEGRA (in blue) and VERITAS Cygnus (in green) surveys.
A catalogue of 77 cosmic accelerators was derived from the HGPS using a semi-automatic pipeline, out of which 6 are new sources that were previously unknown or unpublished. Less than half of the sources (31) are firmly identified, the largest population consisting of pulsar wind nebulae followed by supernova remnants and binary systems. Most of the remaining sources are confused (having several possible counterparts) while a significant fraction (12 sources) have no known counterparts at other wavelengths.
Maps and tables from the HPGS were released, for the first time, as FITS file to allow better exploitation by the community 2 .

Population Studies
For the first time, the existence of a large scale catalogue of TeV sources made population studies on the dominant source classes, PWN [6] and SNRs [7] possible. One major result is that most young and powerful pulsars have a PWN detected in VHE. H.E.S.S. data also exhibit a correlation of TeV surface brightness with pulsar spin-down powerĖ, likely caused by an increase of extension with decreasingĖ, and hence with time, and also by a mild decrease of TeV gamma-ray luminosity with decreasingĖ. This allows a PWN evolution model to be constructed for the first time, from which it is possible to estimate how many of these objects will be visible by CTA. Regarding SNRs, typical ambient density values around shell-type SNRs can be constrained to n ≤ 7 cm −3 and electron-toproton energy fractions above 10 TeV to ep ≤ 5 × 10 −3 . Comparisons of VHE with radio luminosities also provide support for the theory of magnetic field amplification at shell-type SNRs.

Resolving sources down to arcmin scale
The resolution power of current Imaging Atmospheric Cherenkov Telescopes is presently restricted to scales of a few arc-minutes, and is limited first by statistics and second by the uncertainties in the characterisation of the instrument Point Spread Function (PSF). A new paradigm for the simulation of IACTs has been developed during the last years [15], aiming at overcoming the limitations of current approaches, and to more efficiently cover the simulation phase space. Indeed with the advent of hybrid systems consisting of a large number of IACTs with different sizes, types, and camera configurations, the complexity of the interpolation and the size of the phase space becomes increasingly prohibitive.
In the Run-Wise simulation (RWS) paradigm, in contrast with traditional approach using a fixed grid of simulation parameters, the actual observation conditions as well as the individual telescope configurations of each observation run of a given data set, are taken into account. These run-wise simulations exhibit considerably reduced systematic uncertainties compared to the existing approach. They are also more computationally efficient and simple and therefore opens new physics cases. In particular, the PSF is much better understood and reproduced with the RWS simulation, allowing for the first time the study of the extension of sources down to arc-minute scale, well below the actual PSF size.
Using the RWS simulation scheme,the extension of the Crab nebula at TeV gamma-ray energies was resolved for the first time [16], with a width of 52 assuming a Gaussian source shape (Fig. 2).  [19]. The detection of VHE emission from FSRQs during their outbursting episodes was unexpected due to the presence of the broad line region (BLR), an intense radiation field in the vicinity of the AGN, which acts as a barrier to the escape of VHE emission from within it. The detection of VHE emission from these sources can therefore be used to place strong constraints on the position of the emission site with respect to the BLR location. With gamma-ray emission beyond 200 GeV detected from each of these systems during their outbursts, the emission site is found to be constrained to sit at a distance beyond r BLR ≈ 10 17 cm L disk /10 45 ergs −1 1/2 , where = L disk is the thermal luminosity of the accretion disk [20].

Multi-Wavelength, Multi-Messenger astronomy
Due to its very large effective area and low threshold, H.E.S.S.-II has a competitive advantage over space instruments such as Fermi-LAT for transient phenomena in the overlapping energy range. As speed of reaction is critical for GRBs and other transient phenomena, a rapid repointing system for the telescopes has been implemented [21], which has been designed to respond as fast as possible, without human intervention, to targets of opportunity (ToOs). The average overall response time has been reduced to a timescale of ∼few minutes, half of the sky being reachable in less than 60 s. An example of the recent improvement in response time to neutrino alerts is the follow-up observations of the ANTARES neutrino event on the 30th January 2017. These observations took placed only 32 seconds after the reconstructed neutrino event occurred. Although searches for a gamma-ray counterpart within this data set were unsuccessful, the success of the automatic response system has already proved itself through this considerable reduction in the follow-up delay time to neutrino event alerts. The same applies to gravitational waves, where automatic scheduling is done for events occurring during the night. In the case of the neutron star merger GW170817, the H.E.S.S. telescopes were on target 5.3 h after the neutron star merger [1,2], but less than 5 min after the revised LIGO-VIRGO alert was distributed and a handful of seconds after the start of the astronomical night. These 5 min latency was used to implement an observation strategy based on the probability density region obtained by weighting the three-dimensional GW map with galaxies from the GLADE catalog.

Electron spectrum
The spectrum of high energy electrons, up to ∼ 1 TeV is traditionally measured using space based multi-purpose instruments such as AMS [22] or Fermi-LAT [23], due to much better hadronic rejection than IACTs. Dedicated space based instruments such as CALET or DAMPE are planning to extend the measurement of the electron spectrum up to ∼ 10 TeV. In contrast, ground-based Cherenkov telescopes, thanks to their very large effective areas, are able to access the highest energies where the flux becomes very low. More than 9 years after the first electron spectrum measurement with H.E.S.S. [24] up to 5 TeV, subsequent observations have increased fourfold the amount of available data. In addition, analysis techniques have improved significantly, leading to a much better suppression of the background of cosmic-ray nuclei. These improvements allow a measurement of cosmic-ray electrons up to energies of ∼ 20 TeV, presented for the first time at the 35 th ICRC, Busan, South Korea (Fig. 3). The measured electron spectrum can be described by two different regimes: up to ∼ 1 TeV, the spectrum appears quite regular with a constant slope, and becomes steeper above this energy. This break in the spectral slope most probably signals the transition between a regime where a large number of sources contribute to the spectrum, to a regime where only the few, closest sources to the Earth are able to contribute. The very high energies reached in this measurement allow models of nearby sources of cosmic-ray electrons in which one source is very prominent to be tested. Such models are often invoked as a possible explanation for the excess of positrons measured by some experiments such as Pamela and AMS.