Hades experiment probing baryonic matter at SIS18 overview of results

. HADES experiment at GSI is the only high precision experiment probing nuclear matter in the beam energy range of a few AGeV. Pion, proton and ion beams are used to study rare dielectron and strangeness probes to diagnose properties of strongly interacting matter in this energy regime. Selected results from p + A and A + A collisions are presented and discussed.


Introduction
In the recent years a significant grow of the interest in nuclear collisions at lower energies is observed.Dedicated experimental programs of Beam Energy Scan at RHIC and NA61-SHINE at CERN are under way to search for the onset of deconfinement and for a critical point in the phase diagram of nuclear matter.New experimental facilities FAIR at Darmstadt [1] and NICA in Dubna [2] are under construction to scan the lower energy regime.This enormous activity is motivated by the very challenging but fundamental question of the understanding of structure and phases of strongly interacting matter.At high temperature and zero µ b lattice QCD, has provided strong evidence that the transition from a hadron gas to a partonic phase is a smooth crossover for physical quark masses.One expects that this crossover continues at non-zero µ b and eventually ends in the critical point.Furthermore, it has been shown that at vanishing µ b the phase transition is associated by another fundamental phenomenon, the restoration of chiral symmetry which is spontaneously broken in vacuum.It has also been suggested that beyond the critical point the restoration of chiral symmetry could fall apart from the transition to deconfined matter, giving rise to a confined phase with partially restored chiral symmetry.This "Quarkyonic" matter appears in the QCD limit of a large number of flavours and can be understood as a gas of confined quarks with a rich excitation spectrum.At very high µ b and low temperatures condensation of quark pairs can take place and lead to formation of colour super-conductive phase.This gives rise to a rich phase structure in that region, depending on the various flavour-colour symmetry structures, the so-called Colour-Flavour-Locking scenarios (for more details see [3], [4]).Such phase may also exist inside the neutron stars and hence are of interest also for astrophysics.Although, in this range of T ,µ b lattice QCD calculations are so far impossible, approaches based on QCD-inspired effective models are in use and provides some guidance.In this context experimental input is of largest importance to provide some constraints and to drive developments in this field.
HADES at SIS18 at GSI Darmsatdt is currently the only experiment studying properties of strongly interacting matter in a few AGeV energy regime.The strategy of the experiment is to use rare and penetrating probes like dielectrons and hadrons containing strangeness to diagnose the phase diagram at high µ b and moderate temperatures.Using the variety of proton, deuteron, secondary pion and ion beams HADES experiments are ideally suited for systematic studies.Using heavy ion reactions a dense (up to 3ρ 0 ) and hot (with temperatures up to T=80 MeV) fireball with relatively long life time (∼ 10 fm/c) can be created.Experiments with pion and proton beams also allow for the study of cold-matter effects.Further experiments at SIS100 at FAIR also planned and will cover, together with the new CBM detector the 8 − 10 AGeV energy range and bridge to the energy domain of BES at RHIC and SPS.

Experimental programme of HADES
The HADES programme realized so far can be divided into three reaction classes.Experiments studying dielectron, pion and baryon resonance productions in proton-proton (at 1.25, 2.2 and 3.5 GeV), d + p (at 1.25 AGeV) and recently π − p reactions provided important constraints on contributions of various e + e − sources and allowed to establish model independent reference spectra for studies of proton-nucleus and nucleus-nucleus collisions.Studies of p + p collisions at 3.5 GeV provided also valuable new data on hyperon, Σ(1385) and Λ(1405) production.In particular measurements of the mass distribution of Λ(1405) shows differences as compared to the one obtained from photon and pion induced reactions and calls for explanation particularly in the context of the on-going discussion about possible molecular nature of this resonance.Using Partial Wave Analysis excitation of baryon resonances via one pion and pK + Λ(1192) final states have been determined for proton-proton collisions (for recent review and references see [5]).These studies provide also an important input to transport model calculations where the pion and kaon production in nuclear matter is strongly influenced by couplings of the ∆, N * resonances.
The dielectron, neutral kaon and hyperon productions were investigated in p + Nb collisions at 3.5 GeV.Small C + C, medium size Ar + KCl and recently Au + Au collision systems were explored in the 1 − 2 AGeV energy range to study emissivity of nuclear matter and strangeness production (φ, K −,+,0 , Λ, Ξ − (1321)).Below we present highlights of the results obtained in p + A and A + A collision systems 3 Results from p+A collisions p + Nb collisions at a beam energy of 3.5 GeV have been studied by HADES with the main goal of searching for in-medium modification of vector mesons in cold nuclear matter [6].The large acceptance of the detector and the low energy of the beam allow for a detection of e + e − pairs down to low momenta (p e + e − < 1.0 GeV/c) which was not possible in the similar experiments performed by CLAS/JLAB (with photon beams [7,8]) and E325/KEK [9] collaborations.Differential e + e − production cross sections as a function of the e + e − invariant mass (shown in Fig. 1), the momentum and the rapidity have been measured and compared to those obtained from p + p.The direct comparison of the measured distributions to the yields expected from the known hadronic sources (from a PYTHIA calculation) is shown for p + p collisions in Fig. 1 [10].It reveals an unexplained strength below the vector meson pole which becomes even more pronounced in the proton-nucleus collisions.Such additional strength can be explained by a strong coupling of the ρ meson to low-mass baryonic resonances, lating known pion production cross sections [23].The corresponding total reaction cross section amounts to σ pN b = 848 ± 127 [mb].A detailed description of the procedure is given in [24].In addition, the trigger efficiency of the first-level trigger, asking for a charged particle multiplicity larger than three, on inclusive e + e − pair production F (e + e − )= 0.92 was extracted from simulations and has been taken into account for the normalization of the dielectron distributions of the p+Nb run.In p+p collisions the normalization was obtained via the exclusive measurement of elastic p+p collisions and the known integrated cross section inside the HADES acceptance [21].The efficiency corrected invariant mass distributions of e + e − pairs are shown in Fig. 2 for both collision systems.The colored horizontal bars represent the systematic uncertainties, which result from the quadratic sum of errors estimated from the different particle identification methods (10%), from consistency checks of the efficiency correction (10%), including the uncertainty due to combinatorial background subtraction as well as the uncertainty from the normalization (15%).The total systematic error amounts to 21% in case of the p+Nb data while for the p+p data the systematic uncertainty is 20%.For the comparison of the spectral shape of the invariant mass distributions only the systematic errors of the normalization are taken into account as the other sys- tematic errors are assumed to cancel to first order.For the p+p data a dielectron cocktail was generated using an adapted version of the event generator PYTHIA, see [21] for details.There are four distinct mass regions: M ee [GeV/c 2 ] < 0.15 (dominated by neutral pion decays), 0.15 < M ee [GeV/c 2 ] < 0.47 (η Dalitz decay dominated), 0.47 < M ee [GeV/c 2 ] < 0.7 (dominated by direct ρ decays and Dalitz decays of baryonic resonances and ω mesons) and 0.7 < M ee [GeV/c 2 ] (ρ and ω dominated) as can be seen from the cocktail.Moreover, around 1 GeV/c 2 a low statistics φ signal is visible, which will be discussed in a future publication making use of additional information from its hadronic decay channel.The underestimation of the dielectron yield in the mass region from 0.47 < M ee [GeV /c 2 ] < 0.7 points to an insufficient description of the coupling between ρ mesons and baryonic resonances.This coupling will enhance the e + e − yield mainly below the ρ pole mass due to kinematical constraints given by the mass distribution of the resonances as well as the ones of the vector mesons [25,26].Following vector meson dominance, the coupling of the vector mesons to baryonic resonances is related to the electromagnetic structure of the corresponding baryonic transitions.There is then no distinction between the direct Dalitz decay of baryonic resonances (N * → N γ * ) and the intermediate coupling to the rho meson decay (N * → N ρ → N γ * ) and we will refer to them in the following as "ρ-like contribution".
In order to compare the spectral shapes, the p+p data are scaled up according to the nuclear modification factor R pA , defined as: While A pp part = 2, a value of A pN b part = 2.8 is estimated with the help of a geometrical nuclear overlap model [27].We use σ pp reaction = 43.4mb from [28].This choice of scaling lating known pion production cross sections [23].The corresponding total reaction cross section amounts to σ pNb = 848 ± 127 [mb].A detailed description of the procedure is given in [24].In addition, the trigger efficiency of the first-level trigger, asking for a charged particle multiplicity larger than three, on inclusive e + e − pair production F(e + e − )= 0.92 was extracted from simulations and has been taken into account for the normalization of the dielectron distributions of the p+Nb run.In p+p collisions the normalization was obtained via the exclusive measurement of elastic p+p collisions and the known integrated cross section inside the HADES accep-FIG.3: Left: Comparison of the invariant mass spectra for e + e − pairs with P ee > 0.8 GeV/c from p+p and p+Nb.The inset shows a linear zoom into the vector meson region together with a fit to the ω structure for the p+Nb data.Right: For pairs with P ee < 0.8 GeV/c.The p+p data have been scaled according to a Glauber model.tematic errors are assumed to cancel to first order.For the p+p data a dielectron cocktail was generated using an adapted version of the event generator PYTHIA, see [21] for details.There are four distinct mass regions: M ee [GeV/c 2 ] < 0.15 (dominated by neutral pion decays), 0.15 < M ee [GeV/c 2 ] < 0.47 (η Dalitz decay dominated), 0.47 < M ee [GeV/c 2 ] < 0.7 (dominated by direct ρ decays and Dalitz decays of baryonic resonances and ω mesons) and 0.7 < M ee [GeV/c 2 ] (ρ and ω dominated) as can be seen from the cocktail.Moreover, around 1 GeV/c 2 a low statistics φ signal is visible, which will be discussed in a future publication making use of additional information from its hadronic decay channel.The underestimation of the dielectron yield in the mass region from 0.47 < M ee [GeV/c 2 ] < 0.7 points to an insufficient description of the coupling between ρ mesons and baryonic resonances.This coupling will enhance the e + e − yield mainly below the ρ pole mass due to kinematical constraints given by the mass distribution of the resonances as well as the ones of the vector mesons [25,26].Following vector meson dominance, the coupling of the vector mesons to baryonic resonances is related to the electromagnetic structure of the corresponding baryonic The inset shows a linear zoom into the vector meson region together with a fit to the ω peak for the p+Nb data.Right: for pairs with p e + e − <0.8 GeV/c.The p + p data have been scaled as described in the text [6] which was not included in PYTHIA.Modification of the invariant mass spectrum are expected due to electromagnetic transition form factors of R → Ne + e − decays which according to Vector Meson Dominance (VDM) proceed via intermediate vector mesons.Indeed, calculations [11,12] including the form factors of ∆ → Ne + e − obtained from the two component model [13], and contributions from higher mass resonances N * → ρN → e + e − N describe the missing yield.However, this result has to be taken with caution since the exact production cross sections of the resonances and their decay branches into ρN are subject to large uncertainties.It is expected that the recently collected data on dielectron and pion production in π − p reaction channels and the planned future campaigns with the pion beam (see contribution of F.Scozzi to this conference) will shed more light on this important aspect.
Coming back to the p + Nb data, we show in Fig. 2 the comparison of the e + e − invariant mass distribution to the one measured in p + p reactions for two momentum bins of the outgoing dielectron pair.Here, the p + p cross sections have been scaled up by the ratio of the total cross sections for both reactions and the averaged numbers of participants calculated with a Glauber model: With such scaling π 0 production measured in the p + p describes (see Fig. 2) pion Dalitz yield in p + Nb.On the other hand, an increase of the e + e − yield below the vector meson pole above the p + p reference is visible for low momenta p e + e − < 0.8 GeV/c (see also Fig. 3, left panel).
In order to better quantify this excess we subtract first, the ω peak in both data samples and further subtract the scaled p + p dielectron yield from the p + Nb yield.The difference, shown in Fig. 3-right panel, represents the additional e + e − radiation excess due to the medium.It shows an exponential decrease with an additional enhancement directly below the vector meson pole mass, i.e. between 0.6-0.7.Note that this enhancement is exactly at the position where a discrepancy is observed when comparing the p + p data with the PYTHIA calculation (Fig. 1), indicating that both observations is justified by the agreement of the scaled p+p data with the p+Nb results in the invariant mass region below 150 MeV/c 2 (see Fig. 5 below) and the calculations in [26].A unique feature of the HADES setup is its coverage for low momentum pairs.This allows for the first time to compare the invariant mass distributions for e + e − pairs with momenta, down to 0.2 GeV/c and larger than 0.8 GeV/c.The respective contributions are shown in Fig. 3; for pairs with P ee > 0.8 GeV/c (left panel) the dielectron yield from p+Nb is slightly lower compared to the scaled p+p data, pointing to absorption of produced mesons inside the nucleus and subsequent particle production in secondary reactions.These second generation particles have then on average smaller momenta and therefore contribute more to the low momentum dielectron sample.The shape of the spectrum is identical to the reference p+p data within errors.Moreover, the width of the ω peak can be deduced by fitting a Gaussian function to the peak, assuming a smooth background underneath.The corresponding fit, together with a linear zoom into the vector meson region for the p+Nb data, is displayed  The excess observed in the case of p + Nb might be interpreted as a fingerprint of the contribution of the multi-step processes processes of the type p + p → πX, πN → R → Ne + e − adding also to this mass region because of the aforementioned strong resonance-ρ couplings.On the other hand, model calculations based on hadronic many-body interactions also predict that such couplings strongly modify the in-medium ρ meson spectral function [14].Therefore, final conclusions about in medium modifications of the ρ meson in cold nuclear matter can be derived only if the ρ-baryon resonance couplings are fully understood.In this context the data from pion induced reactions will play a decisive role.
As the ω meson is concerned, we observe that for slow pairs the yield at the ω pole is not reduced, however, the underlying smooth distribution is enhanced.Thus, the yield in the peak is almost zero within errors.This indicates a strong ω absorption in contrast to the pairs from the underlying continuum.Assuming that the ω cross section scales with the mass number as σ pNb = σ pp × A α we obtain α =0.38± 0.29 for slow pairs and α=0.67 ± 0.11 for p e + e − > 0.8 GeV/c.Furthermore, an analysis of the ω width shows, within the error bars, no significant broadening.Both observations are in line with the results of the CBELSA-TAPS experiment [15], although one should note that in contrast to the p + A reactions for the photon induced reactions no initial state effects and consequently a stronger scaling could be expected.
The large statistics collected for the p+ Nb system allowed also for studies of the hyperon-nucleon interactions via two-particle correlation function [16].The Λ − p interaction has become more and more crucial in recent years due to its connection to the modelling of astrophysical objects like neutron stars.It appears that in the inner core of the star the existence of hyperons is possible since their creation is often energetically favoured in comparison with a purely nucleonic matter composition.However, the appearance of these additional degrees of freedom leads to a softening of the nuclear

VI. SUMMARY
To summarize, we presented the hitherto first measurement of the pΛ correlation function in pA reactions.
able to provide data which can be investigated with a theoretical framework with the necessary sensitivity to study carefully final state interactions if the size of the particle emitting region is known beforehand.The femtoscopy technique to study interactions between particles can be applied to many colliding systems at very different energies, which can help to improve the understanding of hyperon-nucleon interactions.With the planned update of the HADES setup including a electromagnetic calorimeter the measurement of the pΣ 0 correlation function is accessible and it is a planned analysis in the HADES strangeness program.to the LO (green) and NLO (red) scattering parameter set (see ref [17] for details.The error bands in the theory curves correspond to the errors of the Λ − p source size determination [16] imum value of the Λ decay vertex distance to the primary vertex (Λ-VerToPrimVer).Here, the off-vertex cut iv) is the main condition responsible for the extraction of a Λ signal.Starting with the moderate conditions as used in the previous high-statistics analysis of the Λ phase-space distribution and polarization [31], a clear Λ signal could be separated from the combinatorial background in the p-π − invariant-mass distribution.While in that analysis a signal-to-background ratio in the order of unity was sufficient, for the present Ξ − search we start with a higher Λ purity (>85 %, cp.[12]).Hence, with the stronger cuts and the requirement of an additional π − meson, the number of reconstructed Λ hyperons decreases from about 1.1 million to 300,000.(No event containing clearly more than one Λ was found.)Taking this still high-statistics Λ sample, we started the Ξ − investigation by combining -for each event containing a Λ candidate (selected by a ±2σ window around the Λ peak) -the Λ with those π − mesons not already contributing to the Λ.The result was a structureless Λ-π − invariant mass distribution.Hence, additional conditions were necessary: v) a lower limit on the 2nd π − (potential Ξ − daughter) track distance to the primary vertex (π 2 -VecToPrimVer), vi) an upper limit of the distance of the Ξ − pointing vector w.r.t. the primary vertex (Ξ-VecToPrimVer), vii) a maximum value of the minimum track distance of the Λ and the 2nd π − (π 2 -Λ-MinVecDist), and viii) a minimum value of the distance of the Ξ − vertex relative to the primary one (Ξ-VerToPrimVer).
Starting with the cut settings used in our previous analysis of deep-subthreshold Ξ − production in collisions of Ar + KCl at 1.76A GeV [12] and optimizing further for the present experiment which exhibits different multiplicities and phasespace distributions of the involved particles, we find a signifi- 1 With "track" we mean the trajectory of a particle track extrapolated up to the relevant vertex.
binatorial background (bg).Integration around the peak max-imum within a window of ±5 MeV (≈ ±2σ, with σ being the Gaussian width) we find N Ξ − = 90 ± 18 with the statistical error given.The signal-to-background ratio and the significance, signal/ √ signal + bg, amount to 0.39 and 5.0, respectively.Note that the raw Ξ − yield per LVL1 event of N Ξ − /N LVL1 = 2.8 × 10 −8 is yet a factor seven smaller than the corresponding yield in Ar + KCl reactions at 1.76A GeV [12].We studied also the raw phase-space distribution of the Ξ − baryons.To that purpose, the yield within a window of equation of state (EOS) incompatible with the observation of the neutron stars of two solar masses.This leads to the 'hyperon puzzle'.Many attempts were made to solve this puzzle, e.g. by introducing three-body forces leading to an additional repulsion that can counterbalance the large gravitational pressure and finally allow for larger star masses.Another possibility is offered by calculations using a chiral effective field theory approach.The results [17] with leading order (LO) and next-to-leading order (NLO) demonstrate an attractive interaction for low hyperon momenta (p < 600 MeV/c) but for higher momenta NLO results show a repulsive interactions.Using the femtoscopy technique we performed first studies of the scattering lengths and effective ranges for hyperon-nucleon pairs in p+ A collisions.The p − Λ correlation function are shown in Fig. 4 in comparison to the calculations with LO and NLO results (both versions are plotted separately).The statistical error of data are too large to derive conclusions but shows sensitivity of the method.Future experiments are planned which thanks to the improved DAQ of HADES can provide data sample with statics larger by an order of magnitude.
Another interesting result obtained from studies of Λ particle correlations reveals a significant Ξ − (1321) peak signalling sub-threshold (70 MeV below the threshold of free p + p) production [18].This observation is particularly interesting in connection with the Ξ − signal measured in Ar + KCl reactions at 1.76 GeV and will be discussed in the next section.

Results from A-A collisions
In the 1 − 2 AGeV energy range, particle production in heavy-ion collisions is dominated by pion production which originates mainly from the ∆( 1232 mainly η, are already very low (of order 1 − 2%).Production multiplicities for π 0 and η mesons are known from their decay into real photons from former TAPS measurements at GSI [19].The dielectron invariant-mass distributions measured with HADES in the light C + C 1 and 2 AGeV collisions can be well described by a superposition of N − N collisions [20].However, radiation from the medium-heavy Ar + KCl (at 1.756 AGeV) [21] systems show a significant contribution from a dense collision phase.
and is applied for each time step ∆t.

III. RESULTS
For the results presented here we used calculations with an ensemble of 1000 UrQMD events.However, several runs using different UrQMD events as input had to be performed to obtain enough statistics especially for the non-thermal ρ and ω contributions.Note that in case of the experimental Ar+KCl reaction we simulated the collision of two calcium ions instead, as this makes the calculation easier for symmetry reasons.Effectively it is the same as the Ar+K or Ar+Cl reactions that were measured in the experiment and the size of the system remains identical.To simulate the correct impact parameter distribution, we made a Woods-Saxon type fit to the HADES trigger conditions for Ar+KCl [  Fig. 6 shows the dielectron invariant mass distribution normalized to the mean of the charged pion (π + , π − ) yields, measured independently by HADES, and extrapolated to the full solid angle.At this energy and this collision system it is a good measure of neutral pion multiplicity.The differential distributions obtained in such a way are compared to the expected mesonic e + e − cocktail from the π 0 , η Dalitz and ω decays according to the measured (for π 0 and η) and extrapolated from the m T scaling for the ω multiplicities.One should underline that the ω peak visible in the invariant mass distribution in Ar+KCl collisions constitutes the first measurement of meson production at such a low energy (below its free N − N threshold).As one can see, the e + e − cocktail composed from the meson decays does not explain the measured yields for both collision systems and leave a room for a contribution expected from baryonic sources like Dalitz decays (mainly ∆(1232)), nucleon-nucleon bremsstrahlung and inmedium radiation.The latter one appears to be a dominant source of the radiation, as shown in Fig. 7 by a coarse-grained transport calculation [27].This contribution accounts for a thermal radiation from a fireball [28], in a similar fashion as for SPS [22,23] and RHIC data [24,25].It is mainly given by a radiation from the ρ-meson with spectral function strongly modified due to the meson-baryon resonance couplings [14].This multi-body hadronic interactions lead to a dramatic broadening of the ρ mass distribution and allows for a remarkable consistent explanation of the thermal dilepton gether with the mass spectrum in Fig. 3.The ω decays contribute evidently only a small part to the total pair yield at intermediate and low masses.Note finally that the average ω momentum in the nucleus-nucleus centerof-mass within the HADES acceptance is found from our data to be p = 0.43 GeV/c.This is at least a factor two smaller than the momenta typically observed in ω photoproduction experiments [31][32][33].The ω multiplicity can be discussed in the context of either a scenario of complete thermalization at freezeout or, in the other extreme, of production in elementary N +N collisions.As HADES is a general-purpose charged-particle detector, besides the dielectron results presented here, a wealth of information has been obtained as well on hadron production in Ar+KCl.These findings have already been published in [20] on π ± , in [35] on K + , K − , and φ, in [21] on K 0 s , in [36] on Ξ − , and finally in [37] on Λ and Σ ± .
In particular, from our K + − K − correlation analysis [35], a LVL1 φ multiplicity of M φ = (2.6 ± 0.7(stat) ± 0.1(sys)) • 10 −4 has been found as well as a transversemass slope at mid-rapidity of T φ = 84 ± 8 MeV.Together with the ω multiplicity, this gives a φ/ω ratio of R φ/ω = 0.043 +0.050 −0.015 (stat) ± 0.011(sys).The experimental ratio can be compared to various predictions, running from pure m ⊥ scaling in 4π solid angle, giving R ≃ 0.042, to a full-fledged statistical hadronization model calculation performed with the THERMUS code [38] fitted to our hadron yields [37] and resulting in R = 0.063 ± 0.008.Hence, statistical descriptions agree within error bars with the experimental R φ/ω .As already Figure 8.Comparison of the R φ/ω ratio (HADES) and model value (THERMUS fit) as as a compilation of data from the p + p and π + N reactions (see text).The ratio is plotted as a function of the excess energy above the threshold for the exclusive production in p + p and π + N reactions [30].
The comparison of the extracted chemical freeze-out temperature T chem to the ones extracted from the inverse slope T eff of transverse mass spectra at mid-rapidity for various particles listed in Tab. 1 is not straightforward.In a naive picture the extracted inverse slope parameter T eff include a pure kinetical component T kin plus an additive term, depending on the particle mass m and the square of the radial expansion velocity β.In addition effects like resonance decays deform the spectra complicating this naive interpretation.

p+Nb at
For the fit to the yields obtained from p+Nb reactions we add the charge chemical potential µ Q as an additional free parameter due to the strong asymmetry of the collision system.Apart from the charge chemical potential µ Q , we use the same parameters as above.The extracted parameters are other, which one expects as the suppression of strange particles compared to non strange particles depends mostly on the absolute value of R c and only very weakly on the ratio of R c /R. rates over a broad energy range from SIS18 (HADES), SPS(NA60/CERES) to RHIC energies [28].The underlying connection of the hadronic model to the chiral symmetry is nicely provided by the QCD and Weinberg sum rules relating the spectral functions of the vector (ρ) and the axial-vector (a 1 ) mesons, which mass splitting in vacuum is a direct consequence of the chiral symmetry breaking, with the expectation value of quark condensate also appearing due the symmetry breaking.The latter one can be calculated by lattice QCD in the limit of vanishing µ b while the meson spectral function can be provided by the aforementioned many body hadronic theory.Using such connections, the evolution of spectral functions of the a 1 and ρ meson was calculated and shown to be consistent with chiral symmetry restoration [26].
These remarkable results nicely demonstrate penetrating nature of the dileptons which allows to observe an effect of "shining" of the baryonic matter integrated over the whole collision time.As shown by the model calculations [27,29] the yield of radiation is a direct measure of collision time and hence can serve as a chronometer of the reaction.A further important test of this scenario will be provided by data recently obtained from Au + Au collisions at 1.25 GeV where even larger yield of thermal radiation is expected.
Interesting results on the vector meson production in heavy-ion collisions at SIS18 have also been obtained from analysis of Ar + KCl data.Besides the ω signal discussed above, a surprisingly strong φ meson production has been found from an analysis of the K + K − final state [30].The acceptance corrected φ/K − ratio is found to be 0.37 ± 0.13 which translates into a fraction of 18 ± 7% of negative kaons coming from φ decay.Furthermore, assuming that non-resonant K + K − production is of the same size, as it is known from N + N reactions, even larger contribution of these type of reactions to the anti-kaon production can be deduced.This conclusion is further supported by the observation of the different slopes of transverse mass distributions of kaon and anti-kaon spectra which can naturally be explained as the effect of the φ feed-down [31].This is surprising since strangeness exchange   [1,34] (cross), RHIC [2,3] (stars), SPS [4,5] (triangles), AGS [6] (square), and SIS18 [12] (circle).The filled cross depicts p + p collisions at LHC [35], while the downward and upward pointing filled triangles are for p + A reactions at DESY [36] and SPS [18], respectively.The filled circle shows the present ratio (2) for p (3.5 GeV) + Nb reactions (statistical error within ticks, systematic error as bar).The full curve is a parameterization (see text) of the proton-induced reaction data.The asterisk, diamond and filled star display the predictions of the statistical-model package THERMUS [37], the GiBUU [38,39], and the UrQMD [16,17] transport approaches, respectively.
proach is the UrQMD model [16,17] (version2 3.4).For Ξ − hyperons, we derived a yield of (6.9 ± 2.8) × 10 −7 per event which is more than two orders of magnitude lower than the experimental yield (1) and decreases only by a factor of two, if the channels YY → ΞN (with cross sections from [13]) are deactivated; i.e. in the model hyperon-hyperon fusion is of minor importance for Ξ production in proton-nucleus reactions at 3.5 GeV.The Λ rapidity distribution, however, was fairly well reproduced by UrQMD [31].The resulting Ξ − /(Λ + Σ 0 ) ratio amounts to (3.1 ± 1.2) × 10 −5 (filled star in Fig. 3).The second transport approach we used is the GiBUU model [38,39] (release 3 1.6).We estimated a Ξ − yield of (6.2 ± 0.9) × 10 −6 , a value being considerably higher than the prediction by the UrQMD model, but still significantly lower than the experimental yield (1). Also here, the (π − hyperon → K − N) has been assumed before to be the dominant process in K − production.The HADES result indicate that the φ meson is also produced in multi-step processes involving shortlived resonances.Such scenario is corroborated by BUU and UrQMD transport calculations [32], [34] which reproduce the yields and spectral distributions of K + K − and φ mesons.Fig. 8 shows the ratio of the φ to ω multiplicities measured in Ar + KCl collisions at 1.756 AGeV, together with predictions of the statistical model THERMUS [35] and results from elementary reactions [30,31].The data points are plotted as a function of the excess energy above the production threshold for the exclusive φ production in p + p and π + N reactions, respectively.One can see from this comparison that in the heavy-ion reaction R φ/ω is more than one order of magnitude larger than in N + N collisions and also at least a factor 3 − 5 larger than in pion-induced processes.On the other hand, the ratio is consistent with the calculation of statistical thermalization model assuming no suppression due to OZI rules [31].
The ability of HADES for the selection of displaced secondary vertices arising from weak decays and the high statistics accumulated for the collision system Ar + KCl at 1.76AGeV allowed to investigate the deep-subthreshold production ( √ s NN − √ s thr =-640MeV) of the double-strange Ξ − (1321) hyperon [37].The Ξ − (1321) was reconstructed in the Λ − π − invariant mass distribution thanks to a high-purity signal of Λ identified in the p − π − invariant mass distribution.The obtained yield of the cascade has been fitted together with the other hadron yields measured by HADES by the THERMUS and the results are shown in Fig. 9.The fit describes all particle yields, except Ξ − (1321), where an enhancement of 15 ± 6 is observed.The freeze-out condition corresponds to T=70±3 MeV and the chemical potential µ b = 748 ± 8 MeV [31] and line up with a general freeze-out line obtained from the fit to word data.
Increase of strangeness production (particularly with multi strange content) has been recently predicted to be an ideal probe for quark deconfinement in baryon-rich matter in the FAIR/NICA energy range [38].Hence pioneering results on Ξ − production of HADES are of large importance.In the works [39], [40] the strong Ξ − (1321) production observed in Ar + KCl collision system has been accounted for by strangeness exchange reactions of the type hyperon-hyperon → NΞ − (1321), where hyperons are produced in the earlier reaction stage.However, our measurement of Ξ − production in p + Nb, presented in previous chapter, shows similar enhancement over the statistical hadronization model [31].Fig. 10 shows the ratio of production rates of Ξ − (1321) and Λ + Σ 0 as a function of the total CM energy in N + N collisions measured by HADES and other high-energy experiments.The reconstructed strength of the signal is compared to calculations performed for Au + Au collisions with the statistical model of [41].While high-energy data are well described, the present experimental ratio is underestimated, by the model by an order of magnitude.The similar enhancement over the statistical model observed in p + Nb on seems to be in contradiction to the explanation offered for the Ar + KCl case since the contribution of the multi-step processes should be negligible in p + A reactions.Alternative explanation suggesting production from high mass (> 2 GeV/c 2 ) baryonic resonances with significant branches for the Ξ − decay has been proposed in [32].Furthermore, similar mechanism seems to be also instrumental for the φ meson production measured by HADES.However, it remains to be shown by dedicated measurements with proton-proton or pion-proton experiments that resonances with such decay indeed exist.Nevertheless, the proposed mechanism seems to be attractive since it explains not only the production of hadrons with multi-strange quark content in p+A and A+A reactions at low energy but also provides an interpretation for the remarkable success of the statistical hadronization description.A dedicated calculations with URQMD show that 2 − 3collisions between nucleons and nucleon resonances are sufficient to produce such high mass resonances and reproduce particle yields which are consistent with predictions of statistical hadronization models [33].It is expected that new data obtained with Au + Au collisions where kaons, η, Λ and φ have been identified will bring another interesting insight into the production of strangeness at SIS18 energies.
In-medium effects on kaons have been studied by means of K 0 s meson transverse momentum distributions in Ar+KCl collisions at 1.76AGeV taking advantage of the good acceptance of HADES at low transverse momentum for the K 0 s → π + π − reconstruction [42].We compared p t distributions for different rapidity bins with the corresponding results by the IQMD transport approach with and without taking into account a repulsive K 0 -nucleus potential.For all rapidity bins, but most evidently at mid-rapidity (shown in Fig. 11), data support calculations with the repulsive potential.Our data suggest a repulsive in-medium K 0 potential of about 40MeV which is slightly higher as compared to results obtained from experiments studying K 0 s production off nuclei [43].Simlar studies have also been performed in p + Nb collisions [44] confirming presence of a repulsive momentum-dependent kaon potential as predicted by the Chiral Perturbation Theory (ChPT).For the kaon at rest and at normal nuclear density, the ChPT potential amounts to ∼ 35 MeV.

Conclusions
Selected results on dielectron and strangeness production in p + Nb and nucleus-nucleus collisions has been reviewed.
We find that ρ meson production in our energy range is strongly affected by a strong coupling to low-mass baryonic resonances which is reflected in a significant broadening of the meson mass distribution visible in p + A and A + A reactions.Particularly, results from A + A demonstrate a significant contribution from the dense phase of the collisions which can be successfully described by a thermal radiation from the in-medium ρ in agreement with the higher energy data.An important verification of this scenario will be provided by Au + Au data recently collected by HADES.However, also the vacuum spectral function, as concluded from p + p reactions, has a non-trivial shape.It is expected that results from pion induced reactions will allow to understand details of the meson-baryon couplings and further scrutinize the underlying microscopic model of the meson-baryon interactions.
Studies of the ω production off nucleus show a strong absorption of the meson in nuclear matter in accordance with the results from photon induced reactions.The ω and φ signals have also been reconstructed in Ar + KCl collisions at energies below the N − N production threshold.A surprisingly large R φ/ω production ratio (more than one order of magnitude larger than in N + N collisions) has been found, indicating no suppression for the φ production and consequently a significant contribution to the K − production.The latter one is particularly surprising in view of previous interpretations assuming dominant role of strange exchange reactions.Also strong enhancements of the double strange Ξ − (1321) production above predictions of statistical hadronization models have been found as well in p + Nb as in Ar + KCl collisions.These intriguing results call for further experimental studies of strangeness production in his energy regime.New data obtained with Au − Au system will provide additional valuable information.High purity and statistics Λ sample obtained in HADES experiment allowed also for the first studies of Λ − p interactions in p + A collisions with femtoscopy methods revealing sensitivity to disentangle repulsive and attractive contributions.
Investigations of in medium kaon potential have been performed with the measurement of K 0 production in Ar + KCl and p + Nb supporting a strong (U=40 MeV) repulsive potential.
The author thanks GSI Darmstadt for support for this work.

FIG. 2 :
FIG. 2: Comparison of dielectron cross sections as a function of the invariant mass measured in p+p and p+Nb collisions.The p+Nb data are displayed with full circles and red horizontal lines indicating the systematical errors, while the p+p data are displayed with open circles and yellow horizontal lines.For the p+p data a PYTHIA dilepton cocktail is displayed in addition.

FIG. 3 :
FIG. 3: Left: Comparison of the invariant mass spectra for e + e − pairs with Pee > 0.8 GeV/c from p+p and p+Nb.The inset shows a linear zoom into the vector meson region together with a fit to the ω structure for the p+Nb data.Right: For pairs with Pee < 0.8 GeV/c.The p+p data have been scaled according to a Glauber model.

Figure 1 .FIG. 2 :
Figure 1.Comparison of dielectron cross sections as a function of the invariant mass measured in p+ p and p+Nb collisions at beam energy of 3.5 GeV.For the p + p data, a PYTHIA dilepton cocktail composed of various e + e − sources, defined in the legend, is displayed in addition[6,10]

Figure 2 .
Figure 2. Comparison of the invariant mass spectra for e + e − pairs with p e + e − > 0.8 GeV/c (left panel) from p + p and p+ Nb.The inset shows a linear zoom into the vector meson region together with a fit to the ω peak for the p+Nb data.Right: for pairs with p e + e − <0.8 GeV/c.The p + p data have been scaled as described in the text[6]

FIG. 4 :
FIG. 4: Left: Same as in the right side of Fig. 3 but zoomed into the vector meson region.The shaded bands represent the systematic uncertainties due to the normalization.Right: Excess yield in the p+Nb data after subtraction of the scaled p+p reference data (the ω contribution has been subtracted in both data samples).The grey region corresponds to the invariant mass range plotted in the left picture.

Figure 3 .
Figure 3. Left: Same as in Fig. 2 but zoomed into the vector meson region.The shaded bands represent the systematic uncertainties due to the normalization.Right: Excess yield in the p + Nb data after subtraction of the scaled p+p reference data (the ω contribution has been subtracted in both data samples).The grey region corresponds to the invariant mass range plotted in the left picture

3 FIG. 11
FIG. 11. online).Comparison of the experimental Λp correlation function (open circles with error bars) to the LO (green) and NLO (red) scattering parameter set included in Eq. (4).The error bands in the theory curves correspond to the errors of the Λp source size determination.

Figure 4 .
Figure 4. Comparison of the experimental Λ − p correlation function (open circles with error bars)to the LO (green) and NLO (red) scattering parameter set (see ref[17] for details.The error bands in the theory curves correspond to the errors of the Λ − p source size determination[16]

FIG. 1 :
FIG. 1: The experimental Λ − π − invariant-mass distribution.The error bars show the statistical errors.The curve represents a combination of a Gaussian and a polynomial function used to fit the data.

Figure 5 .
Figure 5. Λ − π − invariant-mass distribution measured in p + Nb collisions at 3.5 GeV.The error bars show the statistical errors.The curve represents a combination of a Gaussian and a polynomial function used to fit the data.[18]

Figure 7 .
Figure 7. Similar distributions but compared to coarse-grained transport calculations which account for the radiation from dense collision phase mediated by in-medium ρ and, to smaller extent, ω meson [27].

FIG. 8 :
FIG.8:(Color online) Comparison of the R φ/ω ratio obtained in this work with its statistical model (THERMUS fit) value as well as with a compilation of data from elementary p+p and π+N reactions (see text).The ratio is plotted as a function of the excess energy ǫ in the NN → NNφ and the πN→Nφ reactions, respectively.

6 ) fm and χ 2 15 Fig. 1 .
Fig. 1.Yields (filled red circles) of hadrons in Ar+KCl reactions and the corresponding THERMUS fit values (blue bars).The lower plot shows the ratio of the experimental value and the THERMUS value.For the Ξ − the ratio number is quoted instead of a point.

Figure 9 .
Figure 9.Yields (filled red circles) of hadrons in Ar+KCl reactions and the corresponding THERMUS fit values (blue bars).The lower plot shows the ratio of the experimental value and the THERMUS value.For the Ξ − the ratio number is quoted instead of a point.[31].