################################################################
Page 1
Good afternoon everybody, today I would like presents the results of the OLYMPUS experiment which measured the hard two-photon contribution to elastic lepton-proton scattering.
################################################################
Page 2
The motivation for the OLYMPUS experiment can be seen in this figure. It shows multiple measurements of the proton space-like electromagnetic form factors. Although the FF have been studied for decades we can't say that we know everything about them.

The very first measurement method was elastic electron-proton scattering or such a called Rosenbluth separation method. Where the cross section had to be measured at multiple energies and scattering angles. Its results results, shown in red, are consisted with the form factor ratio being equal one.

In the beginning of the century, new experimental techniques were developed - polarized targets and beams became available. This allowed to measure the proton FF with using recoil polarization technique - scattering of a polarized electron beam off a proton target and measuring the transverse and longitudinal polarization of the recoil proton. Results of this method is shown in blue.

As you can see, while Rosenbluth data show the same FF ratio at all Q^2, the polarized data show a linear decrease of the FF ratio with an increase of Q^2. After the discrepancy was found, new measurements using Rosenbluth separation were performed to exclude possible experimental errors but their results only supported the discrepancy at large Q^2.

- How two methods which measure one physical quantity can give different result?
- If both types of experiments are correct we should look if something else is missing.
- The answer may be ...
################################################################
Page 3
The hard TPE contribution which, until recently was neglected, is considered to be responsible for this discrepancy. The TPE has an effect of few percent on the Rosenbluth separation method while the polarization techniques is affected only by 1% or less.

The TPE effect can be directly estimated via measurement of relative positron-proton to electron-proton elastic cross section. As this equation shows, the interference term between one- and TPE changes its sign with the lepton species it is possible to determine the size of the TPE from the cross section ratio.
################################################################
Page 4
Event though multiple measurements of the TPE are available, as shown in the figure, all the data comes from sixties and the results, taking the uncertainties into account, are not conclusive.

Additionally, a number of theoretical calculations are shown. Depending on the approach, models predict that the effect of the TPE can be as high as 15% at the lower epsilon, while some show very small and even negative impact on the positron-proton to electron-proton cross section ratio.

The lack of precise experimental data and strong model dependence called for new experimental efforts to solve the proton form factor discrepancy.
################################################################
Page 5
Three groups answered the call - CLAS, VEPP-3, and OLYMPUS.

- CLAS experiment in JLab utilized tertiary simultaneous electron and positrons beams with energies from 0.5 to 3.5 GeV
- VEPP-3 in Novosibirsk had electron and positron beams with fixed energy of 1 and 1.6 GeV
- OLYMPUS experiment at DESY used a fixed energy 2 GeV electron and positron beams

As you see on the figure, all three experiments have different kinematic ranges. Having fixed beam energies VEPP-3 (blue lines) and OLYMPUS (red line) have tightly constrained Q^2 and epsilon range while CLAS have wide bins covering big Q^2 - epsilon areas (golden). The white-golden gradient corresponds to the cross section values, therefore the results are dominated by the low Q^2 data.

A comparison of the results produced by all three experiment will be shown at the end of the presentation.
################################################################
Page 6
Let me introduce you the OLYMPUS detector. You can see the beam pipe with a lepton beam coming from left to right through the target chamber. A hydrogen gas target 60 cm long was used during the experiment with a typical operation temperature slightly below 70 K.
################################################################
Page 7
The beam pipe was surrounded by the eight coil toroidal magnet. The nominal magnet current during data taking was 5000 A what created fields of about 0.28 T in the tracking region.
################################################################
Page 8
Two drift chambers were placed symmetrically in the left and right horizontal sectors between toroid coils. They covered polar angles from 20 to 80 degrees and from -15 to 15 degrees in azimuth. The drift chambers provided measurement of momentum, charge, polar and azimuthal scattering angles, and vertex position of outgoing charged particles. 
################################################################
Page 9
Behind the drift chambers, time of flight detectors were installed. Having 18 scintillator bars in each sector, the TOFs served as the trigger for most OLYMPUS detectors as well as for particle identification (proton-positron).
################################################################
Page 10
Luminosity during data taking was controlled by three different systems.
- Slow control luminosity
- 3 multi-write proportional chambers placed at the polar angle of 12 degrees in each sector
- Two symmetric Moeller/Bhabha monitors were located in both sectors at the polar angle of 1.29 degrees. This is symmetric Moeller/Bhabha scattering angle taken the beam energy of 2 GeV
################################################################
Page 11
- Slow control allowed an on line luminosity measurement which was determined based on the beam current, target density (from molecular flow calculation), and the detector live time. It provided luminosity determination with the uncertainty of 5% for each beam species and 2% relative positron to electron uncertainty.
- 12 degree multi-wire drift chambers measured elastically scattered electrons and positrons in coincidence with the recoil proton detected in drift chambers. Its absolute and relative uncertainties are 2.4 % and 0.46 %, respectively.
- The symmetric Moeller/Bhabha monitor was originally designed to determine luminosity through measurements of elastic Moeller and Bhabha scattering. Due to very big uncertainties another method was developed. Multi-interaction lepton-lepton and lepton-proton events detected by the SYMB monitors were used to measure the luminosity.
################################################################
Page 12
- Multi-interaction event methods was chosen as the most accurate among three. At the scattering angle of 1.29 degrees where the average Q^2 is equal to 2 MeV^2 and the epsilon almost unity the TPE is negligible.
- The data collected by the MPWC were used to determine the size of the TPE at the scattering angle of 12 degrees.
################################################################
Page 13
Here I would like to outline the time line of the OLYMPUS experiment.
- The letter of intent was prepared in 2007 followed by the approval of the OLYMPUS proposal in 2008
- In the BLAST detector was shipped from MIT to DESY
- The commissioning was completed in 2011...
- ... followed by two data taking runs in 2012-2013. Between two runs a number of improvements were done to the DORIS storage ring what significantly improved running conditions
- After the shut down we continue to collect cosmic data and surveyed all detectors as well the magnetic field. Beam position monitors were calibrated.
- Since than the data have been analyzed with the first results being recently published.

As you can see the OLYMPUS experiment was conceived, prepared, and realized in a very short time frame.
################################################################
Page 14
The figure shows integrated luminosity collected in two runs. The data was collected alternating between electron and positron beams on a daily basis. Both positive and negative toroid polarities were used. Due to very high background rates in the drift chambers at negative field most data were collected using positive toroid polarity.

Due to the target cell damage, during the first run only about 10% of the total data was collected. Thanks to multiple improvements to the DORIS ring the designed total luminosity of 4 fb-1 was achieved during the second data taking period.
################################################################
Page 15
The analysis of the OLYMPUS data involved both experimental and simulated data. Simulation is closely tied to the collected experimental data, slow control information, and the beam parameters provided by the DORIS storage ring.

On a run by run basis the operational data is feed to event generators. The primary particles are propagated using Geant4 engine. The signal produced in the detector systems is then digitized to simulate the response of the real detector. Thus, the simulated data is converted to the same format as the experimental data.

That allowed to analyze the experimental and simulated data with the same code.
################################################################
Page 16
A radiative event generator for the elastic lepton-proton scattering was developed by the MIT group. On the figure you can see the ratio of elastic positron-proton to electron-proton scattering cross section with the lepton scattering angle plotted on the X axis and its momentum on the Y axis. The black color corresponds to the cross section unmodified by the radiative corrections, the orange stripe shows the cross section ratio enhancement, while the blueish area shows the reduced ratio. 
################################################################
Page 17
Four different prescriptions were used to determine the size of the TPE effect. With the epsilon on the X axis and the deviation from the Born level cross section on the Y axis you can see that depending on the epsilon and the model the TPE can affect the cross section ratio by 1% to 6.5%.
################################################################
Page 18
The results I'm about to show you are a combination of four different analyses performed by different people. They all share the same experimental and simulated data as well as binning in Q^2 and epsilon. The results were normalized to the same integrated luminosity determined with the SYMB monitors using MIE method.

The difference between these analyses were in the different particle identification methods, different cuts and selection procedures.
################################################################
Page 19
This is the OLYMPUS paper which was published around a month ago where you can find a more detailed description of the data analysis and TPE determination procedures.
################################################################
Page 20
Here are the results of the OLYMPUS experiment using Mo-Tsai to all orders together with a theoretical calculations by Blunden and Tomalak, and the phenomenological predictions from Bernauer. The results indicate small size of the TPE effect at these beam energies with most data points lying below unity with an exception of two points where the TPE contribution increases up to about 2% at epsilon=0.46. The results are mostly consistent with the Tomalak's calculations and Bernauer predictions while being consistently lower than the latest calculations of Blunden.
################################################################
Page 21
This figure shows a difference between Blunden's latest calculations (N+Delta) and results of the OLYMPUS, CLAS, and VEPP-3 experiments. The theoretical TPE contribution was calculated for each data point individually to take into account different Q^2 and epsilon values.

We can see that the results of all three experiments are agree between each other and consistently smaller the theoretical predictions which are necessary to reconcile two proton form factor measurement methods.

Thus, we can say that the available data doesn't provide a definitive answer to the discrepancy between the proton form factor ratio obtained using Rosenbluth separation method and polarization transfer technique. At least not the in lower Q^2 region. And the contribution of the TPE at large Q^2 has to be determined in future measurements.
################################################################
Page 22
Shorter repetition of the last paragraph from the previous slide.
################################################################