Faites connaître cette offre !
Reference : UMR8609-REJBOD-007
Workplace : ORSAY
Date of publication : Monday, June 10, 2019
Type of Contract : FTC Scientist
Contract Period : 24 months
Expected date of employment : 1 October 2019
Proportion of work : Full time
Remuneration : Monthly gross salary between 2695€ and 3841€ according to experience
Desired level of education : PhD
Experience required : 1 to 4 years
The activity will be carried out within the framework of AGATA (Advanced GAmma Tracking Array) and GRETA (Gamma Ray Energy Tracking Array), high performance gamma ray detectors for nuclear physics studies. The focus of the project consists in developing a new gamma-ray tracking algorithm using machine learning. The Post. Doc. researcher will develop the new algorithm using simulated data in order to capture aspects beyond Compton scattering (materials, geometry, etc.) and also ensure the proper functioning of neural networks before going further using experimental data. He/she can be involved in the performance upgrade of AGATA by scanning the many data sets previously accumulated under very different conditions (trigger, counting rate, multiplicity, etc.) in order to evaluate / better understand the instrument. This systematic study is a very important aspect for AGATA in order to identify possible routes to reach the ultimate performance of the instrument.
The successful candidate will work on the aspects mentioned above and will naturally participate in the AGATA physics campaign at GANIL, as well as in the physics experiments conducted by the group using Gammasphere or GRETINA / GRETA in the USA.
Adapt and apply novel tracking techniques using Machine Learning (Training, validation and tests)
Optimization/upgrade of AGATA performance
Work in collaboration with the AGATA and GRETA members
Participate in the AGATA physics campaigns at GANIL
Prepare research articles for publications
Present the work at scientific conferences, workshops, meetings and seminars
The candidate will be a member of the SNO group and will naturally participate to nuclear structure experiments/analysis with AGATA, GRETA or Gammasphere
The successful candidate should possess excellent technical skills, as demonstrated by scientific publications in major refereed journals, conference presentations, etc., and a general knowledge of physics.
Minimum qualification : PhD in physics or computer science/machine learning
-programming in C, C++, ROOT
-experience in artifical intelligence techniques is highly desirable
-experience in semiconductor detectors for gamma-ray spectroscopy is highly desirable
The position will be held within the SNO (Nuclear Structure) research group at Orsay. The group is part of the CSNSM (Centre de Sciences Nucléaire et de Sciences de la Matière).
The laboratory comprises about 80 staff and is supported by the IN2P3/CNRS, Université Paris-Sud and Université Paris-Saclay.
The laboratory is situated on the Orsay Campus south of Paris. The SNO group focuses on nuclear structure research studies of nuclear shapes, of exotic nuclei, and of heavy and super-heavy nuclei. The group members are also strongly involved in development of instrumentation for nuclear physics studies such as those using germanium detectors for high-resolution gamma-ray spectroscopy.
To date, the most advanced implementation of the gamma-ray tracking concept can be found in two arrays: AGATA (Advanced Gamma Tracking Array) and GRETA (Gamma Ray Energy Tracking Array) in the EU and USA, respectively. These spectrometers are being designed and built to eventually achieve nearly full 4π coverage.
A number of methods have been developed to track the data from these arrays and they are mostly based on, or make use of, at least the Compton–scattering formula. Both AGATA and GRETINA/GRETA are routinely employing what is known as the forward tracking algorithm.
As Machine Learning techniques are used to detect γ-ray bursts in optical surveys, it might also be possible to employ this new technology for tracking of γ rays in AGATA. The Compton–scattering formula, Klein–Nishina formula, as well as probabilities for photoelectric absorption and Compton scattering could be imposed as rules for this clustering machine learning. Using the Machine Learning concept represents a new challenging exploratory R&D for γ-ray tracking. This new concept should improve the efficiency, P/T and energy resolution metrics to obtain the best resolving power. First, it will be necessary to carry out the effort of learning / training algorithms of tracking by using simulated data in order to capture aspects beyond the Compton diffusion (materials, geometry, etc ...) but also to ensure that learning neural networks behave properly before going further with experimentally obtained data.
Furthermore, the increasing number of AGATA detectors and its coupling to various ancillaries (with different trigger conditions) has clearly demonstrated a growing complexity and even sometimes a misunderstanding of the resulting performances of the array. With this complexity, the need and the possibilities for a better characterization of the instrument come naturally (?) and even become a necessity for the acceptable performance of the important investment that AGATA represents.
Constraints and risks
This position has an anticipated start date of October 1, 2019 and is for an initial duration of two years.
Short duration trips to France and abroad.
We talk about it on Twitter!