In this project we propose to develop a software attached to an existing experimental setup. At the partner institution, National Institute for Nuclear Physics and Engineering – Horia Hulubei (IFIN-HH) in the research infrastructure Extreme Light Infrastructure – Nuclear Physics facility (ELI-NP, financed by the project with the same name ELI-NP), a specific measurement setup for the temporal and spectral characterization of the ultrashort and ultra-intense laser pulses has been installed (or is in the process of installation), namely the D-scan and FROG setups. The conventional reconstruction algorithm – which is based on the mathematical equations describing the physical process – is at low level of implementation, it is not straightforward to use and it converges slowly. This is the main motivation of this proposal: we intend to develop a fast pulse reconstruction method, based on high-performance machine learning algorithm, in order to make the best use of the existing experimental setup. The main purpose is to optimize and make faster the process of temporal-spectral reconstruction and characterization of the laser pulses generated at the ELI-NP research infrastructure. The scope of this project extends to the testing of the proposed algorithm, demonstration of its feasibility through proof-of-concept measurement and reconstruction algorithm, by confronting the preliminary results with measurements and pulse reconstruction performed previously with other methods.

An adaptive mirror will be inserted into the system, this will facilitate the fast pulse reshaping in a controlled manner. With the use of the adaptive mirror one can correct the pulse’s phase front, and in this way it is possible to avoid the damage of further optical elements due to unwanted pulse peaks. The machine learning algorithm modified for this purpose will allow the pre-compensation of the phase front distortions occurring farther along beam steering and focusing, facilitating the controlled pulse shape on target.


General objective:

The general objective of this project is to develop a fast numerical instrument which performs the temporal-spectral reconstruction of ultrashort laser pulses at ELI-NP and to experimentally validate it.


Specific objectives:

O1. Build a computer code with artificial neural network (ANN) algorithm for the reconstruction of the laser pulse from its D-scan / FROG trace.

This objective practically contains a software development. Our research group at INCDTIM has some experience in programing ANN [9,10]. We have a module of code done at a basic level, where both the input and output parameters are scalars. The major development will be to modify the ANN such to be able to receive as input any 2D picture (D-scan / FROG trace), and learn to predict (i.e. reconstruct) the spectral amplitude and phase which are functions of laser wavelength (i.e. vectors). At first stage of development we will train the ANN with simulated traces, provided by a computer code having a 3D non-adiabatic model of laser pulse propagation implemented.

O2. Attach the functional software to the existing D-scan / FROG setup at ELI-NP.

The ANN based software will be suitable for fast pulse reconstruction. Its role will be accomplished if it will be working on-site and reconstruct the pulses shot-by-shot. The reconstruction procedure will become faster and more accurate as more and more pulses will have been reconstructed, since all previously reconstructed pulses enrich the pool for teaching the ANN.

O3. Study the influence of the laser pulse wavefront on focusing and reconstruction process

Most characterization devices assume flat wavefront of the pulse at the entrance, although (un)intended wavefront distortions characterize the pulses. The software will be extended to assess the effects of wavefront distortions on the focusing and pulse measurements. Controlled wavefront distortions can be introduced on all six arms of the HPLS laser system at ELI-NP. Adaptive mirrors are available at ELI-NP, making possible the experimental tests at the AVESTA laser facility.

Text Box: Laser pulse shaping and characterization: measurement, 
reconstruction software and adaptive optics — Pulse-MeReAd
Text Box: National Institute for Research and Development  of Isotopic and Molecular Technologies Cluj-Napoca