Towards smart lasers: nonlinearity management in optical fibers for ultrashort laser pulse generation supported with machine learning


The Project focuses on understanding and exploiting the nonlinear effects in optical fibers to generate ultrashort laser pulses. Nonlinear effects in fibers, such as self-phase modulation or four-wave mixing are usually avoided since they lead to distortions of the spectral shape of a laser pulse. Here, we will investigate the possibility of creating “smart” ultrafast lasers, exploiting the nonlinear effects, and turning them into an advantage.

The research goal is the development of autonomous, intelligent ultrafast laser systems, generating the shortest possible pulses without manual adjustments of the operator. To achieve the goal, we propose the use of arbitrary phase shaping of ultrashort pulses, combined with machine learning for self-optimization of the laser system. In our approach, we will use programmable optical filters (so-called pulse shapers) to fully control the spectrum, phase, and dispersion of ultrashort pulses. Our studies will lead to the development of self-optimized ultrafast laser systems. Finally, we will use the optimized lasers as drivers for various nonlinear processes, showing that simple fiber technology is suitable for generating ultrashort pulses in non-typical spectral bands. Therefore, we believe that the Project results will strongly contribute to the current knowledge in ultrafast laser technology.

The proposed 5-year research program responds to the current needs of ultrafast laser technology: development of novel techniques of ultrashort pulse generation in different spectral ranges, being essential for applications in biomedicine (multi-photon imaging) and spectroscopy (cavity-enhanced detection of molecules). The techniques, which will be investigated and developed during the Project include:

  • Coherent spectral broadening and temporal compression in optical fibers – to obtain few-cycle optical pulses in the spectral range of 1.55 μm,
  • Application of machine learning in ultrafast photonics: investigations will include recurrent/feed-forward neural networks, genetic algorithms, evolutionary algorithms, and others,
  • Gain-managed nonlinear amplification in Yb-doped fibers – for the generation of intense, ultrashort pulses beyond the gain bandwidth of Yb ions,
  • Gain-managed nonlinear amplification in Er-doped fibers – for the generation of intense, ultrashort pulses beyond the gain bandwidth of Er ions,
  • Nonlinear spectral conversion: intra-pulse difference frequency generation (IDFG) and second harmonic generation (SHG), for the development of robust, simple, ultra-broadband mid-infrared radiation sources, and ultrashort-pulsed NIR/VIS sources.

The 5-year research plan is divided into four tasks. Each task focuses on a specific scientific problem related to ultrashort laser pulse generation: Task 1: Few-cycle pulse generation at 1560 nm wavelength by spectral phase correction Task 2: Amplification of the 1060 nm pulses in Yb-doped fibers the gain-managed nonlinearity (GMN) regime Task 3: Amplification of the 1560 nm pulses in Er-doped fibers in the gain-managed nonlinearity regime Task 4: Nonlinear conversion of optimized ultrashort pulses to other spectral ranges Within our Project, we will apply the self-optimization algorithms to much more complex systems than reported so far in the literature. By using phase shaping, nonlinear spectral broadening and nonlinear amplification, we will study the possibility of generating few-cycle, high-power pulses in different spectral ranges. To the best of our knowledge, self-tuning machine learning algorithms were not yet applied to highly nonlinear Er- and Yb-doped fiber amplifiers and GMN amplifiers. Therefore, we believe that our research and will deliver new, fundamental knowledge in the field of laser technology.

The Project is carried out at the Wrocław University of Science and Technology. The team members involved in the project are:

- Alicja Kwaśny
- Mikołaj Krakowski
- Grzegorz Soboń

The project is funded by the National Science Centre (NCN)

Project duration: 05.04.2022 - 04.04.2027
Amount of funding: 2 081 400,00 PLN
Contract no.: UMO 2021/42/E/ST7/00111