AiPT Research Seminars

From wave turbulence to integrable turbulence and soliton gases
Prof Stéphane Randoux, Université de Lille
Abstract:
The nonlinear propagation of random dispersive waves has been an active research topic in nonlinear physics since the 1960s. Historically, a very important part of the work on this subject has been focused on weak wave turbulence. Wave turbulence theory (WT) deals with the non-equilibrium statistics of incoherent and weakly nonlinear dispersive waves in non-integrable systems. On the other hand, many physical systems are described at leading order by partial differential equations (such as the nonlinear 1D Schrodinger equation) that are integrable in the sense that they can be solved using the inverse scattering transform (IST) method. Nowadays, the theoretical description of nonlinear random wave fields in integrable systems is addressed in the framework of so-called “integrable turbulence”, a research area introduced by Zakharov in 2009. In this talk, I will review experimental and numerical developments on the subject of integrable turbulence with a focus on the topic of soliton gas.

TRENDS IN SCIENTIFIC PUBLISHING
Oliver Graydon, Chief Editor of Nature Photonics
Abstract:
An overview of the latest innovations, trends and future direction of journal publishing with a focus on the relevance to the optics and photonics sectors. Topics to be covered include the growth of journals in the optics sector, the rise of open access publishing, the evolution of peer review, the emergence of open data and code and advice on writing high quality scientific papers.

Computation with degenerate optical parametric oscillator networks
Prof Hiroki Takesue, NTT Basic Research Laboratories, NTT Corporation
Abstract:
As the progress of the digital computers shows the signs of saturation, exploration of new computing schemes based on the dynamics of various physical systems is becoming an important research field. Among many schemes, a coherent Ising machine (CIM), which is an Ising model simulator based on a network of optical parametric oscillators (DOPOs), is drawing attention as a way to solve combinatorial optimization problems efficiently. A CIM employs a km-scale fiber cavity including a phase sensitive amplifier (PSA) based on degenerate optical parametric amplification in a periodically-poled lithium niobate (PPLN) waveguide, where thousands of time-multiplexed DOPO pulses are generated. In the paper “A coherent Ising machine for 2000-node optimization problems”, we reported a CIM that enabled flexible (from very sparse to all-to-all) couplings among 2,000 DOPO pulses using measurement-feedback technique. In this talk, I will describe the basic concept of the CIM, and report our recent progress on the CIM research, including a comparison of computational performance with a quantum annealing machine and a realization of 100,000 spin CIM. I will also describe an implementation of a spiking neuron using a pair of DOPO pulses, which may be useful for future research on brain-inspired computations.

Nd doped fiber for E-band amplification
Dr Leily Kiani, Lawrence Livermore National Laboratory (LLNL), US
Abstract:
There are a few fiber design approaches for suppressing competitive gain in doped core fibers by spectrally filtering light the core. Fibers having concentric index profiles like the W-profile, sometimes referred to as depressed-well or trench profiles, have been developed for inducing loss of the fundamental guided mode at long wavelengths. Here the confinement losses grow as the mode field expands for long wavelengths generating an extended short-pass filter. An alternative filtering scheme based on high dispersion elements located adjacent to the core can provide spectrally selective filtering. Dr. Leily Kiani will review the design, fabrication, and test results of a spectrally selective filtering Nd-doped silica fiber for E-band amplification. The waveguide design includes elements that induce core loss at selective wavelengths in competitive bands, enabling efficient amplification of 1400 nm.

Fiber lasers activity at Tampere University
Dr. Regina Gumenyuk, Tampere University
Abstract:
Fiber lasers have already gained an excellent reputation due to a number of features such as good beam quality, high plug-in efficiency, stability and compactness. However, there is a constant demand for advanced performance. Our research focuses on investigating novel fiber lasers for emerging applications such as green material processing, high-precision bioimaging, and sensitive spectroscopic techniques or new energy sources. Our primary interests are new wavelength range, nonlinear intracavity soliton dynamics, the investigation of novel laser regimes and high-power lasers based on an active tapered amplifier technology and coherent beam combining. Recently, we have launched a new research direction – direct generation of the structured light in an active helical fiber.

Long wavelength InAs-based quantum cascade lasers
Prof Alexei Baranov, Research Director, IES – Institut d’Electronique et des Systèmes Université de Montpellier
Abstract:
Quantum cascade lasers (QCLs) have now become a versatile laser technology covering an extremely broad spectral range from mid-infrared to THz frequencies. These devices can be referred to as GaAs-, InP- and InAs-based QCLs depending on the substrate used for their growth. IES is a recognized leader in the development of InAs-based QCLs. The initial interest to this material system was due to the giant conduction band offset of 2.1 eV between InAs and AlSb, which made it possible to obtain high energies of intersubband transitions and to demonstrate the shortest QCL emission wavelength of 2.6 µm.

The small electron effective mass in InAs is another property of antimonides attractive for use in QCLs, which can provide a higher intersubband gain compared with other materials. This advantage can be fully exploited in long wavelength devices emitting beyond 10 µm, where the active quantum levels of the laser transition are close to the bottom of the conduction band and the electron effective mass is not increased by the nonparabolicity effects. Even the very early long wavelength InAs-based QCLs outperformed InP- and GaAs-based counterparts operating in the 16-24 µm range. With these QCLs we demonstrated the first semiconductor lasers operating at room temperature up to a wavelength of 21 µm. Using a dielectric waveguide instead of the usually employed in long wavelength QCLs double metal surface plasmon waveguide we have succeeded to further improve the device performance. The waveguide improvement was then completed with a design and doping optimisation, which resulted in the achievement of room temperature continuous wave (cw) operation at wavelengths up to 17.8 µm, thus establishing a new record for any semiconductor laser. We have obtained cw operation up to 320K for QCLs emitting near 11 µm and up to 240K at 21 µm. Pulsed operation at 25.5 µm at temperatures up to 240K has also been demonstrated. The state-of-the art of the long wavelength mid-infrared QCLs emitting below the reststrahlen band will be reviewed. Issues related to the multiphonon absorption in the considered spectral region will also be discussed.

Tissue spectroscopy for the development of noninvasive cancer detection
Prof Luís M. Oliveira, Porto University
Abstract:
Tissue spectroscopy is one of the Biophotonics methods used to acquire information about the internal composition of biological tissues and fluids. Although this method provides valuable information to develop clinical procedures based on the use of light, it presents certain disadvantages, such as the necessity of tissue excision to perform measurements. Using a fast calculation method based on the photon diffusion theory it was possible to calculate the spectral optical properties of various animal and human tissues from the ultraviolet to the infrared. The results obtained from the spectral absorption coefficient of different tissues provided physiological and diagnostic information associated with cancer development. In the case of rabbit pancreas, a similar accumulation of melanin and lipofuscin demonstrated that these pigments accumulate in tissues as a result of the aging process. The observed higher content of melanin relative to the lipofuscin content in the rabbit brain was associated with the occurrence of neurodegenerative processes that occur in the brain. Considering both healthy and cancer tissues from the human colorectal mucosa and kidney, it was possible to detect the formation of melanolipofuscin granules as a biomarker of cancer development. Other cancer biomarkers, such as higher blood content and higher mobile water content were also identified from the analysis of the spectral optical properties and application of optical clearing treatments. By using diffuse reflectance spectral measurements in the development of machine learning algorithms it was possible to reconstruct the absorption spectra of human healthy and diseased tissues with good accuracy, which leaded to the recovery of similar results about the blood and pigment contents in each case.

Fibre optic sensing to inform underground construction operations
Dr Brian Sheil, Laing O’Rourke Associate Professor in Construction Engineering, University of Cambridge
Abstract:
Uncertainty surrounding how structures interact with soil remains a key barrier to project optimisation on the basis of cost, time and emissions. However, robust and reliable sensors to measure soil pressures during underground construction do not exist. This presentation describes the development of a new more sophisticated generation of force and pressure sensors by fusing fibre optic sensing with machine learning. The deployment of prototypes on a recent UK construction project and the corresponding impact is also discussed.

Universal light encoders: artificial intelligence hardware for machine vision, sensing, and universal metrology
Prof Andrea Fratalocchi, KAUST University
Abstract:
In this invited talk, I will summarize recent research results in the field of universal light encoders. These components represent a particular class of metasurfaces implementing optical hardware feedforward neural networks that can universally approximate any user-defined input-output function. In the first series of applications, I will discuss the exploitation of this technology to implement ultra-flat (60 nm thick) optical components for vectorial light control with near-unity experimental efficiencies in the visible. In the second part, I will discuss a new area of research that focuses on the implementation in the hardware of software machine learning concepts for various applications in machine vision. I will summarize the fundamentals of HyplexTM, a new hyperspectral technology platform that allows the acquisition and processing of high-resolution videos in real-time for a broad spectrum of applications, including remote sensing, medical diagnosis, precision agriculture, and security.