Researchers take control of the properties of light

Scientists can now customize the properties of broadband light using machine learning and an integrated photonic chip. Also known as "supercontinuum," these sources are at the center of new imaging innovations, and the methodology proposed by scientists will provide additional insights into the basics of light-matter collaborations and ultra-fast non-linear optics.

During the study, scientists were ready to design and control extreme ultrashort pulse designs, used to create a broadband optical spectrum. Recently, the improvement of laser sources, including exceptional and ultra-short laser pulses, led to the Nobel Prize for Physics in 2018, in parallel with the approaches to conserve and spatially guide light (fiber optics and waveguides), which allowed to go back to extremely powerful optical structures. .

With these new systems, a variety of potential outcomes are developing, for example the age of supercontinua, that is, expanded light spectra produced through serious light-matter connections.

These incredible and complex optical systems, along with their associated procedures, are shaping the fundamentals of all applications, from laser science to metrology, advanced detection methods and biomedical imaging.

To continue to push back the points of confinement of these advances, it is further necessary that the lighting properties are adapted. With this work, the universal research group discovers a functional and versatile answer to this problem.

Scientists have demonstrated that various models of femtosecond optical pulses can be prepared and handled with care.

Benjamin Wetzel (University of Sussex), principal investigator of this research led by Professors Roberto Morandotti (INRS) and Marco Peccianti (University of Sussex), said: "We have benefited from the compactness, stability and sub- nanometer resolution offered by integrated photonic structures to generate reconfigurable packets of ultra-short optical pulses. The exponential scaling of the parameter space obtained gives more than 1036 different configurations of impulse patterns that can be achieved, more than the estimated number of stars in the frame. ;universe. "

With such a large number of combinations to seed an optical system deemed deeply sensitive to the underlying conditions, analysts turned to an automatic learning system with the ultimate goal of examining the results of light control.

Specifically, they have demonstrated that the control and customization of output light is certainly effective when they jointly use their system and a suitable computation to study a large number of accessible light pulse designs used to fit complex physical dynamics.

These exciting results will have an impact on basic research as well as applied research in a number of areas, as much of today's optical systems use the same physical and nonlinear effects as those underlying supercontinuum generation.

The work should also support the development of other intelligent optical systems through self-optimization techniques, including the control of optical frequency combs (Nobel 2005) for metrology applications, self-adjusting lasers, the treatment and amplification of impulses (Nobel 2018). as the implementation of more fundamental approaches to machine learning, such as photonic neural network systems.