Structured light,where light is tailored in all its degrees of freedom,has shown tremendous power in unlocking new modalities of light,with its impact felt across dimensions,disciplines,and applications.This richly te...Structured light,where light is tailored in all its degrees of freedom,has shown tremendous power in unlocking new modalities of light,with its impact felt across dimensions,disciplines,and applications.This richly textured light comes with deeply embedded complexity,making the design,analysis,and recognition of such complex light patterns highly non-trivial.In recent years artificial intelligence(AI)has come to the fore,not only for the design,characterization,and optimization of structured light but also for increasingly important roles in adding new functionalities and breaking old paradigms.An exciting twist is the flip side of the coin,where complex light in complex media acts as a light-speed neural network,ushering in a new era of ultrafast optical-based“machines”for intelligence and learning.In this review,we focus on how AI has enhanced structured light technologies,and vice versa,touching on imaging,microscopy,sensing,communication,and optical neural networks as topical application areas,while covering scales from the macroscopic to the microscopic,and from classical to quantum.We highlight the symbiotic relationship between intelligence and light in these processes and offer a perspective on the open challenges and future prospects of this emerging research direction.展开更多
Maxwell's demon(MD)has proven to be an instructive vehicle for exploring the relationship between infor-mation theory and thermodynamics.A long-standing debate has been the concern of entropy violation,now resolve...Maxwell's demon(MD)has proven to be an instructive vehicle for exploring the relationship between infor-mation theory and thermodynamics.A long-standing debate has been the concern of entropy violation,now resolved by the introduction of a quantum MD that can enact reversible operations on a system.However,implementing it experimentally is challenging,as it demands precise control over multi-particle entangled states and the execution of entangling and disentangling operations with high accuracy.Here,we show how this can be emulated using vectorial structured light that is nonseparable in the spin and orbital angular momentum(OAM)internal degrees of freedom of each photon in a classical laser beam.We experimentally demonstrate that the demon's classical entropy,linked to the uncertainty in spin degree of freedom of each photon,increases during the process while that in the system's state(represented by OAM per photon)de-creases.This is achieved by entangling the demon's memory with the system,allowing the demon to acquire quantum information and utilize it to control the OAM states of the system after a disentangling operation.As a result,we demonstrate that the quantum demon can emulate the extraction of useful work from the system in the form of OAM,thereby opening a path to information-driven optical spanners for the mechanical rotation of objects using light.Our demonstration can easily be extrapolated to other degrees of freedom,for robust and scalable implementations of MDs at both the classical and quantum realms,enlightening the role of a struc-tured light as a tool for exploiting principles in thermodynamics to control and measure information.展开更多
Transparent objects are invisible to traditional cameras because they can only detect intensity fluctuations,necessitating the need for interferometry followed by computationally intensive digital image processing.Now...Transparent objects are invisible to traditional cameras because they can only detect intensity fluctuations,necessitating the need for interferometry followed by computationally intensive digital image processing.Now it is shown that the necessary transformations can be performed optically by combining machine learning and diffractive optics,for a direct in-situ measurement of transparent objects with conventional cameras.展开更多
基金National Natural Science Foundation of China(62375015)National Natural Science Foundation of China(U23A20481,62275010)+7 种基金Open Fund of the State Key Laboratory of Precision Space–time Information Sensing Technology No.STL2023-B-03-01(J)Fundamental Research Funds for the Central Universities(2025CX11009)South African Quantum Technology Initiative(SAQuTI),OpticaNanyang Assistant Professorship Start Up Grant,Singapore Ministry of Education(MOE)AcRF Tier 1 grant(RG157/23)MoE AcRF Tier 1 Thematic grant(RT11/23)Imperial-Nanyang Technological University Collaboration Fund(INCF-2024-007)Nanyang Technological University SPMS Collaborative Research Award 2024Singapore Agency for Science,Technology and Research(A*STAR)MTC Individual Research Grants(M24N7c0080).
文摘Structured light,where light is tailored in all its degrees of freedom,has shown tremendous power in unlocking new modalities of light,with its impact felt across dimensions,disciplines,and applications.This richly textured light comes with deeply embedded complexity,making the design,analysis,and recognition of such complex light patterns highly non-trivial.In recent years artificial intelligence(AI)has come to the fore,not only for the design,characterization,and optimization of structured light but also for increasingly important roles in adding new functionalities and breaking old paradigms.An exciting twist is the flip side of the coin,where complex light in complex media acts as a light-speed neural network,ushering in a new era of ultrafast optical-based“machines”for intelligence and learning.In this review,we focus on how AI has enhanced structured light technologies,and vice versa,touching on imaging,microscopy,sensing,communication,and optical neural networks as topical application areas,while covering scales from the macroscopic to the microscopic,and from classical to quantum.We highlight the symbiotic relationship between intelligence and light in these processes and offer a perspective on the open challenges and future prospects of this emerging research direction.
基金South African Quantum Technology Initiative(SA QuTI)Department of Science and Innovation(South Africa)+2 种基金National Research Foundation(South Africa)CIOSECIHTI(formerly CONAHCYT).
文摘Maxwell's demon(MD)has proven to be an instructive vehicle for exploring the relationship between infor-mation theory and thermodynamics.A long-standing debate has been the concern of entropy violation,now resolved by the introduction of a quantum MD that can enact reversible operations on a system.However,implementing it experimentally is challenging,as it demands precise control over multi-particle entangled states and the execution of entangling and disentangling operations with high accuracy.Here,we show how this can be emulated using vectorial structured light that is nonseparable in the spin and orbital angular momentum(OAM)internal degrees of freedom of each photon in a classical laser beam.We experimentally demonstrate that the demon's classical entropy,linked to the uncertainty in spin degree of freedom of each photon,increases during the process while that in the system's state(represented by OAM per photon)de-creases.This is achieved by entangling the demon's memory with the system,allowing the demon to acquire quantum information and utilize it to control the OAM states of the system after a disentangling operation.As a result,we demonstrate that the quantum demon can emulate the extraction of useful work from the system in the form of OAM,thereby opening a path to information-driven optical spanners for the mechanical rotation of objects using light.Our demonstration can easily be extrapolated to other degrees of freedom,for robust and scalable implementations of MDs at both the classical and quantum realms,enlightening the role of a struc-tured light as a tool for exploiting principles in thermodynamics to control and measure information.
文摘Transparent objects are invisible to traditional cameras because they can only detect intensity fluctuations,necessitating the need for interferometry followed by computationally intensive digital image processing.Now it is shown that the necessary transformations can be performed optically by combining machine learning and diffractive optics,for a direct in-situ measurement of transparent objects with conventional cameras.
