Fig. 1. Multiple dimensions of light

Fig. 2. Schematic of prediction based on artificial neural network

Fig. 3. Intelligent laser with low delay depth reinforcement learning algorithm based on depth deterministic strategy gradient^[53]. (a) Algorithm; (b) experimental setup

Fig. 4. Configuration of self-optimizing mode-locked laser^[48]

Fig. 5. Self-tuning laser^[54]. (a) Experimental setup; (b) schematic of phase delay due to fiber birefringence

Fig. 6. Self-optimization of breathing soliton in mode-locked fiber lasers^[56]. (a) Experimental setup; (b) flow chart of evolutionary algorithm

Fig. 7. Intelligent mode-locked fiber laser with embedded time-stretch assisted real-time pulse controller^[58]

Fig. 8. Multi-dimensional fiber laser based on wavefront shaping^[69]

Fig. 9. Trapping potential and force distributions in nonlinear regime for gold nanoparticle^[77]. (a)-(c) Trapping potential distributions on trapping plane; (d)-(f) spatial variation of gradient force on trapping plane; (g)-(i) distributions of force at 200 nm behind trapping plane

Fig. 10. Experimental setups for optical field characterization and pulse compensation^[78]

Fig. 11. Experimental setup of dissipative solitons in multidimensional systems^[80]. (a) Spatio-temporal-spectral laser dynamics; (b) schematic diagram of spatio-temporal-spectral compressed ultrafast photography (STS-CUP)

Fig. 12. Control of spatiotemporal nonlinear frequency conversion in multimode fibers^[81]. (a) Experimental setup; (b) experimental training control principle

Fig. 13. Obtaining training data consisting of input phase modulation images and output speckle images through multi-mode fiber^[89]. (a) Experimental setup; (b) optical speckle image of face phase coded image reconstructed by neural network

Fig. 14. Force field analysis of COVBs and time-lapse images of particles moving along cycloidal tracks^[100]. (a1)-(d1) Experimental intensity patterns of COVBs under different n when m is constant; (a2)-(d2) enlarged gradient force at box in Figs. 14 (a1)-(d1); (a3)-(d3) enlarged OAM at box in Figs. 14 (a1)-(d1); (a4)-(d4) delay images of particles along COVBs trajectory corresponding to Figs. 14 (a1)-(d1)

Fig. 15. Dynamics of particles in Laguerre-Gaussian beam superposition^[107]. (a) Experimental diagram; (b) particle rotation rates obtained from experiment and simulation results based on neural network; (c) experimental trajectories of particles and resulting limiting potential; (d) simulated trajectories corresponding to Fig. 15(c); (e) average of 100 simulated trajectories

Fig. 16. Schematics of elliptic focus locus formation^[111]. (a)-(d) Polarization distributions and periodic changes of constructed light field when n' is 1, 13, 25, and 37, respectively; (e)-(h) focal field distributions corresponding to focal planes in Figs. 16 (a)-(d); (i)-(l) corresponding focus locus in Figs. 16 (a)-(d)

Fig. 17. Laser machining results^[112]

Fig. 18. OAM beams produced by DOVG^[119]. (a) Planar phase wavefront Gaussian beams are diffracted by DOVG and coded as OAM; (b) multiple Gaussian beams incident in multiple diffraction directions are combined in zero-order direction to form optical vortex beams with multiple OAM states; (c) OAM beams are diffracted by same DOVG and converted back to Gaussian beams in corresponding direction

Fig. 19. Image transmission based on deep learning data transmission system^[128]