Broadband mode converters are essential devices for space-division and wavelength-division multiplexing systems. There are great challenges in the generation of higher-order modes above the third order with low loss and high mode purity employing all-fiber devices. In this paper, an all-fiber LP₄₁ mode converter is proposed and fabricated by tapering a nine-core single-mode fiber bundle. Experimental results indicate that this all-fiber LP₄₁ mode converter is low-loss, high-purity, and ultrabroadband. The insertion loss is less than 0.4 dB. The purity of odd LP₄₁ at 1310 nm is 95.09%, and the operating bandwidth exceeds 280 nm.

In this article, we use a convolutional autoencoder neural network to reduce data dimensioning and rebuild soliton dynamics in a passively mode-locked fiber laser. Based on the particle characteristic in double solitons and triple solitons interactions, we found that there is a strict correspondence between the number of minimum compression parameters and the number of independent parameters of soliton interaction. This shows that our network effectively coarsens the high-dimensional data in nonlinear systems. Our work not only introduces new prospects for the laser self-optimization algorithm, but also brings new insights into the modeling of nonlinear systems and description of soliton interactions.

We report a regime of the loose soliton bunch in an erbium-doped passively mode-locked fiber laser. In this state, every soliton bunch consists of multiple pulses. The amount of multiple pulses inside the soliton bunch increase as the pump power rises. Moreover, the temporal average pulse-to-pulse separation decreases in general with the increase of the pump power. Further, the spatial-temporal sequences based on the dispersive Fourier transformation technique show that pulse-to-pulse interactions and time jitter can result in pulse forking inside the soliton bunch. Finally, we theoretically demonstrate the soliton bunch with different pulse-to-pulse separations.