On-demand orbital angular momentum modes through hollow-core multimode fiber via a neural network model

We present a machine-learning approach to generate and dynamically reconfigure orbital angular momentum (OAM) beams in a multimode inhibited-coupling hollow-core photonic crystal fiber (IC-HCPCF). To achieve this, we trained a neural network-based digital twin for a 139 µm core IC-HCPCF guiding over 60 LP-like modes per polarization in the green spectral range. The accuracy of the neural network was confirmed by the high fidelity between experimental and predicted output intensity patterns in Fourier-conjugate planes, with median Pearson correlation coefficients exceeding 98%. These results highlight multimode IC-HCPCFs as a versatile platform for structured light transport and on-demand beam shaping.