Machine learning based regression of complex coefficients of a linear combination of spatial modes from multimode optical fiber

1. A method for classifying higher-order spatial modes comprising:

providing a light beam emitted from an output of a multimode optical fiber, the light beam having two orthogonal polarization components, each including a linear combination of higher-order spatial modes;

separating the two orthogonal polarization components into two separate light beams;

recording digital images of the two separate light beams; and

classifying, using a machine learning based classifier, the recorded digital images.

2. The method of claim 1 further comprising training the classifier using numerically calculated training examples.

3. The method of claim 1 wherein the recording digital images is effected by a digital camera.

4. The method of claim 3 further comprising:

optically imaging the emitted light beam through the effect of imaging optics prior to the separation of the two orthogonal polarization components into two separate light beams.

5. The method of claim 4 wherein the imaging optics include one or more lenses.

6. The method of claim 5 wherein the multimode optical fiber and the digital camera are not placed in a focal region of the lenses.

7. The method of claim 3 wherein the camera records digital images of the transverse, spatial dependencies of the intensities of the two orthogonal polarization components of the light beam, whose alignment, size, and wavefront vary with respect to the digital camera.

8. The method of claim 7 wherein the center of an imaged light beam is displaced in the x-direction.

9. The method of claim 7 wherein the center of an imaged light beam is displaced in the y-direction.

10. The method of claim 7 wherein the direction of propagation of an imaged light beam forms an angle in the y-z plane.

11. The method of claim 7 wherein the direction of propagation of an imaged light beam forms an angle in the x-z plane.

12. The method of claim 7 wherein a waist size of an imaged light beam varies in time.

13. The method of claim 7 wherein a wave front of an imaged light beam exhibits an aberration.

14. The method of claim 1 wherein the higher order spatial modes are Hermite-Gaussian modes.

15. The method of claim 1 wherein the higher order spatial modes are Laguerre-Gaussian modes.

16. The method of claim 1 wherein the two orthogonal polarization components of the light beam are one selected from the group consisting of: horizontal/vertical, diagonal/antidiagonal, and right/left circular.

17. The method of claim 1 wherein the classifier is trained using numerically calculated training examples including varying alignments, sizes, and wave fronts.

18. The method of claim 17 wherein the numerically calculated training examples include higher order spatial modes having waists that vary.

19. The method of claim 17 wherein the numerically calculated training examples include at least one of: higher order spatial modes displaced in the x-direction, or higher order spatial modes displaced in the y-direction.