Machine learning a manifold
https://arxiv.org/pdf/2112.07673.pdf
Abstract: We propose a simple method to identify a continuous Lie algebra symmetry in a dataset through regression by an artificial neural network. Our proposal takes advantage of the O(ε2) scaling of the output variable under infinitesimal symmetry transformations on the input variables. As symmetry transformations are generated post-training, the methodology does not rely on sampling of the full representation space or binning of the dataset, and the possibility of false identification is minimised. We demonstrate our method in the SU(3)-symmetric (non-) linear Σ model.
It is quite interesting for me that the machine learning can help on identifying underlying symmetry. This topic brings me the question raised by my supervisor how to confirm the symmetry of a dynamics, for example SU(3)_color.
Using Covariant Polarisation Sums in QCD
Abstract: Covariant gauges lead to spurious, non-physical polarisation states of gauge bosons. In QED, the use of the Feynman gauge, ε(λ)ε(λ)∗ = −ημν, is justified by the Ward identity which ensures that the contributions of non-physical polarisation states cancel in physical observables. In contrast, the same replace- ment can be applied only to a single external gauge boson in squared amplitudes of non-abelian gauge theories like QCD. In general, the use of this replacement re- quires to include external Faddeev-Popov ghosts. We present a pedagogical deriva- tion of these ghost contributions applying the optical theorem and the Cutkosky cutting rules. We find that the resulting cross terms A(c1,c ̄1;…)A(c ̄1,c1;…)∗ be- tween ghost amplitudes cannot be transformed into (−1)n/2|A(c1, c ̄1; . . .)|2 in the case of more than two ghosts. Thus the Feynman rule stated in the literature holds only for two external ghosts, while it is in general incorrect.