|WEPAF041||Use of Dimension-Reduction Techniques With Multi-Objective Genetic Algorithms to Improve the Vertical Emittance and Orbit at CESR||1901|
|SUSPL064||use link to see paper's listing under its alternate paper code|
Funding: DOE DE-SC0013571 NSF DGE-1650441
In order to reduce the vertical emittance at the Cornell Electron Storage Ring (CESR), we first measure and correct the vertical orbit, dispersion, and coupling. However, due to the finite resolution of our optics measurements, we still retain a significant residual emittance. In order to correct this further, we made use of the theory of sloppy models, according to which certain high-dimensionality systems can be modeled with significantly fewer "eigenparameters" that still contain most of the effect on the desired objective, in this case, the emittance.* However, we noted that using these knobs for tuning often resulted in increased vertical orbit errors. In an attempt to constrain these, we have applied multi-objective genetic algorithms to this problem. We have found that it can be more efficient to run such algorithms using our eigenparameters as the genes to be varied, as opposed to the raw magnet values. When running with the first 8 such knobs as genes, we can get either orbits or beam sizes as good as we obtain with our regular emittance-tuning algorithm which uses all the corrector magnets.
*K.S. Brown and J.P. Sethna, Phys. Rev. E 68, 021904 (2003).
|DOI •||reference for this paper ※ https://doi.org/10.18429/JACoW-IPAC2018-WEPAF041|
|Export •||reference for this paper using ※ BibTeX, ※ LaTeX, ※ Text/Word, ※ RIS, ※ EndNote (xml)|