Compression and noise reduction of field maps

Errors from discretization and large data volume of field maps is a concern for beam dynamics simulations with respect to achievable accuracy and to the required amount of time. High-order singular value decomposition (HOSVD) has recently emerged as simple, effective, and adaptive tool to extract the essentials from multidimensional data. This paper is on the feasibility of compression and noise reduction of electromagnetic field map data with HOSVD. The method has been applied to an electric field map of a DTL cavity with 11 m in length comprising 55 rf-gaps. The original field map data of 220 MB was converted into practically noise-free data of just 20 KB. Noise was reduced by 95% as demonstrated using a cubic cavity for which the analytical field map is available.