| 題名: | Efficient scheme for parametric fitting of data in arbitrary dimensions |
| 作者: | Pang, Ning-Ning;Tzeng, Wen-Jer;Kao, Hisen-Ching |
| 貢獻者: | 淡江大學物理學系 |
| 日期: | 2008 |
| 上傳時間: | 2013-07-09 15:20:10 (UTC+8) |
| 出版者: | College Park: American Physical Society |
| 摘要: | We propose an efficient scheme for parametric fitting expressed in terms of the Legendre polynomials. For continuous systems, our scheme is exact and the derived explicit expression is very helpful for further analytical studies. For discrete systems, our scheme is almost as accurate as the method of singular value decomposition. Through a few numerical examples, we show that our algorithm costs much less CPU time and memory space than the method of singular value decomposition. Thus, our algorithm is very suitable for a large amount of data fitting. In addition, the proposed scheme can also be used to extract the global structure of fluctuating systems. We then derive the exact relation between the correlation function and the detrended variance function of fluctuating systems in arbitrary dimensions and give a general scaling analysis. |
| 關聯: | Physical Review E 78(1), 011112(11pages) |
| DOI: | 10.1103/PhysRevE.78.011112 |
| 顯示於類別: | [物理學系暨研究所] 期刊論文
|