1 / 17

Fitting (special modeling)

Fitting (special modeling). 预习 BR2003, Chap. 11. xbdong@ustc.edu.cn 2009.12.09. 通知. 《 天文数据分析方法 》09 考核方法 平时成绩   40%: 作业 + 案例讨论表现; Project    40%: 书面报告 + 口头报告(分组); 开卷考试  20% . 郑永飞院士如是说.

Download Presentation

Fitting (special modeling)

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Fitting (special modeling) 预习BR2003, Chap. 11 xbdong@ustc.edu.cn 2009.12.09

  2. 通知 《天文数据分析方法》09考核方法 • 平时成绩  40%: 作业 + 案例讨论表现; • Project    40%: 书面报告 + 口头报告(分组); • 开卷考试  20% 

  3. 郑永飞院士如是说 • “这个‘油炸冰淇淋’模型只是理论推断,当时属于科学假说。要使科学假说上升成为科学理论,必须经过科学验证和时间检验。” 郑永飞说,自己不去做实验,做研究肯定深化不了,只能是从假说到假说,既成不了科学模型,更上升不了科学理论,别人也不会信服。 • “重要的实验必须自己去参加,重要数据的处理必须自己参与分析,而且全过程都应该是一丝不苟,这样得来的数据才直观,才有感性认识,对解决的科学问题才有把握。” 郑永飞如是说。 http://news.ustc.edu.cn/?article=00021553

  4. Overview of Chaps.8,9 • P.145, Method of Least squares Eqn.(8.5), Fig. 8.3 • P.148, Searching parameter space starting values, step sizes, bounds, searching methods (cf. Fig.8.4) • Direct, brute-force method; MPFIT package; forget those searching algorithms. • It is ok to skip Chap. 9 totally.

  5. Question: How to find the minimum? (i.e. the searching method) The answer: 八仙过海,各显神通。

  6. Comments of last case-discussion • Must be familiar with svdfit (or mpfit). • Must know how to synthesize the total measurement errors. • As brave/ambitious as possible: “just do it!” – the easiest, brute-force and powerful method.

  7. Overview of Chap.11 • $11.1, only Eqn. (11.6), and from Eqn. (11.10) through the end (p.196-197). • Skip $11.2, $11.3 totally. • $11.4: next class. • $11.5: important.

  8. Prob. density distr. FunctionEqn. 11.6

  9. Eqn. 11.10

  10. Essence 1: chisq test显著性水平的含义 Cf. 陈希孺书p.262-264 • 检验什么? 检验:是否model不符合数据? [p.197; 参考p.70] • 三句话理解chisq概率(显著性水平): 1. 假设model符合data,那么由于测量不确定性,多次实验的统计量zmin将符合自由度n-k的卡方分布(Eqn. 11.6)。 2. 现我们在一次实验中得到的z0min;而由上我们知道如果多次实验的话,由于测量不确定性zmin大于或等于z0min的概率P(z>=z0min; n-k)可以根据(Eqn. 11.6)算出。 3. 那么,如果这个概率很大,说明在这次实验得到的z0min是很正常的(大概率事件);如果这个概率很小而竟然出现在我们的一次实验中了(小概率事件出现了!) ,这表明我们的假设很可能错了,显著性水平就是这个概率P(z>=z0min; n-k) 。

  11. $11.5 confidence intervals on estimated model parameters • Confidence for what? 对拟合参数值的信心。 Const. a2: 对于每一个a2值,通过最小化chisq来确定其它参数最佳值。

  12. Essence 2 Question: “四大定理”成立的前提是什么? k k k k k k k

  13. background

  14. Question: 分别在什么情况下采用单个参数的 置信区间、联合参数的置信区间?

  15. Case discussion 箴言 • “首先投入战斗,然后再见分晓。” (拿破仑,列宁;即是两湖地区俗语“老子搞了再说”的意思。) • 如載一車兵器,逐件取出來弄,弄了一件又弄一件,便不是殺人手段。我只有寸鐵,便可殺人。 ——宋·禅师宗杲 。 朱熹引(《朱子语类·卷八·学二·总论为学之方》)

  16. Assignment for next class • 上次的case必须人人过关:调通程序,并现场运行演示。下次课检查! • Self-study section 11.4 of BR2003 book, and Section 15.6 of Press etal. Book. 透彻理解“四大定理”。 • 熟悉IDL中的chisqr_pdf, chisqr_cvf, f_cvf等等函数,并写小代码生成BR2003书中的Table C.4, Figure C.4, Table C.5, Figure C.5, Table C.6, Figure C.6, Table C.7, Figure C.7 。(Matlab等也可。)

More Related