點樣prove 到In(exp(-x^2/2))=sqrt(2*pi())?

2008-07-24 6:26 pm
In=Integration
Domain 係由-infinity 到+infinity
簡單啲講,就係點prove integrate normal distribution係等於1?

回答 (2)

2008-08-01 3:30 am
✔ 最佳答案
你識唔識double integral先?識唔識jacobian?

[In(exp(-x^2/2))dx]^2
=[In(exp(-x^2/2))dx][In(exp(-x^2/2))dx]
=[In(exp(-x^2/2))dx][In(exp(-y^2/2))dy]
=InIn(exp(-x^2/2)exp(-y^2/2))dxdy
=InIn[exp(-(x^2+y^2)/2))]dxdy

Then use the change of variable x= r cos(s), y= r sin(s).
The jacobian of the change of variable is
(其實唔係dx/dr, 係 partial differentiation)
| (dx/dr)(dy/ds)-(dx/ds)(dy/dr) | = r

And the integration limit for r is from 0 to infinity and for s is 0 to 2*pi
because the integration domain is the whole R^2 plane.

Then
InIn[exp(-(x^2+y^2)/2))]dxdy
= InIn[exp(-r^2/2))]rdrds
= In{In[exp(-r^2/2))] d(r^2/2)}ds (again change of single variable)
= In{ -{ lim(r->infinity)[exp(-r^2/2)] - exp(-0^2/2)}} ds
= In(1)ds
= 2*pi

So [In(exp(-x^2/2))dx]^2=2*pi

Hence, In(exp(-x^2/2))=sqrt(2*pi)
2008-07-27 4:53 am
做乜問兩次?


收錄日期: 2021-04-23 18:03:18
原文連結 [永久失效]:
https://hk.answers.yahoo.com/question/index?qid=20080724000051KK00635

檢視 Wayback Machine 備份