Vietnam Quantitative Finance Society

User Name
Password
Go Back   Vietnam Quantitative Finance Society > Forums List > Interview Questions
Reply
10-18-2008, 11:21 PM   #101
shinichi9htv

Member
Sergeant
No Avatar
 
Join Date: Jul 2007
Posts: 148
Thanks: 13
Thanked 7 Times in 6 Posts




Default

nice post bro!
30/ you can define it from a complex error function
31/ Excellent! Now, let's compute

32/ E(|W_t|*N(a*W_t^2)
shinichi9htv is offline Reply With Quote
10-18-2008, 11:30 PM   #102
nguyenxuanson

Core Member
Corporal
No Avatar
 
Join Date: Aug 2007
Posts: 58
Thanks: 1
Thanked 1 Time in 1 Post




Default

hi Quan
30/ Error function is defined on real line, you answer has no sens becuase the way you prolong it will define the "complex error function". I don't know this function and do not know how you problong it i.e which are propert of this new abstract function? If you define it by the integration representation, it does not converge, you have to do something....
nguyenxuanson is offline Reply With Quote
10-19-2008, 12:31 AM   #103
shinichi9htv

Member
Sergeant
No Avatar
 
Join Date: Jul 2007
Posts: 148
Thanks: 13
Thanked 7 Times in 6 Posts




Default

hi Son
I think you're confused between the error function and the normal cdf. The error function is defined by 2/sqrt(PI)int_0^z exp(-y^2) dy, hence it's well defined in the complex plane. Hope it helps
shinichi9htv is offline Reply With Quote
10-19-2008, 12:46 AM   #104
shinichi9htv

Member
Sergeant
No Avatar
 
Join Date: Jul 2007
Posts: 148
Thanks: 13
Thanked 7 Times in 6 Posts




Default

you're absolutely right Son, forget about my non-sense question
shinichi9htv is offline Reply With Quote
10-19-2008, 01:03 AM   #105
nguyenxuanson

Core Member
Corporal
No Avatar
 
Join Date: Aug 2007
Posts: 58
Thanks: 1
Thanked 1 Time in 1 Post




Default

I google and found only complex prolongation for erf(x) and it is defined in http://en.wikipedia.org/wiki/Error_function If you says:N(x)=1/2(1+\erf(x/\sqrt{2}) then you can defined the prolongation of N(x) to the complex plan and you can easily have the formula for it, something proportional to erf(1/sqrt(2))..

PS: in which context you need to evaluate N(i)?
nguyenxuanson is offline Reply With Quote
10-19-2008, 01:54 AM   #106
nguyenxuanson

Core Member
Corporal
No Avatar
 
Join Date: Aug 2007
Posts: 58
Thanks: 1
Thanked 1 Time in 1 Post




Default

32/ not very smart solution
a/ check that the expectation is well defined (easy becasue the N(x) is bounded)
b/ express N(x) by erf(x)
c/ erf(x) has well defined Taylor expansion, on the whole real line
d/ integrate each term...easy part
e/ sum it up, you may simplify it somehow, have not checked

If you have smart solution, let me know
Son
nguyenxuanson is offline Reply With Quote
10-19-2008, 04:28 AM   #107
shinichi9htv

Member
Sergeant
No Avatar
 
Join Date: Jul 2007
Posts: 148
Thanks: 13
Thanked 7 Times in 6 Posts




Default

Thanks Son for your solution,

I believe the solution is compact and beautiful (haven't checked either ). If you could give us your final result, that would be great
shinichi9htv is offline Reply With Quote
10-19-2008, 03:43 PM   #108
nguyenxuanson

Core Member
Corporal
No Avatar
 
Join Date: Aug 2007
Posts: 58
Thanks: 1
Thanked 1 Time in 1 Post




Default

integral by part is the other way, but i do not bother to do it as well. But what happen if we need to calculate
E(N(aW^{2}_{t}). Can we still do it?
nguyenxuanson is offline Reply With Quote
10-19-2008, 06:37 PM   #109
shinichi9htv

Member
Sergeant
No Avatar
 
Join Date: Jul 2007
Posts: 148
Thanks: 13
Thanked 7 Times in 6 Posts




Default

One powerful method to solve this kind of problem is to use Itô lemma. (The key idea is that the first and second derivatives of N(x) are simple and easy to integrate) Once the solution is obtained, there might be a simpler solution to come
shinichi9htv is offline Reply With Quote
Reply


Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump

Save or bookmark this site with: Del.icio.usStumble It!AddThis.com
All times are GMT +7. The time now is 06:25 AM.
Powered by vBulletin Version 3.7.2