 Martingale
|
| NP House Mouse |
| Total Posts: 2527 |
| Joined: Jun 2004 |
| |
|
Consider the following C program for producing Fibonacci numbers
Int Fibonacci(int n)
{ if (n <=011 n==1) return 1;
else return Fibonacci(n-1)+Fibonacci(n-2);}
If for some large n, it takes 100 seconds to compute Fibonacci(n), how long will it take to compute Fibonacci(n+1), to the nearest second? |
|
|
|
 |
 Energetic
|
| Forum Captain |
| Total Posts: 1381 |
| Joined: Jun 2004 |
| |
|
| 162 seconds. |
Quote me as saying I was misquoted. |
|
 |
 Martingale
|
| NP House Mouse |
| Total Posts: 2527 |
| Joined: Jun 2004 |
| |
|
| Kinda of (1+sqrt(5))/2 * 100 , right? |
|
|
|
 |
 Martingale
|
| NP House Mouse |
| Total Posts: 2527 |
| Joined: Jun 2004 |
| |
|
| what about this one? Suppose that 2 cylinders each with radius 1 intersect at right angles and their centers also intersect. What is the volume of the intersection ? |
|
|
 |
 Energetic
|
| Forum Captain |
| Total Posts: 1381 |
| Joined: Jun 2004 |
| |
|
a. You got it.
b. I certainly was able to solve this stuff in high school but now I'm afraid to get a headache in the process. Run a MC simulation  |
Quote me as saying I was misquoted. |
|
|
 |
 drews
|
|
| Total Posts: 276 |
| Joined: Jun 2004 |
| |
|
Intersecting cyllinders problem can be solved by a very simple integration, google for 'Steinmetz solid'. Crossections of that object by 3 perpendicular planes are: 2 circles and a square |
|
|
 |
 Martingale
|
| NP House Mouse |
| Total Posts: 2527 |
| Joined: Jun 2004 |
| |
|
| Here is another one, consider X_t = m*t+ B_t where B is a standard Brownian motion, what's the probability of X_t hit 3 before it hit -5? I know a very nice way of this for m=0, but for m<>0, is there a elegant answer? |
|
|
|
 |
 chiral3
|
| Founding Member |
| Total Posts: 4345 |
| Joined: Mar 2004 |
| |
|
| Martingale, that one was all you, considering your background. Is the elegant way doob? |
What do you mean by **drag** the range"? If you just mean expand the column width, that doesn't work. NUM! |
|
 |
 Martingale
|
| NP House Mouse |
| Total Posts: 2527 |
| Joined: Jun 2004 |
| |
|
| Yes, Doob' way, for m=0 you have a neat answer, but what about for m<>0? |
|
|
|
 |
 Energetic
|
| Forum Captain |
| Total Posts: 1381 |
| Joined: Jun 2004 |
| |
|
| Now, for the rest of us, doob it out, pls. |
Quote me as saying I was misquoted. |
|
 |
 Martingale
|
| NP House Mouse |
| Total Posts: 2527 |
| Joined: Jun 2004 |
| |
|
Let's just stick to m=0 case where X_t is a standard BM.
Let tau =inf{t: X_t = 3 or X_t = -5}, tau1 = {t: X hit 3 before -5}, tau2={t: X hit -5 before 3}
Consider the martingale M_t = X_min{tau, t}, by Doob's sampling theorem E X_tau = X_0 =0. X_tau = 3 on {tau = tau1} and X_tau = -5 on {tau = tau2}; we get two equations
P{tau=tau1}+P(tau=tau2) = 1
E X_tau = 0 can be written as
3P{tau = tau1} +(-5)*P{tau=tau2} = 0
you get P{X hit 3 before -5} = P{tau =tau1} = 5/8. |
|
|
|
 |
 monkeyA
|
| Mr. Ass to you |
| Total Posts: 834 |
| Joined: Apr 2004 |
| |
|
| very neat! I like. |
If there was problem, Yo I'll solve it |
|
 |
 chiral3
|
| Founding Member |
| Total Posts: 4345 |
| Joined: Mar 2004 |
| |
|
| admittedly I don't know how to do the m<>0 case |
What do you mean by **drag** the range"? If you just mean expand the column width, that doesn't work. NUM! |
|
|
 |
 Martingale
|
| NP House Mouse |
| Total Posts: 2527 |
| Joined: Jun 2004 |
| |
|
| I am wondering if there's some neat way, but after all, we know the joint distribution of minimum, maximum and BM itself for some linear barrier, we can do something from there, just wonder if there's any other short cut............ |
|
|
 |
|
I remember seeing this (I think in mj's book) - the derivation took a few pages and involved 2 applications of Girsanov, one to remove the drift, one to put it back in.
Yeah the doob trick is neat, easy once you know how, and it did come up in an interview! |
|
|
|
 |
 Energetic
|
| Forum Captain |
| Total Posts: 1381 |
| Joined: Jun 2004 |
| |
|
Very neat!
It's a great problem for a take-home test or, maybe, for an open book test (though a big stretch, IMHO). But I'll tell you one thing. I don't want to work for a sonofabitch who throws in this sort of shit during regular interview. There can be no purpose other than to humiliate the candidate. |
Quote me as saying I was misquoted. |
|
 |
 Martingale
|
| NP House Mouse |
| Total Posts: 2527 |
| Joined: Jun 2004 |
| |
|
| I have thought about the Girsanov trick, yet to pull all things together. E, I was thown problems like this even at Phone interviews...oh well.... |
|
|
|
 |
 Energetic
|
| Forum Captain |
| Total Posts: 1381 |
| Joined: Jun 2004 |
| |
|
M, I'm serious: don't deal with the phuckers. You'll be unhappy with them.
|
Quote me as saying I was misquoted. |
|
 |
 Martingale
|
| NP House Mouse |
| Total Posts: 2527 |
| Joined: Jun 2004 |
| |
|
| I know.... in fact , I have spoke to quite some big names and they never torture people like this over the phone, but some phuckers enjoy this and they always wanted to kill you.... |
|
|
|
 |
 kr
|
| Founding MemberNP Raider |
| Total Posts: 3560 |
| Joined: Apr 2004 |
| |
|
I'm wondering if there is a clever way to do this using Fourier transforms... initially I thought one could quickly work out the case where you have to hit the barriers before a finite time T_0, then figure out how to piece a couple of these together. Then take the limit and see how the thing turns out. But for this one needs the distribution of starting points for the second segment conditional on staying in the channel - I think this is not too pleasant.
This is a problem I've been interested in b/c I was thinking that views on the trajectory of the underlying could be quickly translated into a kind of option valuation... I.e. guy says: I am 101% sure that X will go up 5% in the next 2 months. But, in the -1% chance that it falls by 10%, I'm gonna close out and eat my hat... Anyhow, translate these views into some ugly knockout option, which you're not actually going to buy, but which you are going to value in a risk-neutral way... i.e. Market says prob(this guy wins) is 50-50, he is a dumb analyst, don't listen to him. |
my bank got pwnd |
|
 |
 Martingale
|
| NP House Mouse |
| Total Posts: 2527 |
| Joined: Jun 2004 |
| |
| |
 |
 BigApple
|
|
| Total Posts: 1 |
| Joined: Oct 2004 |
| |
|
The driftful BM problem does have an elegant solution basically following the martingale method that Martingale outlined above. Just 3 or 4 lines more! Certainly not a few pages nor Girsanov theorem involved!
On my way to provide the reference or even a refined solution right here, but this whole thread sounds like being taken from a screening (take-home) test of Goldman Sachs. (Fibonacci, two cylinders, and this m0 BM one. Wouldn't be surprised if the best-of-seven game series question pops up here tomorrow.) For your knowledge, I wouldn't mind to tell you the details; but for your job interview where your true abilities are under test, it'll be unethical if I do so. Sorry... A close friend of mine was even let to struggle with these questions when he ran into Goldman, despite that I prepared him from his resume to interview practice. Anyhow, good luck!
Energetic, the Goldman test was take-home, although some sadists out there take great joy humiliating (out of revenge, perhaps???)
|
|
|
 |
 Patrik
|
| Founding Member |
| Total Posts: 1043 |
| Joined: Mar 2004 |
| |
|
Everyone seems to have stories about tougher-than-hell interviews with Goldie Sticks. Are they really that bad or is there an element of myth involved here?
From the stories I've read here and at the other place it sounds like they have a 100 people grilling candidates for days, without any food or water, ask hard things, etc. You almost get the feeling that they have a dark room and one of these very bright lights in your eyes etc - sort of like the strategies carried out by the bad guys in a movie when they are trying to get information from some poor hero 
|
Paper Trading - Capital Structure Demolition, LLC  |
|
|
 |
 DW
|
|
| Total Posts: 504 |
| Joined: Jun 2004 |
| |
|
| Having placed a number of guys in their quant groups I can say that all the stories are true! Their IRD and Credit groups seem to be slightly tougher than Equities. Also, there are members of this forum who have successfully made it through the hoops, perhaps they can shed some light on whether they thought it was tough or not. |
|
|
 |
 Johnny
|
| Founding Member |
| Total Posts: 4331 |
| Joined: May 2004 |
| |
|
Everyone seems to have stories about tougher-than-hell interviews with Goldie Sticks. Are they really that bad or is there an element of myth involved here?
I heard about a guy that had to spent his entire summer there from morning til night in a darkened room having to programme weird commodities sheaaa in a weird homemade script. Terrible. I don't know why anyone would put themselves through that. Maybe Patrik you heard a story like that too?
|
Stab Art Capital Structure Demolition LLC |
|
|
 |