<t>say you have three nodes at the first time: A1, B1 and C1 which go over in the nodes A2, B2, and C2.Then define the transition probabilities of going from node A1->A2 as p(A1,A2).Say A1 is two-dimensional A1 = (x1,y1), A2 = (x2,y2), then p(A1,A2) = P[ (X2,Y2) = (x2,y2) | (X1,Y1) = (x1,y1) ] ('Biv...

<r>A few practical tips:- As the derivative in the root finding procedure for y can be easily calculated, a Newton-Raphson procedure seems the most obvious.- Note that the integral is actually a bounded function against a normal distribution (make a plot of the integrand and you'll be surprised how ...

The paper of Lord and Kahl or the paper of Schoutens might help.see http://ssrn.com/abstract=1105998 or http://perswww.kuleuven.be/~u0009713/HestonTrap.pdf

I already suspected that ;-)Perhaps you can use the fact that both the zero-coupon bond option as well as the stock price follows a log-normal distribution which is equivalent to a two-stock Black-Scholes economy (with correlated assets) ...

Can't you use RN valuation and use exp[-int r(u) du] (i.e. one over the money market account) as stochastic discount factor ? \edit typo

<t>No problem, I was also confused initially and it took me some time to figure it out. Indeed the restriction is very odd and unnecessary. I can only guess for the reason behind it, but it might be that the resulting ODE's become a little bit messier ... anyway the author could have been more clear...

<r>QuoteOriginally posted by: piterbargjust use this:<URL url="http://papers.ssrn.com/sol3/papers.cfm?abstract_id=946405that"><LINK_TEXT text="http://papers.ssrn.com/sol3/papers.cfm? ... 946405that">http://papers.ssrn.com/sol3/papers.cfm?abstract_id=946405that</LINK_TEXT></URL> should sort you outVV...

<r>QuoteOriginally posted by: piterbargare you sure abut that? that was not my impression at all. Perhaps the vector notations confused you? I shall go back and check-V- First of all, imho, the author is not really explicit about the correlation assumptions, so one has to distill it from the formula...

<r>QuoteOriginally posted by: piterbargSomebody just published the same in a journalhttp://<URL url="http://www.worldscinet.com/cgi-bin/details.cgi?id=voliss:ijtaf_1103&type=tocInternational"><LINK_TEXT text="www.worldscinet.com/cgi-bin/details.cgi ... ernational">www.worldscinet.com/cgi-bin/det...

Ok thanks: that is pretty clear ... since it might indeed be hard to align the variates, JoeKuo07 might indeed be the better option if these results hold in general.

<t>QuoteOriginally posted by: bluffbodyNumerically speaking I experience difficulties to compute the closed-form characteristic function (of the domestic T-forward log FX-rate) of the HWHWSchobel-Zhu model. Problems come of course from the hypergeometric functions contained in the A(t) coefficient o...

<r>The built-in function lsqnonlin (available in the optimization toolbox) can also use the LM-algorithm.See <URL url="http://www.mathworks.com/access/helpdesk/help/toolbox/optim/index.html?/access/helpdesk/help/toolbox/optim/ug/lsqnonlin.html"><LINK_TEXT text="http://www.mathworks.com/access/helpde...

mj,how do the newly added initialization numbers (in QL-SobolGenerator) of Joe and Kuo compared those of Jäckel, is this difference noticeable ?

<r>Interesting results Vladimir, which I was not aware off: I have only calibrated the model to equity quotes and not FX ones. For the equity market the SZ-HW model can fit the markt quite well (similar to Heston), but I will need to test it on FX quotes.Anyway, we are a bit in parallel (or off-topi...

<t>The distribution of the variance in the Heston model is Non-central Chi-squared distributed with (usually) a small degrees of freedom, while the distribution of the SZ model is distributed as an ordinary Chi-Squared distribution. Then your statement would imply that for high vol of variance the n...