Hello, I am trying to find the Expected Shortfall corresponding to a 95% confidence level on a lognormal distribution. Given x~LogNorm solve for Alpha. E[ x | x > LogNormInv( Alpha%) ] = LogNormInv(95%) Any idea?Many thanks

Cool thks.

Hi all, I just can't remember how to express 1+exp(-x)+exp(-2x)+exp(-3x)+...+exp(-nx) as a simple function of x.Any help?

<t> I chose to change the initial matrix for communication purpose. Its is difficult to explain what level of stress you are actually applying if you play with the eigen values/vectors. I managed to successfully use the rescaling method but I will definitely try the second one, that sounds indeed mo...

Note that I already did the Eigen vector decomposition.

By rotation matrix method, arerefering to the hyperspher decomposition i.e the first method proposed by this paper?What is a "random" rotation matrix?Why do you consider the rescalling mentioned previously as rubbish?Thks.

<r>Indeed the sum of the eigen values should be equal to the number of nows. But it is not sufficient. I found a paper explaining the method :<URL url="http://citeseer.ist.psu.edu/cache/papers/cs/25768/http:zSzzSzwww.rebonato.comzSzcorrelationmatrix.pdf/rebonato99most.pdfBasically:Target"><LINK_TEXT...

<t>Hi all, I am trying to stress a correlation matrix. To make sure the Cholesky decomposition will work afterwards, the matrix has to be positive definite, i.e. the eigen values have to positive. But by setting the negative Eigen values to 0 (or a small positive number) and recalculating the matrix...