Reputation
436
Next privilege 500 Rep.
Access review queues
Badges
4 12
Newest
 Yearling
Impact
~16k people reached

  • 0 posts edited
  • 0 helpful flags
  • 22 votes cast
Jul
21
comment Is there a map that maps the mean vector and the variance matrix of a multivariate lognormal to its location vector and diffusion matrix?
Good that you found a potential solution.
Jul
21
comment Is there a map that maps the mean vector and the variance matrix of a multivariate lognormal to its location vector and diffusion matrix?
Also, check this post just in case it is useful to you.
Jul
21
comment Is there a map that maps the mean vector and the variance matrix of a multivariate lognormal to its location vector and diffusion matrix?
I would suggest you could do this to make sure the resultant transformation is correct: Sample some data from the known mean and std dev of the obtained lognormal transformation. Take log of those data and check if their distribution is normal or not. if your transformed lognormal distribution is correct then the log of those data should give you a normal distribution.
Jul
21
comment Is there a map that maps the mean vector and the variance matrix of a multivariate lognormal to its location vector and diffusion matrix?
Nicola: To clarify my few "side comments" that I gave earlier, I would like to make two points. 1) not all covariance matrix you obtain numerically will have the property of "covariance" or "correlation" matrix, which is that the covariance matrix is supposed to be positive semi-definite and not all estimates are guaranteed to have that property. 2) I checked if the transformation is PD provided it was PD before, however, I could not find any simple explanation for that.
Jul
20
comment Is there a map that maps the mean vector and the variance matrix of a multivariate lognormal to its location vector and diffusion matrix?
If the covariance is standard then it'll have sill, hence, it'll be positive definite. However, if you're numerical covariance does not have a sill then it's not positive definite. It's the same thing I believe as stationarity
Jul
20
comment Is there a map that maps the mean vector and the variance matrix of a multivariate lognormal to its location vector and diffusion matrix?
I believe since you're transforming from one standard distribution to another standard distribution so that should be fine. Maybe either you're 1) missing log while transforming. Sometimes the distribution you have is normal so make sure you're taking care of transforming it to log normal first by taking log. 2) it could also be an error in coding somewhere.
Jul
20
comment Is there a map that maps the mean vector and the variance matrix of a multivariate lognormal to its location vector and diffusion matrix?
Where did you get that transformation from? Does the field exhibit stationarity or in other words the variances of blocks of data should be constant. It'll be positive definite if it's stationary.
Jul
20
comment Bounded Matrix-Vector Multiplication
Zircht: I don't think the condition you proposed on boundedness includes all possibilities. $\| Ax \|$ will, in fact, be always bounded for given conditions. You may want to check the explanation below in the answer.
Jul
20
comment Solution of Second order ODE: theoretical question
Okay. You may want to edit your question then to clarify that those solutions which do not have exponentials in them.
Jul
20
comment Is there a map that maps the mean vector and the variance matrix of a multivariate lognormal to its location vector and diffusion matrix?
You could probably get more help if you edit your question in a more structured format ( e.g. a) Your problem setting, b) What issues you have, and c) If you have tried something) than writing long paragraphs.
Dec
23
comment Can decimal numbers be considered “even” or “odd”?
@Ben: I appreciate your publicaly revealing your downvote, but I still feel you took a very subjective approach (probably based upon above comments?) to this answer because there so many answers which get voted up even though they do not appear to be in the spirit of seeking understanding. :)
Dec
18
comment What is the CDF for the following PDF of a cut-off log-normal distribution (in Matlab)?
The question is what is the CDF of the given distribution? If I know CDF then I can follow the rest of the steps...
Dec
18
comment What is the CDF for the following PDF of a cut-off log-normal distribution (in Matlab)?
Would it be correct if I find truncated normal random variables by generating the untruncated values and replacing those that are outside the limits. Then, finally use the truncated normal random variable, say y, as follow: r = rch * exp(y).?
May
18
comment How to express the traditional variogram for a non 2nd order stationary random function?
@Shai: Can you figure out from that expression why the difference between expected values at both points should be added?
May
17
comment How to express the traditional variogram for a non 2nd order stationary random function?
hmm..i'll see..thanks Shai Covo!
Apr
4
comment How to deduce the CDF of $W=I^2R$ from the PDFs of $I$ and $R$ independent
@Shai: Can you post the complete solution? Thanks!
Apr
4
comment How to deduce the CDF of $W=I^2R$ from the PDFs of $I$ and $R$ independent
@Shai: How would you back transform once you have pdf or cdf of log(W)?
Apr
4
comment How to deduce the CDF of $W=I^2R$ from the PDFs of $I$ and $R$ independent
@wnoise: Can you edit your equations to look clear. It doesn't looks readable to me. Thanks!
Apr
1
comment solving quadratic equations by completing the square
@Myself: hehe...I haven't learned latex!!
Mar
24
comment Proof of upper-tail inequality for standard normal distribution
Also, the integral in first step is definition of CDF, but LHS i.e. P(X>x) is not CDF. How is that?