Re: converting a R code to fortran
- From: "e p chandler" <epc8@xxxxxxxx>
- Date: 3 Mar 2007 12:22:15 -0800
On Mar 1, 2:20 pm, "fortee" <hongch...@xxxxxxxxx> wrote:
On 3월1일, 오후12시42분, Michel Olagnon <molag...@xxxxxxxxxxxxxxxxx> wrote:
I tried this code, but somehow I can't get what I want. I am
doublechecking on the R code, If my code is correclty structured by
the R code, then I should get the same table with n=40 and r=2000.. But
I can't get that result. I don't know why. It there anyone who can
look at two code and find something different or something wrong in my
How different are your results ? Monte-Carlo simulations with different
languages use different random number generators and may give different
though right results.
I started from your program (except that I did not have the poisson
number generation), and kept checking that the results were the same
as with a previous version when I made changes. But of course, at some
stages I had to change N and r, so at some point I may have made a mistake.
I don't mean your code is wrong. I mean the difference between the
result of my code and the table from the webpage are huge. I think I
did exactly the same structure as in the algorithm of the webpage.
Given a critical value, say 0.96, I have 0.5 but 0.05 in the table.
There must be something wrong in my code. I am now checking upon the R
code with replicating the code in R. Thank you.
I assume you mean the table found at http://www.mcs.vuw.ac.nz/~ray/Brownian/.
This does not look like what I would expect from a table of KS
statistics. IIRC, the critical value (for "two sided") KS/sqrt(n) at
the 5% level is 1.36 for large n. Yet in this table the corresponding
value for which p=.95 is approximately 2.48.
- Prev by Date: Re: When will this ng come to accept that Fortran needs to go "back to the future" ?
- Next by Date: Re: Division of large negative number leads to integer overflow
- Previous by thread: Re: converting a R code to fortran
- Next by thread: Re: converting a R code to fortran