[R] Questions about ks.test function {stats}

Jin, Ziyan z|y@n@22 @end|ng |rom |nt|@zju@edu@cn
Sun Apr 7 10:18:40 CEST 2024

Dear R-help,

Hope this email finds you well. My name is Ziyan. I am a graduate student in Zhejiang University. My subject research involves ks.test in stats-package {stats}. Based on the code,  I have two main questions. Could you provide me some more information?

I download different versions of the r language source code through r language website (https://www.r-project.org/). By reading R-4.3.3/src/library/stats/R/ks.test.R, I encounter the following problem: before call the default psmirnov function (in two-sample case), z <- NULL. According to the T and F the TIES, determines the value of z assigned to w or not. However, when psmirnov is called, z=w is always used. I am curious whether the TIES parameter can be omitted.

Compared with the previous ks.test, such as version 4.1.3, the method of calculating p value in version 4.3.3 has changed a lot. ks.test() now provides exact p-values also with ties. For the psmirnov_exact_uniq_upper function in the ks.c file (R-4.3.3/src/stats/src/ks.c), could you please provide some details for the mathematical basis used to calculate the p-value? If you could provide me with some references, I would be grateful .

Thank you for your patience. I am eagerly awaiting your response.


	[[alternative HTML version deleted]]

More information about the R-help mailing list