Bivariate regression with dichotomous (probit link) and continuous dependent variables
57 views
Skip to first unread message
k...@psy.senshu-u.ac.jp
Jul 30, 2015, 10:39:32 AM7/30/15
to Stan users mailing list
Hi all,
I would like to write a bivariate regression model, in which one of the dependent variables is dichotomous and the other is continuous real valued.
In the attached stan file, I modified the multivariate probit regression code in the section 5.14 of the manual, so that one of the dependent variables (y2) takes any real value.
However, when compiling, it generates the error message: "ERROR(s) during compilation: source code errors or compiler configuration errors!", with long lines of C++ source code.
I never saw it before, and I would appreciate any help/suggestions that could be offered.
Thank you,
Ken
Daniel Lee
Aug 12, 2015, 10:31:15 AM8/12/15
to Stan users mailing list
Hi Ken,
Sorry for the delay. This is a bug and I'm filing a bug report.
The problem is that sum is a built into the Stan language and overloading is currently not supported. If you remove the sum function from your functions block, this should work. (it's never used.)
Daniel
Sorry for the delay. This is a bug and I'm filing a bug report.
The problem is that sum is a built into the Stan language and overloading is currently not supported. If you remove the sum function from your functions block, this should work. (it's never used.)
Daniel
Daniel Lee
Aug 12, 2015, 10:40:15 AM8/12/15
to Stan users mailing list
Hi Ken, just to let you know, the bug is here: https://github.com/stan-dev/rstan/issues/186
It's an RStan specific bug, not a problem with the Stan language. In other words, your Stan program does work with CmdStan and it should have worked with RStan.
Daniel
It's an RStan specific bug, not a problem with the Stan language. In other words, your Stan program does work with CmdStan and it should have worked with RStan.
Daniel
Reply all
Reply to author
Forward
0 new messages