RStan cropping warnings?

[Daniel Lee]

Is this intentional?

Begin forwarded message:

From: Ben Goodrich <goodri...@gmail.com>
Date: October 29, 2015 at 6:40:09 PM EDT
To: Stan users mailing list <stan-...@googlegroups.com>
Subject: Re: [stan-users] Would you consider this mixing?
Reply-To: stan-...@googlegroups.com

On Thursday, October 29, 2015 at 3:36:53 PM UTC-4, Bob Carpenter wrote:

But what happens to the exception messages that caused the divergences?
They'll indicate where things are going wrong for a user.  

Is there a way to do the R-like thing of buffering them so someone
can look at them later.    I don't think they should just be suppressed
and then thrown away altogether.

They are neither suppressed nor thrown away. The currently are collected so that at the end, it should say something like (in the case of stan_demo("inv_wishart"))

The following numerical problems occured the indicated number of times after warmup on chain 4
                                                                                                     count
Exception thrown at line 11: inv_wishart_log: LDLT_Factor of random variable is not positive definit     2
If this warning occurs sporadically, such as for highly constrained variable types like covariance matrices, then the sampler is fine,but if this warning occurs often then your model may be either severely ill-conditioned or misspecified.

followed by

Warning messages:
1: There were 53 divergent transitions after warmup. Increasing adapt_delta may help.
2: Examine the pairs() plot to diagnose sampling problems

and the whole stderr is saved in one of the randomly-named files without an extension in tempdir().

Ben

--
You received this message because you are subscribed to the Google Groups "Stan users mailing list" group.
To unsubscribe from this group and stop receiving emails from it, send an email to stan-users+...@googlegroups.com.
To post to this group, send email to stan-...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Ben Goodrich]

On Thursday, October 29, 2015 at 7:07:07 PM UTC-4, Daniel Lee wrote:

Is this intentional?

Yes, if it is permitted to wrap then it becomes really difficult to see how many of each type of exception there was.

Ben

 

To unsubscribe from this group and stop receiving emails from it, send an email to stan-users+unsubscribe@googlegroups.com.