Inconsistent Hanging During Sampling

Summary:

rstan simulation hangs on one machine (Windows) but not another (Mac).

Description:

Each iteration of a loop, I run the same stan model with a different set of data. The loop completes 320 iterations as expected on one computer (which is a Mac), but it hangs on iteration 83, chain 2 on another computer (which is Windows). Note the loop was structured to avoid recompiling the stan model, which takes much, much longer than running it. It is not clear if the underlying issue is Windows vs Mac - I will include installation details of both machines below.

Reproducible Steps:

I have attached files that reproduce the issue reliably. If you unzip and run /windows/forecastUpdate.R, on the Windows machine it will hang on iteration 83, chain 2. Diagnostic and sample files are included in the directory, and there are noticeable issues in the sample_file_2.csv, but alas I cannot decipher them.

In contrast, /mac/forecastUpdate.R is the same script, but gets through the 320 iterations.

samplerhangs.zip

Current Output:

Loop is successful on one machine, but fails on another.

Expected Output:

I expect the model should run successfully or fail miserably on both machines.

RStan Version:

Windows and Mac: 2.17.3

R Version:

Windows: 3.3.2 (2016-10-31)
Mac: 3.3.1 (2016-06-21)
I am posting this now, but will test downgrading the Windows version to 3.3.1.
EDIT: No dice - the issue persists after downgrading to 3.3.1.

Operating System:

Windows 7 Enterprise, Version 6.1
MacOS High Sierra 10.13.3

Start with the attached samplerhangs.zip, running windows/forecastUpdate.R and comparing it with mac/forecastUpdate.R. Inspect the diagnostic files and sample_file_2.csv around iteration 83, chain 2, then compare the reported RStan, R, and operating-system versions. Done means identifying why the Windows run hangs or documenting a reproducible fix.