I’m not sure how Stan finds the includes, but perhaps the problem is with the ../../../ part? I think the path must be relative to the Stan code that includes it, rather than the location where you run it from. In my code I have a chunk of code in the chunks subdirectory and the following works:
Are you using PyStan, RStan, CmdStan, or something else? Each Stan interface has a different way to specify paths to search for #include'd files:
PyStan’s StanModel constructor takes an include_paths argument.
RStan, AFAIK, still uses the stanc_builder() to account for #includes.
CmdStan has the STANCFLAGS Makefile variable, which can be modified to include an --include_paths option, which takes a comma-separated list of directories as an argument.
There’s something peculiar about the way paths are interpreted, it frustrated me to no end when I first tried. My guess is that when you write /work/etc it gets interpreted as ./work/etc. At least this is what I think it’s happening since for me if I have the stan files in ./src/stan_files and I want to include ./src/stan_files/chunks/fun.stan my include is #include /chunks/fun.stan. I suggest you to try putting the files you want to include in a subdirectory and see if that takes you anywhere.