Because of the style of a journal, and the space limit, I was asked to remove a section and all appendices.
While I have no problem with that, I want to preserve it in the arXiv preprint.
- Is it OK to have different final papers and its arXiv version?
(perhaps with the comment, e.g. "Sec 3.4. and Appendices A and B only in the arXiv version")
- Is it OK to cite in a paper its own arXiv version?
(e.g. derivation of (15) is in ?)
There doesn't need to be a strict correlation between an arXiv publication and an actual publication. It's actually common to use arXiv to publish extended/longer version of a published version (which seems exactly what you want to do). So
- Yes, it's OK to have a final paper that differs from the arXiv version.
- Yes, it's even encouraged to cite the arXiv version to point to appendices/extra material/proofs/detailed examples, etc.
However, the main rule is still that your journal paper must be self-contained, i.e., must be understandable without having to read the arXiv version. In other words, if removing the section implies to remove a definition (for instance), then you must put this definition somewhere else in the paper.