In truth, people do not ask these questions frequently, but it would be better if they did.

The arXiv is a scholarly e-print library with hundreds of thousands of articles in physics, mathematics, computer science, and quantitative biology. It is maintained at Cornell University. The arXiv was started in 1991 at Los Alamos in the area of high-energy theoretical physics and was then known as hep-th. For many years it was also known as the "xxx" archive or the Los Alamos archive. It is freely available and mirrored in 15 countries. New submissions are usually available the next day.

The Front for the arXiv is an overlay maintained at UC Davis. It automatically retrieves the abstracts of new and revised articles in the mathematics arXiv from the main site Sunday through Thursday. It has search and browse facilities but does not distribute full text; instead it has hyperlinks to arXiv mirror sites. It has instructions for submitting to the arXiv, but the arXiv itself processes new submissions.

If it is a serious research article, by all means! This includes not only formal new results, but also summaries, survey articles, expositions, informal notes, and software, as long they are somehow useful for other researchers in participating arXiv disciplines. ArXiv articles do not have to be refereed by a journal. Instead, the system relies on your responsibility as an author, and on a system of moderators.

It doesn't matter to anyone who runs the arXiv whether or not a submission has been refereed, published, submitted for publication, etc. However, other parties, such as publishers, may have legal or other concerns. It is strictly your responsibility as an author ensure that your submissions do not conflict with any copyright agreements that you may have made. For further information, see the arXiv policies on revision and withdrawal, and the legal disclaimer.

TeX source is required only if the article was written in TeX; otherwise PDF format is acceptable. The most original form of the article is generally the most useful for various reasons. However, for the article is written in a proprietary format such as Microsoft Word, then PDF is the best compromise.

If you have confidential material in your TeX source, for example if you commented out an incorrect passage, you should remove them before contributing the article to the arXiv. The following Unix command will properly strip comments from a TeX document 99% of the time:

perl -pe 's/(^|[^\\])%.*/$1%/' < old.tex > new.tex

If derivative formats of an article are less useful than the TeX source, then a URL is the least useful of all. The purpose of the arXiv is to record and distribute the research literature, not merely to announce its location. Abstract-only submissions will be rejected. On the other hand, you are free to include extra URLs in any research submission, as long as it is still plausibly interesting if the URLs stop working.

If you cite an article in the arXiv, you should include its archive number with the prefix "arXiv", even if the article has been published in a traditional journal. This courtesy is very important for readers and for the archive itself. For example, in many previewers you can click on an arXiv citation to retrieve the corresponding article. The recommended citation style follows this example:

[1] Dmitry Kleinbock, Gregory Margulis. Flows on homogeneous spaces and Diophantine approximation on manifolds, Ann. of Math. (2) 148 (1998), 339-360, arXiv:math.NT/9810036.

You don't need to say "preprint" or "e-print", and you should link, but not display, the URL.

Although the arXiv accepts almost any kind of TeX, there are many good reasons to use LaTeX2e (LaTeX version 2 plus epsilon). Both the American Mathematical Society and the American Physical Society support LaTeX versions of their own TeX systems, AMS-LaTeX and REVTeX; they have both frozen development of non-LaTeX equivalents. The original LaTeX manual, LaTeX: A Document Preparation System, by Leslie Lamport, is highly recommended. Another manual, Math into LaTeX: An Introduction to LaTeX and AMS-LaTeX is a good introduction to the current AMS LaTeX packages.

If you forgot the password of one of your e-prints, you can retrieve it with the password recovery form. It will be e-mailed to the address listed with the e-print.

If you forgot your user password, you can get it back by going to the arXiv registration page and clicking on "I forgot my password". It will be e-mailed to the address with your registration.

If you move, you should change the e-mail address of your registered account by following the instructions on the registration page. If you have both moved and forgotten your user password, then you're stuck. In this case, you should apologetically ask the arXiv administrators to bail you out.

The e-mail address of an article is considered a record of where the paper was submitted from, not a current author contact. The arXiv staff will not change it on request. Thus if you move, the system can no longer remind you of the article password automatically by e-mail.

The article may be in one of the four legacy archives (alg-geom, funct-an, dg-ga, or q-alg), or it may be a migrated article. Either way, the easiest thing to do is to write to arXiv administrators to provide you with a password, and to attach the article to your current user registration.

The current size limit for submissions is 1 megabyte for the compressed source. If your submission is oversized for a good reason — if it is a very long monograph or if it has numerous figures — then you can ask the arXiv administrators for an exception to the size limit. You should, in the same letter, explain your situation and send the job number of the rejection.

Many years ago, the Macintosh TeX system called Textures offered an innovative, platform-dependent way to include figures in TeX documents. But the publisher has since endorsed the standard LaTeX graphics/graphicx package. You should follow their advice, including their instructions for converting PICT to EPS.

You can usually reduce or eliminate Postscript bloat if you understand your Postscript. Computers cannot understand Postscript well enough to do it automatically, because Postscript, like TeX, is a full-fledged programming language. However, there are several case-by-case solutions:

Two Unix/Linux utilities can often fix the problem: convert is a bitmap-based image conversion program which is part of the ImageMagick graphics package. It interconverts nearly 100 different image formats, including several kinds of Postscript. eps2eps, which is part of the ghostscript package, converts fancy Postscript to simple Postscript. The output is frequently much shorter than the input, even though it is functionally identical. However, it does not help much with bitmaps.

Here are three possible reasons that your Postscript might be bloated, together with suggested remedies.

• The file has a large header with macros and/or fonts. This is typical for commercial illustration programs such as Adobe Illustrator. One treatment for a file like this is to convert it with eps2eps, which is part of the ghostscript package. For example:
  

  eps2eps fig.old.eps fig.new.eps

  You should check the size of the output (preferably after compressing both the old and new figures) to make sure that it really is smaller.

  Another solution is not to use Adobe Illustrator. The Unix illustration program xfig produces very lean Postscript output. A newer alternative is dia.

• The file is a large bitmap. This is usually because the resolution is substantially higher than necessary, because it is a black-and-white diagram saved as a full-color photograph, or because it is encoded inefficiently (using as many as 12 bytes per pixel). If you include the figure in your TeX document without rescaling it, then you should try:

  convert -density 300x300 fig.old.eps eps2:fig.new.eps

  With these options, convert produces 300dpi output in Postscript Level 2 format. Occassionally 600dpi is necessary for fine detail or 150dpi is necessary to further save space; resolutions other than these three are rarely the best choice. If you rescale the figure when you include it in your TeX document, you should change the parameters in convert so that the scaled resolution matches one of these values, or at least so that it's in a reasonable range.

• The file is an extremely complicated figure with tens of thousands of vertices, each given with many digits of precision. Sometimes you can fix the problem by changing the settings of the program that generated the figure. For example, a 3D plotting program might have an option to omit points that are hidden from view. Alternatively, you can convert it to a bitmap.

• The file was converted to a bitmap from JPEG. JPEG is an imperfect (or lossy) image compression format based on Fourier transforms. It is ideal for continuous-color images, such as photographs, that would be very large as bitmaps. Postscript Level 2 supports JPEG, and the correct way to convert is with a specialized utility such as jpeg2ps.

In choosing from the list of categories for a new article, you should mainly consider which researchers would most like to read it, rather than which terms or objects it mentions. For example, an article that discusses prime numbers might not be appropriate in math.NT, Number Theory, if they are only mentioned for the sake of a question in quantum information theory.

Please do not cross-list excessively. Very few articles merit more than two secondary categories; having no secondaries is usually fine.

Maybe. One of the tasks of the arXiv discipline advisory committees is to review the lists of categories. The lists are not perfect and there may always be a reason to add a category here or remove a category there.

The moderators for each arXiv discipline are listed on the page the describes its categories; for example, see the one for physics. They are volunteers who have been approved by the discipline-level advisory committee and by the arXiv staff.

The arXiv moderators have the authority to reclassify or reject any submission that is not plausibly interesting to the readers of their categories. Or in the case of primary classification, not plausibly of primary interest.

At the same time, authors deserve primary discretion for the content and classification of their submissions. If you object to an action taken on your article, you can write to the arXiv administration address at the bottom of this page. Appeals may be handled the advisory committee of the respective arXiv discipline. Please do not write directly to the Front, to moderators, or to members of advisory committees.

Although "no plausible interest" is the first and most important reason that some submissions are reclassified, moderators can also invoke the following specific rules as reasons. Note that these rules are only intended for blatantly inappropriate submissions. If you are unsure whether your submission crosses the line, it probably doesn't.

Improper form

Each submission should have the basic narrative of a scholarly article. Submissions that consist mainly of bibliographies, computer data, or similar should have at least some human narrative.

Epsilon content

Please don't submit articles with very little scholarly content, even if they technically have a scholarly form.

Inflammatory content

Please don't submit articles with egregiously and blatantly inflammatory content. Note that germane controversy and criticism are just fine.

Excessive cross-listing

Please don't select or add unduly many secondary categories. None is fine; more than two is rarely appropriate.

The drum-bang rule

Please avoid highly dramatic titles, abstracts, and introductions for articles with modest content.

The salami rule

Please don't divide one self-contained research contribution into many thin or redundant submissions.

The flood rule

Please don't flood the arXiv with submissions from a large back catalog, even when it does not violate the salami rule. (Please contact the arXiv administrators if you would like to submit many old papers.)

Although arXiv moderators should not referee submissions in the traditional sense of journal publication, the arXiv is not intended for plainly frivolous or incomprehensible material. If a submission is merely mediocre, speculative, or erroneous, the readers of some particular category could still find it useful. But if a responsible editor in the field would not even send the submission to a referee, then it is not plausibly interesting. The moderators serve as representatives of arXiv readers in deciding this standard.

If you find an article which is of no plausible interest in its listed categories, then usually the moderators and arXiv staff have already received many complaints about it from other users. Please try to respect the judgment and limited time of the staff and moderators, and save complaints for continuing problems that involve many articles.

As explained on the arXiv help pages on version control and withdrawal, when you revise an article, all older versions remain available. If the current version is a withdrawal notice, the article is said to be withdrawn.

You should explain in the abstract of the withdrawal notice why you are withdrawing the article. A candid, technical explanation is the best kind. It is very useful not only to those who already read previous versions of your article, but also to future readers who might learn from your mistakes.

If you are submitting an ordinary revision rather than a withdrawal notice, it is equally important to explain in the comments field or in the abstract why you are revising the article. Brief or clerical comments about the revision should go in the comments field; longer and more technical comments should be added to the abstract.

Remember people who like to check out scores of books from the library so that they can decide later which ones to read? These days, such people write software to do a million times as much of it on the World Wide Web. The entire arXiv is freely available to robots, but not to the "shoot first, ask questions later" kind.

General-purpose "robots", "spiders", and "browser accelerators" should always check the file robots.txt to determine their permitted access to the Front or to any other web site. If you use software that disregards this file, it may make you very unpopular.

If you would like to retrieve a large part of the arXiv (for example if you want to write an overlay similar to the Front), you should get the data from the arXiv and not from the Front. You can harvest arXiv metadata using the standard Open Archives Initiative protocol. If you would like closer access to the arXiv database than what OAI provides, then the best approach is to write to arXiv administrators and explain your access needs.

- for questions or comments about the Front arXiv contact page - for questions about downloading and submitting e-prints