Submitting a problem

Invited physicists with a linked ORCID iD can add open problems (accounts). Each one is signed with an ORCID login and goes into the public log, like every record on PubPhys.

Steps

1. Upload. Choose a field and upload a PDF (up to 50 pages), a Markdown, LaTeX or plain-text file (at most 10 MB), or type the problem. The upload is itself a record in the log at once; only its hash is visible until the problem is published.
2. Fields. A problem has a title, a statement in plain words, a precise question, what would settle it, its kind, the expected answer, references and its status in the literature, under one topic of the field. If you allow it, AI proposes these from your file (what that means); otherwise you fill them in. Formulas are LaTeX between $...$ and must render before you can go on.
3. Sign. Your browser builds the record from the form and computes its hash; you sign it with your ORCID password. You get a receipt to keep.
4. Screening. A quick automatic check (with your AI consent) asks only whether this is an open question in physics, stated intelligibly; not whether it is right or important. If it is confident, the problem is published at once as unreviewed. Otherwise, or without AI consent, it waits for an editor of the field.
5. Check after publication. With your AI consent, a published problem is checked again for gaps and compared with up to five similar published problems. The report appears on your problem's page under Your problems; only you and the field's editors see it, and it changes nothing by itself.

Before preparing the record, Check for similar problems lists published problems with similar words, without AI and without recording anything, so you can see whether yours is already there.

A new problem may build on up to 5 published problems: a special case, a generalization, a combination or a variant of them, each with a one-line note on what differs. Pick them in the form's "Parent problems" block, or start from a problem's page with Make a variant, which opens an empty form without a file (problems built on other problems).

Limits per account: 5 new problems signed and 20 uploads (failed ones too) per UTC day, and at most 3 problems in progress (being written or signed); every limit, with its current value: limits. What you write is shared under the Creative Commons Attribution 4.0 license (CC BY 4.0): anyone may share and adapt it, crediting you (terms of use).

Writing a good problem

Write in English, in the third person and in a neutral tone, without claims such as "obviously" or "groundbreaking". One problem asks one question: split a list of questions into several problems. Define each symbol where it first appears; formulas are LaTeX between $...$.

• Title: one specific line, without a final period. Origin of the large phonon thermal Hall effect in underdoped cuprates, not An interesting question about cuprates.
• In plain words: two to four sentences a physics student can follow: the system, what is observed or unknown, and why it is open.
• Precise question: the question for specialists, with its assumptions and parameters and exactly what is to be found or proved, ending with a question mark. For example: For the 3D incompressible Euler equations with smooth initial data $u_0 \in H^s(\mathbb{R}^3)$, $s > 5/2$, does $\int_0^T \|\omega(t)\|_{L^\infty}\,dt$ diverge for some finite $T$?
• What would settle it: a concrete result that closes the question either way: a measurement, a proof, a calculation or a counterexample.
• Kind: a well-posed question with a definite answer, or a phenomenon: an observed fact that needs a mechanism. Expected answer: yes or no, a value, a mechanism, a classification, a proof, or an identification.
• References: only works you have checked, one per line, the paper that posed the problem first, for example L. Smith et al., Phys. Rev. Lett. 120, 123456 (2018); arXiv:1801.01234.
• In the literature: open; contested when published results disagree; partially resolved when part of the question is answered. An optional status note says where things stand, with sources.
• Posed since: the year it was first posed in the literature; leave it empty if unsure.
• Topic: the narrowest topic of the field that fits. The list also offers the topics you proposed that are published and not yet decided, and, when topics are being proposed, "Propose a new topic", which you sign together with the problem (proposing a topic).

When AI proposed the fields, check the references above all (AI takes them only from your text but may garble them), the kind of problem, and that the statement claims nothing your text did not.

What becomes public, and when

• Before publication, only the log entries are public (they show that a record exists, not what it says). The upload, the signed record and any decision on it are seen by you and the editors of the field only.
• On publication the problem, its signed record and the uploaded file become public. The file is published exactly as uploaded, including metadata it may carry (such as an author name), and cannot be changed afterwards.
• If you abandon a problem before signing it, or leave it for 30 days, the uploaded file is deleted; its log entry stays.

Moderation

Editors are appointed per field (see appointments). For a held problem an editor either releases it (published as unreviewed) or signs that it is not accepted at screening; that decision stays as private as the problem, and you may appeal it within 30 days. On a published problem an editor signs accept or reject, with reasons; both stay public. Editors sign these decisions with their own ORCID iD, and nobody decides on their own problem. Any problem can be removed for copyright, personal data, abuse or spam: its content and file are then withheld for good, and a page of a problem that was public says why. Every stage, who acts at it and what is public: editors and their decisions.

After publication

The text of a published problem is never edited in place. "Edit this problem" on its page signs a new revision with your ORCID iD. A change of wording to a problem an editor has accepted is displayed at once; a change of the question waits for an editor (revising a problem).