A paper is an argument backed by evidence. It should let a reader check the claim, repeat the work and build on it. This page sets out how we write in the NLD. Where a journal’s guide for authors differs, follow the journal.

Structure of a paper

• Title. Specific and searchable. Name the system, the method and, where possible, the main finding. Avoid abbreviations and empty words such as “novel” or “a study of”.
• Abstract. One paragraph: the problem, the approach, the key result with a number, and why it matters. Stay within the journal’s word limit. See writing an effective abstract.
• Introduction. Why the problem matters, what is known (with citations), what is missing, and what this paper does. End with a short list of contributions.
• Methods. What was done, in enough detail to repeat it: models and assumptions, the experimental setup and instruments, signal-processing parameters and numerical settings. Cite published procedures instead of repeating them.
• Results. What was found. Lead each paragraph with the observation, then point to the figure or table. Report uncertainty.
• Discussion. What the results mean. Relate them to theory and prior work, state the limits of each method, and mark speculation as speculation.
• Conclusions. The main findings and their consequences in a few sentences. No new results.
• References. Cite what you used, and only that. Read every source you cite. Use the journal’s citation style and manage references in BibTeX or Zotero.

How we write

• Use “we” for what the authors did. Keep the reader’s attention on the system and the results.
• Use the past tense for what was done (“we measured the response”) and the present tense for established facts and for what figures and equations show (“Fig. 3 shows…”).
• Lead with the point. One idea per paragraph, stated in its first sentence.
• Keep sentences short. Cut words that add nothing: “it is interesting to note that”, “very”, “novel”.
• Use field terms precisely and consistently: backbone curve, frequency-energy dependence, phase resonance, nonlinear normal mode. Define a term once and keep it.
• Keep experimental, numerical and analytical results distinct, and say which is which.
• Name the limits: where a linear model fails, where a method breaks down, what the data cannot show.

Figures, equations and units

• Each figure makes one point, and its caption says what it is.
• Follow the NLD figure preparation guide (PDF): final-size figures, vector PDF, LaTeX labels in Computer Modern, the NLD figure palette, axis labels with units. Style files for Python, MATLAB and Mathematica are in the guide appendix.
• Show uncertainty: error bars, confidence bands or repeated runs.
• Define every symbol at first use and keep notation consistent. Long papers benefit from a nomenclature table.
• Use SI units, with a space between number and unit (6 mm, 20 Hz). Spell out numbers below ten unless they carry a unit.
• Refer to “Fig. 1”, “Table 2” and “Eq. (3)”, or use the journal’s own form.

Reproducibility and data

• Keep code, data and the scripts behind every figure under version control from the first day.
• Every number in the paper should come from a script you can rerun.
• Include a data and code availability statement. Public lab code is released on the lab’s GitHub; see Software and data.
• Deposit the accepted manuscript in DigitalCommons@URI when the publisher allows it.

Generative AI and research integrity

Tools and templates

• Write in LaTeX on Overleaf. Use the NLD document class for reports and theses (preparing documents using LaTeX, NLD templates) and the journal’s template for submission.
• Manage references with BibTeX or Zotero.
• Ask a labmate to read your draft before it goes to your advisor. Read it aloud yourself.

Choosing a journal and submitting

• Scope and audience. Choose the journal whose readers need your result. The journals you cite most are usually the best candidates. Read the aims and scope and check the editorial board for people who work in your area.
• Avoid predatory journals. Be wary of unsolicited invitations, very fast acceptance promises and fees without real peer review. If in doubt, ask your advisor.
• Read the guide for authors first, and prepare the manuscript, figures and files to its rules before you submit.
• Cover letter. State the problem, the main contribution and why it suits this journal, in a few sentences. Suggest qualified reviewers without conflicts of interest if the journal asks.
• Responding to reviewers. Answer every comment, point by point. Quote the comment, say what you changed and where, and stay courteous when you disagree. Mark the changes in the revised manuscript.
• Journal finders can help shortlist candidates: Elsevier Journal Finder, Springer Journal Suggester and DOAJ for open-access journals.

Before you submit

• The title and abstract state the main result.
• The introduction ends with your contributions.
• Every figure is cited in the text, in order, with units and legible labels.
• Every symbol is defined, and notation is consistent.
• Experimental, numerical and analytical results are labelled as such.
• Every reference has been checked against its source, and every listed reference is cited.
• A data and code availability statement is included, plus an AI-use disclosure if the journal asks for one.
• Author contributions follow CRediT, and funding is acknowledged with grant numbers.
• The journal’s guide for authors has been followed: length, format, figure files.

Further reading

• Writing an effective abstract
• Guideline for scientific journal publications
• Guide to abstracting (PDF)