This is a (pandoc) conversion of the document found here, linked from this page.
There, a more in-depth document with guidelines is linked aswell (direct link).
Due to this conversion, figures, refs and equations aren’t properly converted. Additionally, comments in the original .tex document aren’t shown here either. So for the full set of information, visit the links above.
Never use other people’s ideas/text/images without citing and clarifying copyright issues.
Analyze the papers you read with respect to their style.
Start with a logical structure, do exact formulations later.
Have someone revise your paper.
Polish your abstract. Many readers will read only this.
Create a table of contents during drafting, even if it is not needed for the final version. Readers should quickly know where to find which information, so strive for “mutually exclusive” headings.
Strive for symmetry in terms of number and length of sections per chapter, or of subsections per section (introduction/conclusion are exceptions with less subheadings).
Do not place chunks of “loose” text between section heading and subsection heading. One or two sentences are possible, but only for an overview of the entire section.
A section cannot contain only a single subsection.
Do not skip heading levels.
Avoid any more than three levels (section, subsection, subsubsection). If you need to split at a deeper level, revise your structure.
Each paragraph should contain one distinct idea/message. Also, the reader should see the paragraph’s topic in the first three words, but at least in the first sentence.
Ensure logical connection within and between paragraphs. It is easiest to write down the main ideas first, and only then to expand them to paragraphs.
Do not get creative with different types of line breaks between paragraphs. If you find that necessary, revise your structure.
Write in short and concise sentences.
Do not use contractions (don’t) or colloquial speech.
Avoid static verbs and passive tense; try to write actively. As an exercise, start by replacing all forms of “to be” in your first two paragraphs with an active construction.
Look up the rules for hyphenation in English, particularly for “compound modifiers”, to know the difference between a “red haired person” and a “red-haired person”.
Look up the rules for which/that (including punctuation).
Whenever you look something up (like grammar rules), do not trust newsgroups or other sources like that. Get a reliable source, e.g. a book.
Try to use only English words you know!
Merriam-Webster helps with spelling and hyphenation.
Install Grammarly! This is a free browser extension that automatically checks and underlines grammar and spelling errors in the browser. The checks are better than the default Word checks (there is also a Word plugin). It also corrects in Overleaf.
When writing your thesis or any document that is (or could be) disseminated in any way, have all of your figures free of any copyright issues. That means: Either draw it yourself (entirely, taking pieces from other pictures and modifying these is also an issue) or obtain permission and state in the figure caption that you got this permission and from whom. Mostly publishers will tell you how to write it and even have an automated process to get permission, for example via RightsLink/Copyright Clearance Center (mostly instantaneous and free for students’ theses). Also pictures from the public domain need proper referencing. There are also different types of licenses, for example some do not allow modifications to a picture.
Avoid direct citation (“…”) whenever possible, instead rephrase statements when citing
It is clearly preferred to cite primary, original literature, not reviews or textbooks. Go back to the original sources. If you cite reviews or books, do that in an explicit way, such as “an overview is found in…”.
Read until you understand each paper you are citing, or at least the portion of the paper you are referring to. Do not just rely on abstracts or, worse, secondary literature that cites the original source (often incorrectly).
Provide evidence for each single claim that is not an obvious fact, either in the form of a reference, or by your own experiment, deduction, or calculation.
Define each single variable variable explicitly. Do not rely on “self-explaining” indices like or on the reader’s background experience (for example to assume that must be a force). Choose short indices, rather one than multiple letters.
Operators and common mathematical functions are typeset upright [@RedBook2010]. This is particularly often done wrong with the differential operator, or with sin and cos. Example: The derivative of variable with respect to time is $$\dot{x}=\frac{\mathrm{d}x}{\mathrm{d}t} , . \label{eq:derivative}$$
Everything that is text should not be typeset as variables. This also holds for indices of variables. For example, for a variable with intended index “m”, where “m” is an abbreviation of “max”, write and not .
Units: It is a very frequent mistake to put square brackets around the unit. This is incorrect according to the SI [@SIbrochure2006] and the ISO norm [@ISOnormunits]. Example: For a force , correct use of the operator is: . For axes labels in plots, use division as in “” (preferred), or “force (in N)”.
Units are typeset roman (upright), not italics (sloping).
The space between a number and a unit should be slightly less than a normal space. Example: moment .
Be aware that also controller gains have units! Report them in the paper. Understanding the units also helps you choose reasonable gains during tuning.
Do not use “” to indicate multiplication (Reserve this operator for convolution). Use no operator between variables, or “” if needed. Example: .
Do not start sentences with numbers or variable names. Sometimes, it helps to put constructions like “The parameter ” or “The variable ” or “A value of 20…”.
Typeset scalars, vectors, and matrices differently. Example: The mass matrix of a robotic manipulator is a function of the robot’s generalized coordinates .
Make formulae part of sentences. Place full stops when a sentence ends after your equation, or a comma when appropriate. See the above example equation [eq:derivative].
Reference all figures and tables in the text, before their occurrence.
Use vector graphics whenever possible; do not rasterize text. You can export figures as .eps
from Matlab.
Make sure your figures are all perfectly readable in b/w printing and by colorblind people. So, do not (only) color-code information, but use different linestyles or thicknesses or grayscales.
All text in figures should be readable without zoom, also by elderly readers. So, do not use font sizes any smaller than footnotesize.
Use the same font for symbols in figures as you use in your main text. A package that helps here is psfrag
. Example: A disk will change its angular velocity when an external torque acts on it (Fig. [fig:diskwithtorque]).
[l][l]disk [c][c] [c][c] [c][c]
Making tables directly in Latex is a pain. This site is useful to generate tables: https://www.tablesgenerator.com/
You can also copy an existing Excel table to it.
Sometimes the copy to tablesgenerator.com will not work. In this case, or if extra functions are desired, an alternative is to install OpenOffice Calc (Open source Excel) and the related macro CalcToLatex. This works on more tables, and can automatically escape characters Latex would choke on, as well as do some other settings.
OpenOffice Calc download: OpenOffice Calc: https://www.openoffice.org/download/
CalcToLatex download: https://sourceforge.net/projects/calc2latex/
Macro installation instructions: https://calc2latex.sourceforge.net/install_oo20/
Use instructions: https://calc2latex.sourceforge.net/
Comments in the .tex
-file of this document
You can create a tracked-changes PDF document using the old and new .tex file using this online tool
What is the context / motivation of your work?
What is the State of the Art? Cite relevant literature.
What is missing in the literature?
What is your new contribution?
The methods section contains the recipe that allows others to reproduce your results, without reading the results section. So, here you answer questions like:
How do you design your device/your controller?
How do you design experiments/simulations?
How do you collect and analyze data?
Choose descriptive names for new methods (instead of words like “novel”), to simplify citation and future improvements.
How does your method perform? Give quantitative information and illustrate with plots.
Do not give extra information on your methods in the results section (such as method to choose parameters, or the experimental protocol).
Give the raw facts only; do not interpret your results yet. That means no sentences like: “The improved performance demonstrates that the controller is effective…” or “A probable reason for this observation is…”.
Interpret your results. Is the performance acceptable? Go back to initial questions, compare to other methods reported in literature. How do you explain unexpected results? Which shortcomings or limitations are there in your methods and results? Where is room for improvement? Could combination with methods from literature further improve the work? Mention plans, needs, and recommendations for future research.
This is not a summary. Instead, state the take-home message.
Thank everyone who contributed with smaller pieces of advice or non-scientifically (e.g. technical support).