If you use something from another paper, then you should cite it. Make that a general rule. There are a number of reasons for it, only one of which is to avoid a charge of plagiarism when someone notices that they have "seen that before."
The other, perhaps primary, reason is that scholars want to know how a paper is situated within the scientific/mathematical literature and if previous work isn't cited it is difficult or impossible to do so. A future reader of your work may want to know where else this definition appears and what its history is. Citations make this possible.
However, for things that have become common knowledge, such as the definition of the derivative (math) or ph (chemistry) you don't need to give definitions as you can expect that they are known to every scholar in the field. But you probably wouldn't repeat those definitions in the first place unless you were doing a comparative history study of some kind, in which case you would cite the original sources.
Don't quote or paraphrase the work of others without citation.