38 2. Background

A proof is an object of convincing. It should be an explicit,

specific, logically sound argument that walks step by step from the

hypotheses to the conclusions. Avoid vagueness and leaps of deduc-

tion, and strip out irrelevant statements. Be careful to state what

you are trying to prove in such a way that it does not appear you are

asserting its truth prior to proving it. More broadly, make sure your

steps are in the right order. Often, a good way to figure out how to

prove something is to work backwards, in steps of “I get this as a con-

clusion if this other property holds; this property holds whenever the

object is of this kind; and oh, everything that meets my hypothesis is

an object of that kind!” However, the final proof should be written

from hypothesis to kind of object to special property to conclusion.

True mathematical proofs are very verbal, bearing little to no re-

semblance to the two-column proofs of high school geometry. A proof

which is just strings of symbols with only a few words is unlikely to

be a good (or even understandable) proof. However, it can also be

clumsy and expand proofs out of readability to avoid symbols alto-

gether. For example, it is important for specificity to assign symbolic

names to (arbitrary) numbers and other objects to which you will

want to refer. Striking the symbol/word balance is a big step on the

way to learning to write good proofs.

Make your proof self-contained except for explicit reference to

definitions or previous results (i.e., don’t assume your reader is so

familiar with the theorems that you may use them without comment;

instead say “by Theorem 2.5, . . .”). Be clear; sentence fragments and

tortured grammar have no place in mathematical proofs. If a sentence

seems strained, try rearranging it, possibly involving the neighboring

sentences. Do not fear to edit: the goal is a readable proof that does

not require too much back-and-forth to understand. There is a place

for words like would, could, should, might, and ought in proofs, but

they should be kept to a minimum. Most of the time the appropriate

words are has, will, does, and is. This is especially important in proofs

by contradiction. Since in such a proof you are assuming something

that is not true, it may feel more natural to use the subjunctive, but

that can make things unclear. You assume some hypothesis; given

that hypothesis other statements are or are not true. Be bold and let