I am about to begin studying as an undergraduate. Previously, the mathematical exams I have studied have been incredibly lax in terms of notation or neatness of proof - it would not be penalised, for instance, to write $f'(x)$ instead of $\frac{dy}{dx}$ when the function given was in terms of $y$. It was also considered equally valid when given an identity to prove, to work from the LHS and then from the RHS and join the two together in the middle (rather than a coherent working from one side to the other).

Anyway, as it is not something featured in my previous syllabus, I am looking to improve my understanding of notation to use when proving a statement. I have come across two inconsistencies in teaching materials so far:

1. Use of $=$ when $\equiv$ is valid.

Now, I am comfortable with the meaning of these symbols. But my question is when the use of $\equiv$ or $=$ is mandatory, and when it is permissible. An obvious case to me would be something like: $$\frac{3x^2+34x+6}{x^2+2x-24} \equiv \frac{3x+7}{x-4} + \frac{9}{x+6}$$ Here, use of $=$ would imply it is an equation to solve, rather than something true $\forall x \in \mathbb{R}$ ($x \ne -6, 4$). All materials I have come across would use $\equiv$ here.

But why is it common to write $3+2=5$, when clearly this is not an equation to solve, but an equivalency? Would $3+2 \equiv 5$ not be more accurate? In a more borderline case, if part of the answer to a question includes the simplification of $3(x-4)+2$ to $3x-10$, which symbol is most appropriate? I have seen both $ 3(x-4)+2 = 3x-10$ and $3(x-4)+2 \equiv 3x-10$.

The definition of the $\equiv$ symbol (two things are equal for all values of any variables used) seems to contradict the notation I have seen in common use, based on its lack of use outside of proving identities.

2. Use of $\implies$ or $\impliedby$ when $\iff$ is possible.

If I am writing a proof and we progress from one stage to the next, it seems that $\implies$ would demonstrate the flow of logic more accurately, even if $\impliedby$ is also true and thus $\iff$ could be used. For example, if a question asked us to find all possible values of $x$ in some trigonometric equation, and we had simplified down to:

$$\sin(x) = 0 \implies x = n\pi\; (n \in \mathbb{Z})$$

Here, I am conflicted, because $\implies$ seems to be more coherent, as we are interested in progressing from the left to the right but have no use for the information that progressing from the right to the left is valid. However, if $\implies$ contains within it the implicit notion that $\require{cancel} \cancel{\impliedby}$, then the use is incorrect and should be replaced by $\iff$. Which symbol should be used, or are both permissible (and the question down to personal preference)?

  • 2
    $\begingroup$ You say "It was also considered equally valid when given an identity to prove, to work from the LHS and then from the RHS and join the two together in the middle". That is evil $\ddot\frown$. $\endgroup$ Sep 23 '17 at 15:19
  • 1
    $\begingroup$ I think that's perfectly fine, and in many cases more transparent than reverse one of the chains and concatenate to pretend it was a single chain (half of which may seem rather mysterious when written backwards) $\endgroup$
    – user334639
    Sep 23 '17 at 15:33
  • 3
    $\begingroup$ @Lor It's not evil if one does it correctly, i.e. using bidirectional inferneces. In fact it is so useful it even has a name in the (atuomated) theorem proving literature: forward and backward chaining. It can lead to significant speedups when the search trees have large branching factors (.e.g solving chess endgame problems). $\endgroup$ Sep 23 '17 at 16:23

Ad 1. You write

But my question is when the use of ≡ or = is mandatory, and when it is permissible.

There is no absolute answer to this, yet the use of '=' between terms like the one you give, which can be seen as a elements of the 'rational function field' $\mathbb{Q}(x) = \mathrm{Frac}(\mathbb{Z}[x])$ is much more usual in modern mathematics. The use of $\equiv$ you have encountered seems to be some sort of 'subcultural' usage confined to some schoolbooks (I guess).

To explicitly address the example you give: I can assure you that most mathematicians would consider

$$\frac{3x^2+34x+6}{x^2+2x-24} \equiv \frac{3x+7}{x-4} + \frac{9}{x+6}$$

an unusual notation. The most usual notation is to write

$$\frac{3x^2+34x+6}{x^2+2x-24} = \frac{3x+7}{x-4} + \frac{9}{x+6}$$

and interpret this as an equality in the field $\mathbb{Q}(x)$ of rational functions. Yes, in the background, depending on your choice of set-theoretic foundations, there may be equivalence classes somewhere, yet this is irrelevant, and this is a genuine equality.

I would appreciate if some other people would second this (uncontroversial) opinion and talk the OP out of their fear of using = here. There might be good reasons to use other symbols for this sometimes, but recommending the OP to use whatever they see fit is definitely misleading.

In particular, your statement

All materials I have come across would use ≡ here.

if true, implies that you have been brought up on a very limited set of 'materials'.

And in

But why is it common to write 3+2=5, when clearly this is not an equation to solve, but an equivalency?

the statement about the "equivalency" is wrong, at least relative to usual contemporary mathematics. This is an equation between natural numbers.

As a general rule, remember that

  • when you are working with the elements of an algebraic structure (group, ring, field, ...), then the usual 'relation symbol' is '=' and not $\equiv$.

The symbol $\equiv$ has various meanings in various contexts.

Ad 2.


However, if ⟹ contains within it the implicit notion that

Please note that the usual convention is that $\Rightarrow$ never means $\require{cancel} \cancel{\impliedby}$. Never. If it would, it would be impossible to express $\Leftrightarrow$ in terms of $\Rightarrow$ and $\Leftarrow$.


Which symbol should be used, or are both permissible (and the question down to personal preference)?

The latter. Both are permissible. The choice is a choice of emphasis. And of course they do not mean the same, as you know.

Does that help?

  • 1
    $\begingroup$ +1 This is a correct description of mathematical conventions on such matters. $\endgroup$ Sep 23 '17 at 16:13
  • 1
    $\begingroup$ I would add that it is also extremely common (and a good habit to get into!) to explicitly state the quantifiers. That is, basically no one ever uses $\equiv$ the way OP says, but very often they don't just write $=$ either, but instead write $=$ and additionally say in words that the equation is true for all values of the variables (or for all values of the variables in some particular set). $\endgroup$ Feb 18 '18 at 19:52

It is absolutely fantastic that you are thinking about these issues as you start your undergraduate career. If only more students would appreciate these subtleties in mathematics. One nice thing in mathematics is that everything can be made precise.

A couple of remarks though:

  1. Often we will define precise use of notation and then immediately violate the convention. Insisting on precise notation everywhere is not helpful because everything becomes too cumbersome. It can become easy to bury what you are trying to communicate in notation. Abuse of notation is very common and accepted. Also, mathematics is about more than a game of notation. While it strictly speaking might be true at the root, mathematics is also about ideas.

  2. Remember the context. Saying that $x^2 + x + 1=0$ might mean an invitation to solve the equation. It might mean that $x$ is an elsewhere defined number and that the number $x^2 + x + 1$ is equal to $0$.

  3. Ask your teacher. The teacher will probably have certain preferences when it comes to notation. Don't then get mad at your teacher because he/she doesn't follow notation that you have used before. Instead, get used to the change of preferences.

Now, typically $=$ is used to say that two elements in a set are the same element. Saying "solve $x^2 = 2$" can then mean find all elements $x$ in the set $\mathbb{R}$ whose square is the same as the element $2$ in the set of real numbers. Saying that $2(x-3) = 2x -6$ might say that the polynomial $2(x-3)$ is the same as the polynomial $2x - 6$. It might mean that the function $f:\mathbb{R} \to \mathbb{R}$ given by $f(x) = 2(x-3)$ is the same (as element in a set of functions) as the function $g:\mathbb{R} \to \mathbb{R}$ given by $g(x) = 2x - 6$.

Saying that $\frac{x^2}{x} = x$ might again be about an equality of functions. But what it the domain of these functions? both functions would have the same domain, namely $\mathbb{R}\setminus\{0\}$.

You say that "if $\implies$ contains within it the implicit notion that $\require{cancel} \cancel{\impliedby}$ ..." This is simply not true. You can, in fact take the definition of $A\iff B$ as ($A\implies B$ and $B\implies A$). So indeed both of the following is correct $$ x^2 = 1 \implies x\in \{\pm1\} \\ x^2 = 1 \iff x\in \{\pm1\} $$ Both are therefore permissible.

Here is the thing. When you are writing a proof you want to be careful and precise. Getting into the habit of writing $\iff$ everywhere you can will likely lead you to use it wrongly at some point. Being careful is to prove that $A\iff B$ by first showing $A\implies B$ and then $B\implies A$ even if you could do both at the same time.

The symbol $\equiv$ is often used in different ways. It will often depend on the definition. I think that most sources would not use $\equiv$ in the place of $=$.

  • 2
    $\begingroup$ @Adam: on the whole I agree with everything in Thomas' nice answer. A few technical addenda: the distinction in '2.' of this answer is nice and instructive, and I would just like to add that 'invitation to solve' is a non-mathematical meta-concept, and that the distinction in '2.' could more technically put this way: $x^2+x+1=0$ is a formula with $x$ as its only free variable, while in the context defined via "elsewhere defined", the correct technical term for $x^2+x+1=0$ is atomic sentence, wherein the 'sentence' refers to there being no free variable anymore: since [...] $\endgroup$ Sep 24 '17 at 7:21
  • $\begingroup$ [...] $x$ has been 'elsewhere defined', and assuming that 'elsewhere defined' is taken to mean: defined in terms of constants of the structure we are speaking of', it follows that then '$x$' is a constant (in the model-theoretic sense). Also worth pointing out: the definition of $\Leftrightarrow$ that Thomas gives is the, overall, most useful and common one, as a binary operation on the set $\mathcal{P}$ of all propositions within a given context and w.r.t. to classical logic. A useful technical term if you would like to learn more about this is propositional logic. From this [...] $\endgroup$ Sep 24 '17 at 7:27
  • $\begingroup$ [...] point of view, each of the following "symbols" $\Rightarrow$ (implication), $\Leftrightarrow$ (equivalence), $\&$ (logical and), $\vee$ (logical 'or'; the symbol derives from the Latin word '$\underline{\text{v}}$el'), $\downarrow$ ('nor' aka 'Peirce's arrow'), $\uparrow$ ('nand' aka 'Sheffer stroke') is simply a function $\mathcal{P}\times\mathcal{P}\rightarrow\mathcal{P}$. Moreover there is the function $\neg\colon\mathcal{P}\to\mathcal{P}$ ('negation') Please note that [...] $\endgroup$ Sep 24 '17 at 7:37
  • $\begingroup$ [...] some compositions of these functions are 'extensionally equal' (here this means" they 'give the same results'), while 'intensionally different' (roughly: they are 'built' differently). It is a very common convention that functions are considered equal if they are 'extensionally equal'. It is a topic of research to this day if and when and how this convention is to be changed, but you need not worry about this. By and large, you'll only need $\Rightarrow$, $\neg$, $\&$ and $\vee$. Also: the above is only the most usual way of giving meaning to the symbol '$\Rightarrow$'. Others exist. $\endgroup$ Sep 24 '17 at 7:45

It's nice that you have this disquietude at this point.

Mathematical precision and clarity depends on the context, the writer, the intended reader and their common backgrounds.

Absolute precision using just symbols is possible in theory but in practice you will be much better using words instead.

Hence, whether the symbol "=" means an equation (so you are talking about whether there is some $x$ for which equality holds), a statement (so $x$ has been fixed before and you are asserting that both sides of equality are the same number), an identity (so for all $x$ in a certain set the resulting numbers on both sides of the equality will be the same number), or a functional equality (so the functions implicitly defined by both expressions -which is often an imprecise way to define a function- are the same function), etc, should either be clear from the context or it should be made clear on the fly.

When solving an equation such as $\sin(x)=0$, you should use <--> because you want to describe the entire set of solutions, not more, not less.

Now if you are proving an implication with a chain of implications, you'd normally use words instead that of symbols. In any case, you will assert that two things are equivalent instead of just saying that one implies the other if this hint will produce a better impact on the reader. For the logical validity of a proof, if you show that a-->b-->c-->d then you have shown that a-->d, and that's all you need to argue. But telling the reader that actually b<-->c can make their life easier or harder, depending on the context.

In summary, good math writing is not about utmost symbolic precision, it's about understanding who you're writing to.

  • 1
    $\begingroup$ Re "Hence, whether = means a question or an affirmation should either be clear from the context or it should be [...]": The usual convention is that '=' never means a "question". It is a 'logical symbol'. And "affirmation" is not a usual technial term, while of course it's okay to say that $a=b$ is an 'affirmation that $a$ and $b$ are equal. And re "Absolute precision is possible but produces unreadable text and nobody does it.": I think this is not helping the OP much. In situations as easy as this, one can and should be absolutely precise. $\endgroup$ Sep 23 '17 at 15:41
  • $\begingroup$ Thanks for the comments. About question vs affirmation, I have reformulated. About absolute precision, please refer to the accepted answer, where almost every paragraph says exactly the opposite: that in practice the mathematical notation people use can always be made more precise, therefore it is never absolutely precise. $\endgroup$
    – user334639
    Sep 24 '17 at 1:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.