I'm asking this question as a student, wondering what various pros/cons to the given formula for oral exams could be. Let me give some context first. I am a first year mathematics student at a university. The are two approaches available to students for every subject: There is a "basic" group and an "advanced group" and we were able to choose which one we wanted to attend at the beginning of both semesters with the possibility of moving from one group to another if we found the level inadequate for 2 weeks since the semester's start. The difference between groups is really large, to the point where the only people left in the advanced groups are ones who have either studied a significant amount of university-level maths in their free time in high school. So, in short, attending the advanced groups is completely voluntary and the courses can't be really treated as ordinary introductory ones, because while they might theoretically assume any background knowledge, they do require quite some mathematical maturity which is bound to come with the backgorund knowledge.

After this long rant we arrive at the actual question. In these advanced groups, students are required to know all the theorems together with complete proofs of all the theorems/propositions presented during the entire semester in the given course. Learning all that is obviously not an easy task - not only is that a lot of material, it is also significantly harder than expected from an ordinary first year course. What do you think about such a formula? Is it too much? Or is it actually a good thing to expect this much from students who are highly above "average"?

I think I did somewhat benefit from learning all the proofs in the first semester, but on the other hand the task seems like a bit unreasonable, since even professors use notes when presenting proofs during lectures.

  • $\begingroup$ Uhm, what is this? In my university everything seems advanced groups then. $\endgroup$
    – user7171
    Mar 10, 2018 at 12:00

4 Answers 4


This will depend in some part on the course you are taking. If it is an introduction to proofs class, then probably it is reasonable to hold such an expectation. For higher level courses, some of the details that go into a proof can make memorizing a lot of theorems difficult. (Concrete example: I think the proof that permutation groups can be partitioned into "even" and "odd" elements is the sort that, if there were a lot of these to be memorized for an Abstract Algebra final, would constitute asking too much of the student.)

An important difference in how advanced vs. novice students think about proofs is related to how they "chunk" the information. An advanced undergraduate who has finished Calculus in high school might be taking a first course on Real Analysis by semester two of their first year; in such a case, consider the theorem that says every closed and bounded infinite subset of $\mathbb{R}$ contains a limit point.

When I first saw this proof, I was very caught up in the details; I could not chunk it very well, having no real familiarity with this style of argument. Today, I certainly would not memorize how it is proved verbatim. Instead, I would think: Put subset into an interval (boundedness); divide in half and choose left half if it contains infinitely many points, otherwise choose right half (infinite); repeat indefinitely. This zooms in on a limit point [essentially by giving its binary representation if the interval is $[0,1]$, where "choosing left" gives a $0$ and "choosing right" gives a $1$] and that limit point is in the set (closed).

Even in the paragraph above, I have written more details than I keep in my head. My actual thinking is something like: divide in half repeatedly and use the three givens (boundedness, infinite, closed).

To keep many proofs (especially ones that are not being written out formally, but rather explained in an oral examination where they will probably give you hints if you get stuck) in one's working memory is a doable task. But it does, in your own language, "require quite some mathematical maturity."

Summary: This task could be unreasonable if the proofs are overly involved and too many; but, more generally, the ideal for such an examination is that students can grasp the key concepts and reconstruct proofs or proof sketches without needing to store word-for-word arguments in their working memory.

  • 3
    $\begingroup$ This is a fantastic point. I think when I teach a proof based class, I will make sure "breaking proofs down into parsable chunks" is emphasized. In fact, I will have HW questions that say, "Outline the proof of theorem X on pg. y (your outline should contain no more than n steps). We will spend time in lecture discussing people's outlines. $\endgroup$ May 12, 2014 at 18:10
  • 1
    $\begingroup$ I disagree with your concrete example. It is trivial to define parity of a permutation as the parity of the number of inversions in the sequence representing it, and then easily prove that the parity is changed by adding any swap. All a student would have to do is to remember what I just said, and reconstruct the rest. Not hard at all. On the other hand, if the proof you had in mind was more complicated, then the problem is not with the student... Maybe that's what you were trying to say later, but it is hard to tell what you meant by saying that this proof "asks too much". $\endgroup$
    – user21820
    Apr 28, 2022 at 18:29
  • 1
    $\begingroup$ Anyway, I don't even agree with the idea of making students reproduce any proof covered in the course itself. Rather, they should be able to adapt the key ideas in these proofs to solve completely new problems that fall easily to the same key ideas. That really is what we should test, not the ability to repeat something. $\endgroup$
    – user21820
    Apr 28, 2022 at 18:33

I'd certainly expect a student to be able to remember the more important proofs, and remember enough of more run-of-the-mill results to be able to come up with a proof (sketch) on short notice. But I'd also expect quite a bit more, in terms of being able to use them to solve (somewhat) similar problems that weren't covered in class. Given time constraints on exams, I'd probably leave out most of the "state and prove theorem XYZ" type questions. Some memorization is OK, but the focus should be elsewhere.

But then again, this is first term, and I'm spoiled by more advanced students anyway.

[In the above you can freely replace "proof" with whatever analogue may be applicable to some other subject area, be it Computer Programming or Italian Renaissance Poetry.]


This is a good way, in my opinion the best way, to teach somebody that actually wants to study mathematics, especially if they should have the goal of becoming a mathematician. I use the word teach above, since while the question is about the exam, this style of exam also impacts the way students (should) learn the material.

I can see nothing wrong with it. In various systems it is also not at all unusual, and not limited to some "elite" groups or institutions. Obvioulsy, in order to judge if some course is hard or not one would have to know more. But, just the general layout is not that unusual.

The reason why I consider this as a good way is that it makes it necessary for students to really engage with the theoretical core of the material, and not just learn how to do some excercices as they often are used to from secondary education. This is often a fundamentally different approach, and the first year of university seems like a good point for this shift of style.

  • 2
    $\begingroup$ I have no doubt that, in general, requiring students to know some proofs is a good idea. What I'm wondering about is whether it isn't a bit of an overkill to require every single one of the presented proofs. Do you think it's more beneficial than specifying a somewhat limited range? $\endgroup$
    – Ormi
    May 8, 2014 at 18:49
  • 2
    $\begingroup$ Yes I think it is better to ask for everything to be learned rather than starting to specify that this or that will not be asked. On the one hand starting to exclude things can encourage a learning only for the exams mentatlity and drawn out discussions and 'negotiations'. On the other hand, it is not clear which to exclude then. Typically one would want to keep the interesting ones, and the easy ones should become 'obvious' over time. A point of requiring evrything is that it should enforce understanding over learning, see a somewhat related answer on MO $\endgroup$
    – quid
    May 8, 2014 at 19:13
  • $\begingroup$ As far as oral exams go, requiring the students to know EVERYTHING presented in the course and probably simple extensions thereof must be the norm. Personally, it has helped me think through many things and organise my head. $\endgroup$
    – kan
    May 9, 2014 at 0:05
  • 1
    $\begingroup$ I disagree, and I refer to Benjamin's answer below. $\endgroup$ May 9, 2014 at 2:00

I myself am not very interested in seeing whether students (whether exceptionally talented or not) can reproduce long, complicated proofs. Or do long, complicated computations. I am interested in their awareness of what is true, and of the relative difficulty of proof, and of the prerequisites for a good proof. And awareness of the utility of the fact(s). To my mind, "the standard results" are the context in which one does mathematics. Re-establishing the standard context, as opposed to demonstrating awareness, is not very interesting or useful.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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