I imagine many people are familiar with the extremely entertaining article "Ten Lessons I Wish I Had Learned Before I Started Teaching Differential Equations" by Gian-Carlo Rota. (If you're not, do yourself a favor and follow the link I provided.)

I, a number theorist, have been told that I am to teach an undergraduate ODE class one year from now. (Nevermind that my familiarity with ODEs is, to put it mildly, minimal.) In a good faith effort to serve my students as well as possible, I am asking the following question.

Has any ODE textbook been written which addresses and assuages the issues brought forth in Rota's article?

Is there any way for me to teach ODEs next year and not, having read Rota's article, feel dirty about it?

  • 5
    $\begingroup$ I'm making this CW because I doubt there is one optimal answer, but: great question. I instantly love the Rota essay (I've read a fair amount of Rota essays, but this one is now near the top for me). I am reminded that I once taught the dreaded course in the dreaded way as a graduate student, and my friend from grad school days who at MO is alvarezpaiva also taught it that same summer, but infinitely better (probably from his own notes and insights, but maybe referring to Coddington and Levinson here and there). A thoroughly geometric course, and I hope he sees and can answer this question. $\endgroup$ – Todd Trimble Apr 8 '16 at 16:56
  • 6
    $\begingroup$ I like the question very much, but might it be better suited for MESE? $\endgroup$ – LSpice Apr 8 '16 at 22:06
  • 3
    $\begingroup$ A side remark: there seems to be a typo on page 9 about the convolution of two Dirac measures: it should be $\sum_{i,j} \delta_{a_i+b_j}$ rather than $\sum_{i,j} \delta_{a_i}+\delta_{b_j}$. $\endgroup$ – Fan Zheng Apr 8 '16 at 22:32
  • 2
    $\begingroup$ You sound as though you were surprised to learn that you'll be teaching undergraduate ODEs. If so, you've been badly misled about something. $\endgroup$ – Mark Meckes Apr 9 '16 at 17:29
  • 4
    $\begingroup$ >>I, a number theorist, have been told that I am to teach an undergraduate ODE class one year from now. (Nevermind that my familiarity with ODEs is, to put it mildly, minimal.)<< Graduate students in Number Theory (and other specialties) should hear this. Only 20% of math Ph.D.s end up at math departments large enough where they can avoid teaching differential equations (and other undergraduate topics). So when graduate students come up with the argument "I'll never need that in my research" there is this sensible conter-argument, "Mayb not, but you likely will need it in your teaching." $\endgroup$ – Gerald Edgar Nov 26 '18 at 11:54

There are a number of textbooks which are not as guilty:

  • M.W. Hirsch, S. Smale, R. Devaney, Differential Equations, Dynamical Systems, and an Introduction to Chaos
  • G. Teschl, Ordinary Differential Equations and Dynamical Systems

The last one is also freely available here. As the titles say, both take a rather geometric path and give many examples from various sciences.

The first one waits until the final chapter to give the classical existence and uniqueness theorems.


My advice would be to teach the traditional course. Especially because you lack knowledge of ODEs themselves. Deciding that you are going to do something different when you lack knowledge, experience of the topic and likely sympathy with the objectives of the course (teaching engineers).

Rota has a fun to read essay and there are parts of it that I like. For example 2nd order ODE with constant coefficients is very common in science and engineering. however, even here the course really already covers it heavily. In some cases repetitively: calculus primer on ODEs, ODE itself, and then in majors courses in physics, chemistry, engineering. I would also emphasize that Rota's book never had great commercial success either as a common text like Brown or an underground Amazon success like Tenenbaum. Read the reviews of the Rota text on Amazon for some perspective.

Once you are a little more experienced perhaps it makes sense to do something revisionist but I would be very hesitant of the "I'm a number theorist and will teach all the engineers their ODEs differently since it was all wrong before...even though I don't even know much about the topic or about teaching it"

  • 2
    $\begingroup$ +1. Use a text that has been used for that course in your department in the past. I am thinking that if you blindly pick one of the texts mentioned in these answers (even the one in my answer), you will end up with something that is too hard for your students. $\endgroup$ – Gerald Edgar Nov 26 '18 at 11:57

I like these two differential equations books:

  1. Differential Equations and their Applications, by Martin Braun (Amazon)

  2. Ordinary Differential Equations, by V.I. Arnol'd (Amazon)

I don't claim they meet all Rota's criteria, but I think they are somewhat fun and offer insight. Braun is especially recommended for examples like how to detect art fraud by the age of the paint, the galloping gertie bridge disaster in tacoma (although he slightly erroneously names the culprit "resonance"), and predator - prey examples involving sharks in wartime.

The main thing to make ode seem appealing to me is just to point out an ode is a vector field wanting (a family of) parametrizing curves. Arnol'd offers this sort of geometric appeal. It was after reading Arnol'd that I understood why Reeb's theorem is so intuitive, (a compact manifold with a function having just two non degenerate singular points is a sphere).


When I was a grad student, our department taught ODEs in the "bag of tricks" style. Guilty! As a TA I even hammed it up doing an impression of Emeril -- bam! -- saying we were making a recipe book of techniques to attack problems.

But when I was an undergrad and originally learned the material, it was taught in a vastly different way. I took ODEs with Borrelli at Harvey Mudd. They referred to their own book often enough; it's more applications oriented. But we also spent a lot of time with the workbook.

It seems there's a second edition, and Wiley has "companion" sites for the book, instructors, and students. See:

BTW, the workbook link above is to the CODEE site. CODEE is the Community of Ordinary Differential Equations Educators. The CODEE site has a lot of useful resources.


I would suggest

ArnoI'd, V.I. (Vladimir lgorevich) Geometrical methods in the theory of ordinary differential equations. Springer (1988)

of course, it's not to be taken "as is" for an undergraduate ODE class, but a lot of modern important examples (taken from nature) are given and discussed. Overall, there are a lot of well illustrating and expressive drawings which can inspire the lectures.

  • $\begingroup$ Isn't this a graduate level text/monograph? $\endgroup$ – Chris Judge Apr 12 '16 at 19:40
  • $\begingroup$ Yes, that's why "it's not to be taken "as is" for an undergraduate ODE class", but to feed that preparation. $\endgroup$ – Duchamp Gérard H. E. Apr 12 '16 at 22:43

As @AndreiHalanay says, probably the best that one can do here is to give a not-guilty textbook, rather than a perfect one. I haven't seen Teschl's book that he recommends, but I have read part of the Hirsch–Smale–Devaney one. Much as I enjoyed it personally, I got the impression it asked a bit too much of students for me to use it.

A book in a similar spirit (no surprise, since it shares an author), but a little less demanding of students—which it accomplishes largely by being less ambitious (those used to teaching out of Boyce and diPrima will be shocked by the fact that, for example, it treats integrating factors only for linear equations)—is Blanchard–Devaney–Hall. I haven't specifically compared it to Rota's list (which rhymes very well with my woes after having taught the course for years), but I did try going back to Boyce and diPrima once after having taught a few times from BDH, and it really made me appreciate the latter.

EDIT: Having now refreshed my memory of Rota's essay: I had forgotten that he specifically disclaims the utility of integrating factors, so I guess my remark about them can be taken as an endorsement of BDH. I am also disappointed by what seems to me to be BDH's sparse treatment of change of variables; at least they do not explicitly describe the importance of this technique.

  • 2
    $\begingroup$ My department uses the BDH book. It has its detractors, but it fits Rota's advice pretty well, which is one (or ten?) of the reasons I personally like it. $\endgroup$ – Mark Meckes Apr 9 '16 at 17:27
  • 1
    $\begingroup$ @MarkMeckes, I like it not so much because it adheres to Rota's philosophy—although, as I say, my impression is that it does—but simply because using it makes me feel more like I'm teaching mathematics rather than a bag of one-shot tricks (even though I do sometimes miss some of the tricks, despite Rota). I also have to say I'm a sucker for the intrusion of authorial voice; it could be oppressively corny or intrusive, but instead it's just pleasantly corny—like having a conversation with someone who is very passionate about the subject, which seems to be accurate. $\endgroup$ – LSpice Apr 9 '16 at 18:51

I am in the same department as Mark Meckes, and have enjoyed using Blanchard, Devaney and Hall over the decade I've been teaching the course. I appreciate its emphasis on qualitative methods and insights, perhaps because these are the aspects of differential equations I use in my own work (in mathematical neurosocience). Having just read the Rota essay, I don't see it as giving any very particular recommendations vis-a-vis what should go into a textbook. The lectures are where I try to bring the material to life for the students, whatever textbook I use, and the course is more than just the lectures and the textbook. (I recall an evaluation once by an angry student who said if it weren't for doing all the homework problems he would have learned nothing from the course -- as if just listening to lecture and reading the text were supposed to suffice...)
So -- I endorse BDH.

  • $\begingroup$ I came back to this post while thinking over advice to a student. I'd add to this response that I like BDH's emphasis on qualitative methods and insights because those are precisely the aspects of differential equations that we can't just rely on computers for. (More to the point: those are the aspects that our future-engineer students won't be able to just rely on computers for.) $\endgroup$ – Mark Meckes yesterday

I read Rota's essay today, having already taught the "cookbook" ODE course 5 or 6 times. While the essay carries a lot of wisdom, I think the answer of Nov 25 '18 at 23:18 by guest is absolutely the most useful.

With non math majors, you must strike a balance between concrete and conceptual and most of the suggested books are way too advanced for this level (you will lose most of the students).

Finally, while I guess I subscribe to Rotas arguments, another perspective might be that teaching (some) tricks is not necessarily a bad thing especially if you can deconstruct the tricks for the students. There is nothing dirty about such an exercise well done.


"ODE tetbook following Rota" suggests to me

Birkhoff, Garrett; Rota, Gian-Carlo, Ordinary differential equations, Introductions to Higher Mathematics. Boston-New York-Chicago: Ginn and Company, VII, 318 p. (1962). ZBL0102.29901.

But of course that is too advanced for an undergraduate students (except perhaps at places like MIT).

  • 8
    $\begingroup$ If you read Rota's ten lessons, you will find that Rota is claiming guilt about exactly that book, so this is precisely the most wrong answer possible. $\endgroup$ – Ben McKay Nov 26 '18 at 12:15

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.