On an interview of Klainerman

Oliver Knill,


I had been asked to comment about a recent interview with Sergiu Klainerman. Well, the email asking for feedback was most likely a clever marketing trick for spreading the article. But there are some interesting points made in that interview and I would like to comment on two.

Climb the mountain first, the meaning will come

The interview with Sergiu Klainerman was an interesting read for me, as I'm also an European bound mathematician. I myself am a teacher primarily now and not a research mathematician, but there are some interesting things for me which appeared in that interview.

One mention is a prevalent mantra among students ``climb the mountain first and then the true meaning of the subject will come". I actually think this is very good advice. It does not look so first, but an important thing when learning math is to keep doing it and not get discouraged. The above mantra helps. Klainerman himself proves that that it can be a good approach as he became a fantastic mathematician. I myself felt a similar frustration at first as a student at ETH Zuerich. We were fed a lot of information and I missed a bit the creativity I was able to pursue as a high school student. But I appreciate now that learning a lot over a relatively short time was good. Knowledge allows us to build connections and be creative later on. We need a rich background of knowledge in order to grow later.

If we learn a language, or a music instrument, we first practice a lot of old stuff, we read and play the masters for many years, before we go into composing ourselves or try to write poetry on our own. We also just might enjoy a novel or piano concert without first analyzing how it is built and why it works or make an effort to modify it. Understanding is great, creativity is even greater but it can be frustrating for a beginner. Once too much frustration kicks in, the case is almost lost as mastering a subject requires to pull lots and lots of time doing it.

Understanding only comes on a fertile solid of knowledge. There might be psychological reasons playing a role here but the rush to force young kids to understand or to ``be creative" early on can also be a turn-off. Creativity is hard. If you do not believe this, try to be creative and come up with a new theorem in geometry or number theory on your own. Most people who preach creativity have not tried to be creative themselves and have no idea how damn hard it is. Now, we should practice being creative and try as hard as we can. I myself was most successful in this with final creative projects like composing a music piece using a computer algebra system or to build 3D printable objects. But I'm constantly reminded when doing such efforts that being creative is hard for most. During my final (mostly oral) examinations as an undergraduate student I was asked the hardest question of all by Ernst Specker, who tested me in logic. His question was: "Mr Knill, tell us something". It was a hard question to answer and it illustrated me how difficult it can be to tackle open-ended questions.

Also true understanding can be hard. One reason of course is that understanding needs a lot of knowledge beyond the actual subject. For a student, it can be frustrating to hear: ``but why do you not understand this?". It can be very tough to understand, it is even tougher to teach ``understanding". For anybody who does not believe this, it is a good idea to pick up from time to time a math book in a subject one is not familiar with and then solve the task to ``understand". A good approach is to actually solve the problems which are provided in the book without looking up any solutions or ask for advise. It is a good exercise because in most cases, one feels exactly like a student who is first is exposed to some new subject like algebra, geometry or calculus. One first has no clue at all what things are about. It is the feeling Klainerman points out when reading Lars Hoermander or Fritz John. It is what I experience myself if I look now at a paper of Klainerman. I would definitely need to go back and brushing up much more harmonic analysis and functional analysis and learn inequalities to master the reading.

As an educator myself, I feel, that neglecting to teach substantial amounts of material is one of the most important mistakes of modern math education and one of the reasons why other countries do so much better in assessments. We are still indoctrinated as teachers that teaching content is bad, that conceptional understanding has to be fostered before we know how to do things. This is even entrenched in the famous Bloom taxonomy, where ``understanding" is placed before ``algorithmic doing". But understanding can only come once we know how to do this. The reason is that for understanding, we also need to build up interest. We first learn how to drive a car before we know the inner workings of a combustion engine or an electric motor. Actually, by driving a car we might become interested in the inner working of these engineering miracles. Before we do it, it is just some object moving around. We need to have written a few programs in a programming language before starting to understand why the syntax is built as it is or how the compiler for the language works.

Now, Klainerman himself proves the point. Like me, he was frustrated to have not reached good understanding due to all this ``content learning" but by staying ``hungry" and having been provided with a lot of fertile soil as a student, he could grow nice mathematical plants later on. It is the holy grail of education to pass along ``understanding". It is my experience from 30 years of teaching and 50 years of learning mathematics that understanding is difficult to teach, especially if not first some ``how to" is done. Yes, a good teacher can implement understanding also in the early stages of introducing a subject, but not many succeed like that. The vast success of ``how to" guides on the internet prove this.

Learning something well requires some memorization and drill. This sounds like heresy. But we would never question this in music for example. You can not perform a piano concert well if you do not know it by heart and have played it a thousand times. Nobody would tell there ``you shall not memorize" or not practice ``etudes" (finger exercises). Also a rock climber needs to memorize every move of a difficult route as a non-choreographed approach will fail. The recent movies ``Dawn wall" and ``free solo" illustrate how much work and repetition is needed to master a climbing problem. In mathematics, even seemingly boring things like the multiplication table are treasure troves. There are patterns to be discovered, the concept comes back in multiplication table of abstract groups etc, but familiarity with a subject is a prerequisite to become interested in the subject and as we all know, being interested in something allows to learn with almost no effort. If we are interested in something, we even enjoy doing it.

The role of partial differential equations

I very much agree with the Klainerman assessment that partial differential equations are an important topic of mathematics and that it unifies a lot of topics. Partial differential equations are the fabric we are made of. It has connections with lots of other fields of mathematics like geometry and probability theory. It is usually located in the field of analysis, especially functional analysis but there are also connections to combinatorics or algebraic geometry. The name Klainerman has come up at first when taking a graduate course in nonlinear wave equations given by Michael Struve at ETH. It is a subject, where local and global existence of initial value problems come up similarly as in fluid dynamics or in general relativity and this is where Klainerman made important early contributions. It had been interesting for me then as a graduate student because I had been interested in integrable partial differential type systems.

The subject of partial differential equations can become very technical and the work of Klainerman illustrates this. It is also an area where mastery of basic skills, knowledge of many inequalities and analytic techniques are pivotal. This does not work without learning lots of stuff first. The field of partial differential equations is huge, one of the richest and nicest in the entire field of mathematics. It is also a show case for a tool when doing physics. The recent discoveries in physics by the event horizon telescope has invigorated the interest in partial differential equations appearing in relativity. And then of course there is the one million dollar Millenium problem for Navier Stokes. It even has come up in pop culture like in the movie ``gifted" from 2017.