Hacker News new | past | comments | ask | show | jobs | submit login

Can I give a very practical advise to people who are reading this and trying to learn math ? As someone, who received a very strong mathematical training in a former Soviet Union, here is my practical advise:

1. Calculus books, just like this one, are absolutely impractical in real life situation, especially, if your goal is "Industrial Mathematics". All you will learn, are basic calculus notations. You will, at best be able to solve very basic toy problems. 2. Instead, learn basic algebra and combinatorics on extremely proficient level. This is what often is missing in US education.

In order to get to 2. 3. Learn how to do a. complex algebraic manipulations, b. solve complex algebraic inequalities, c. basics of number theory, d. combinatorics. Notice, nothing going beyond Real Numbers and I'm not even including Euclidean geometry.

4. Best sources for that are Math Olympiad problems and technique to solve them. You will learn how to crack extremely complicated algebraic expression, how to factor them and represent them in different forms, how to do tricky substitutions. Same technique is applicable in working with complicated integrals/diff. There is an entire layer of mathematics that devoted to inequalities and they are very applicable in solving calculus problems. Most of the technique and materials to solve those problems aren't taught in high schools and even college course.

Being able to solve moderately complex algebraic problems is must before learning calculus and analysis. Crush your ego, google/amazon for books and materials on how to solve (basic) Olympic problems that are intended for HS 9-12 graders and see what you can do.




This is absolutely wonderful advice for a high school student who wants to get a good foundation in STEM.

I'm not sure that it is applicable to an adult who needs a rough and ready understanding of Calculus.

I personally taught my brother enough Calculus to take a course that had it as a pre-requisite in under an hour. What did I focus on?

  1. The idea of approximations.
  2. The tangent line.
  3. How the tangent line connected to approximations.
  4. The derivative.
  5. The easiest formulas for differentiation and why they are true.  (All handwavy, heuristic big-O arguments.)
  6. That all possible max/min points can be found at the boundaries, or by finding where the derivative is 0 or non-existent.
  7. The Fundamental Theorem of Calculus aka why areas are the reverse of derivatives.
  8. The advice that if he had to actually calculate a derivative or integral, he should use a program like MAPLE.
Did he master the subject? Heck no!

Did he have to review his notes a bunch of times so it stuck? Of course!

But he went on to ace the course. And my guess is that he understood what makes Calculus tick better than most who took the course. (Sanity check. If you do not understand why the tangent line and derivative are connected, then you do not understand Calculus.)


Counting Differential Equations, I've had 5 semesters of calculus. (Calc I,II,III,IV, Diff Equ.) Until reading your comment, item 7, I had never heard this simple explanation of the meaning of the Fundamental Theorem of calculus ("why areas are the reverse of derivatives"). I was taught the fundamental theorem algebraically, and how to apply it, but none of my professors or textbooks ever explained what it meant.


You never had someone in your introductory course show you some problem where you have e.g. a big water tank filling up, and you relate the flow rate with the volume of water in the tank, describing the relation as either sum(flow) = volume or diff(volume) = flow?

That’s quite depressing, since this is really the whole point of calculus.

* * *

Calculus really should be introduced with as much emphasis on physical modeling as possible. Differential equations are at the heart of the past several centuries of science, and understanding the basic ideas involved is crucial for everyone doing any kind of technical work, if not every citizen.

I like this version, which uses discrete computer simulation to cut down on some of the obscurantism of a formal heavily algebraic treatment, and lets students jump right into ideas which are significantly delayed in typical university math curricula. http://www.math.smith.edu/~callahan/intromine.html


I guarantee you it was in your textbook and probably in lecture and you just glossed over and forgot. What does "applying it" even mean if not "finding the intrgral of a dervative or the derivative of an integral"? That's the entire point of the theorem, which is given the most powerful name mathematics pedagogy gives to important ideas, which should motivate you to glance at it more than once.

Or do you mean that you never connected "derivative and integral are inverses" with "integrals are areas" and "inverses are opposites"?


That's really sad.

You might check out the sequence of activities in the Integral chapter of this book to get another perspective on the FTC.

I totally agree. It's the entire point of calculus, and it's what sets the entire field of Differential Equations in motion!


Thank you for your honesty.

I wish it surprised me. But I suspect that your experience is more normal than not.


I think a critical ingredient for him was that he was motivated to get the exam passed. Otherwise new terms (unknown, not heard before, so scary and opaque) would overwhelm and attention would be lost.

How you did that introduction is very important. Every time you tell something new, it better be really small - or explained quickly and well, so the concept would stick before brain would get tired.

Thank you for listing your points.


I think that it really helped that back in grad school I read https://www.docdroid.net/z8ki/knuth.pdf and thought hard about Knuth's ideas. I even went so far as to make a course outline for a first course in Calculus based on his ideas.

The result is that I was able to break my exposition into one piece at a time, starting with ideas that were already accessible. More specifically I started with the idea that f(x) = approx(x) + error(x) where we want approx simple and error small. This motivates a language for describing what "small error" means, which motivates little-o for polynomials.

That's enough to do tangent lines and answer questions like, "Given the equation for position vs time, how fast is the rock going when it hits the bottom?"

Only after he could calculate tangent lines did I introduce the derivative.

Contrast to the usual approach where limits are an abstract concept with no obvious application, and then the derivative is introduced. It actually combines several ideas jumbled together. That's a big mental knot that almost nobody gets. And explaining it more carefully doesn't help because people keep getting to holding too many unintegrated ideas in their head at once.

And seriously, heuristic arguments that you can reproduce whenever you forget them are good. Take the product rule.

  f(x0+h) = f(x0) + f'(x0) h + o(h)
  g(x0+h) = g(x0) + g'(x0) h + o(h)

  (f*g)(x0+h)
    = (f(x0) + f'(x0) h + o(h)) * (g(x0) + g'(x0) h + o(h))
    = f(x0) * g(x0) + f(x0) * g'(x0) h + f(x0) * o(h) +
      f'(x0) h * g(x0) + f'(x0) h * g'(x0) h + f'(x0) h * o(h) +
      o(h) * g(x0) + o(h) * g'(x0) h + o(h) * o(h)
    = f(x0) * g(x0) + f(x0) * g'(x0) h + o(h) +
      f'(x0) * g(x0) h +                o(h) +           o(h) +
              o(h) +            o(h) +        o(h)
    = f(x0) * g(x0) + (f(x0)*g'(x0) + f'(x0)*g(x0)) h + o(h)
And we recognize the form of the tangent line and so the derivative of f*g is f' g + g f'.

I had him do that calculation with only minimal prompting. And I think that this is a calculation that I'd like every Calculus student to be able to do on demand. If you forget the rule and you know that argument, you can figure it out again.


I'm genuinely curious what field you work in and what problems you work on that IMO level algebra and combinatorics skills are frequently useful (IMO = the olympiad, not my opinion).

In the fields I have experience with, the usually approach when faced when something gnarly involves a lot of "to first order," or "assume X is much greater than Y," or simulation.

I'm somewhat doubtful that your advice is widely applicable if the necessary skills aren't taught in college courses.


To answer your question, I work in optimization.

IMO (International Mathematical Olympiad) are extremely complicated problems that even professional mathematicians often struggle with them. IMO is a level on its own - Gold Standard. Not all Mathematical Olympiads are of the same level of complexity as IMO. Good example are Olympiad caliber problems that are not overkill - Hungarian Problem Books.

MO problems are useful, because they make you think out of box, they often times involve several branches of math in one problem, such as number theory problems go in hand with combinatorics; they don't require complex mathematical machinery, and technique of solving problems, directly translates to solving complex problems in analysis/abstract algebra.


But industrial mathematics is numerical methods, a rather different beast from the tricky closed form puzzles of Olympiads. I question your claims of causation vs correlation regarding Olympiad work and industrial math and general effort put into basics of mathematical thinking and general mathematical intelligence.


Not entirely true. A lot of optimizations theory and practice are are straight-up inequalities. My claim is very simple: by being able to solve moderately-hard MO problems, you are developing excellent mathematical problem solving skills that let you solve problems in other branches of mathematics or draw interesting connections, because MO problems require ability to solve problem that are at the intersections of different branches of mathematics.

I frequently see topics similar to this one popping-up on HM and a lot of people interested in learning or re-learning math as an adults. Excellent ! So let them learn discreet mathematics by solving problems in algebra, combinatorics and number theory. Once they are comfortable, they can move on to more abstract subjects. My view is very similar to Concrete Mathematics, by Knuth, Patashnik, Graham. Their "concreteness" is very down to earth: combinatorics, number theory and few other things in the mix.


As a working mathematician who has rarely managed to solve an IMO problem, I have to say this isn't the best advice for everyone, though I agree that focusing on linear algebra and combinatorics is probably a better use of one's time.


I agree Linear and Combi is far more useful once you get going on a technical degree and/or career, but go to a university in the US and check the prerequisites on these courses: CALCULUS.

This was another reason I wrote this book. For people who just need to get through calc, here's some help that you can pick up and read in a couple hours.


Can you propose a specific curriculum?


The Art of Problem Solving series of books is pretty much exactly what was described, is relatively affordable to an adult with disposable income, and they all come with complete solution manuals: https://artofproblemsolving.com/store/list/aops-curriculum

You could spend a year or two doing ~20 problems per day while scheduling review of definitions you've understood and problems you've solved with spaced-repetition software like Anki.

Afterwards, you'd be prepared for any undergraduate mathematics curriculum in the world.


There are numerous books devoted to Olympiad preparation, there are websites completely devoted to Olympiads, there are freely available problems sets (with complete solutions). To name the few: Hungarian Problem Books, Problem-Solving Strategies, Challenging Problems in Algebra, books by Titu Andreescu - a former US IMO coach, who published a lot of books on math and IMO prep, http://www.cut-the-knot.org/


Thank you for your suggestions. I purchased Challenging Problems in Algebra and a text book by Titu Andreescu. Looking at the kinds of problems in Olympiad, it seems like they'd actually be helpful and more fun to solve.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: