Ask HN: Serious mathematics books that can replace a good teacher?

288 points by newsoul ↗ HN
Mathematics is best learned under the guidance of a mentor. But not everyone has access to mentors all the time. That's where books come in. Good books. Books that can be substitute for a mentor or sometimes even better.

Which books (preferably not pop-sci) fall into this category?

173 comments

[ 3.6 ms ] story [ 226 ms ] thread
J.Steward's "Calculus: Early Transcendentals" is pretty on point for calculus. Mathematics is more than calculus though.
Do you have specific mathematical topics you're interested in?

I have tried reading textbooks a few times to teach myself but found it hard to stay motivated, so I found a tutor on Upwork to assign and grade homework problems and answer my questions. Along with math textbooks and YouTube videos, this has been super helpful for fleshing out my knowledge of college-level math that I never learned properly. It's also great because they can go at the pace you want and focus on topics you find challenging, interesting, or useful.

Do you have favorite YT channels that you can recommend?
I believe Richard Feynman has mentioned in interviews that he learned calculus from "Calculus Made Easy".

IIRC there might have been other books in a "Made Easy" series about math, but I'm not sure.

That's one of the books my grandpa had that I wish I had managed to get before that whole collection got dispersed because I remember him reading it over like a ten year time period. He just liked repeating variations on the exercises I think
He also liked "Advanced Calculus" by Woods (possibly not for his first calculus text, but it happened to be six inches from my hand as I read this.)
I liked that book. It uses differentials to give simple explanations of various derivatives, which seems to be frowned upon these days. You can download the book for free if I'm not mistaken. I once recommended it to a calculus class I was TAing for, though I don't think they were very impressed.
Burn Math Class by Jason Wilkes is interesting.
Seconded. also consider Lockhart's "Arithmetic". Bear in mind. These are both non-orthodox style books. Burn Math Class least orthodox.

An orthodox approach to Analysis would be Terence Tao's Analysis books.

Further recommendations depend on your specific goals.

Lockhart wrote another book, called measurement, with a different scope. It covers Geometry and Calculus in his whimsical style. It can give you many aha moments.
This is a nice recommendation! The chapters on circles and Trigonometry were amazing. But, man, it takes a lot of patience to slog through the dialogues.
The question reminds me of Susan Rigetti's recommendations for math self-study. If your goal is to self-study the equivalent of a university undergraduate mathematics degree, this is one approach: https://www.susanrigetti.com/math
his recommendation is quite useful, I've read most of it and found most of them are approachable
Susan Rigetti is a woman.
oops

I found that she was an SRE engineer at uber, later she change corse to full time writing at some newspaper agency, what a change.

From physic to software engineering then fulltime writing, not anyone can do that

> she was an SRE engineer at uber

You might want to look into what she had to go through while she worked there. It's well known.

Also, Ms. Rigetti has similar guides for physics[1] and philosophy[2].

I intend to slowly go through her math guide over the years. I already started working through How to Prove It by Vellman.

[1] https://www.susanrigetti.com/physics

[2] https://www.susanrigetti.com/philosophy

I expected to see something about probability and statistics but they're hidden with dozens of other topics in the electives > Any and every topic imaginable. Are probability and statistics not part of a regular mathematics curriculum?
I ended up with a math degree without taking any of these classes.
No, they're fairly niche topics as far as the rest of mathematics is concerned. You'll see much more emphasis on them, and especially on statistics, in an applied math curriculum.
In this matter, my opinion of statistics is that it does not really teach 'math', it teaches how to use 'math' to derive information. Pretty much applied math.

I see this question as 'Why does math students not take Thermal Mechanics' or 'Why does math students not take Nuclear Methods'. The answer is just that, they do not teach math, they teach an application of math. Your math knowledge has not really grown.

That's quite close to saying that applied maths is not maths, which many people would disagree with. If you have definitions, theorems, proofs, I'd say that's maths, and you definitely have that in a sufficiently rigorous statistics course.

Of course, then there's the fact that statistics builds upon probability theory, and probability theory is, in a sense, a subfield of measure theory which in turn is about as mathematical as it gets (in the discrete setting, it also includes a fair amount of combinatorics).

Pretty much applied math.

Well... yeah. I mean, I might be wrong, but my default assumption is that when people on HN ask about learning math, unless they explicitly say otherwise, they are mainly interested in maths from an applied viewpoint. That is to say, I think most such inquiries are rooted in a basis of "I want to learn the math require to DO 'x'" where x might be "machine learning" or "circuit analysis" or whatever, as opposed to "I want to become a mathematician and advance the overall state of mathematics as a field."

I say that at least in part because of an assumption that people who want to become mathematicians per-se are probably asking their questions on Mathoverflow or whatever, and not HN.

EDIT: to be fair the specific sub-thread we're in here does contain this, which I guess justifies taking a "pure mathematics" position in this part of the overall discussion.

Are probability and statistics not part of a regular mathematics curriculum?

Still though, this seems to be a general issue with any maths related discussion on HN. It seems like a lot of people are commenting from a position of assuming that the initial question was based on an interest in pure / theoretical maths and the "I want to become a mathematician" idea. And I am somewhat skeptical that that is normally what's intended by the person asking the initial question.

It will depend on the university. Where I studied, an introductory class to probability and statistics was mandatory, as was measure theory where probability theory was at least hinted at.

Other universities may have even stronger probability theory requirements, or none at all. But it's certainly not an uncommon specialisation.

Note that probability theory is not necessarily the same thing as statistics. While the latter builds upon the former, statistics is more about "given the data, what's the most likely distribution ", as opposed to "given the distribution, what's the data gonna look like" (probability theory). For mathematicians, the latter seems to be more relevant, as it's a more deductive form of reasoning.

Spivak's Calculus! (It's really a book about real analysis.) It's extremely well written and starts by rebuilding your understanding of your fundamental mathematical building blocks, only using things you can prove. It also teaches you how to prove them.
It is a long time since I learned, so I dont have specific book, but dont underestimate exercises. Whatever math topic you are learning, look for one of those books with only exercises in them and solutions in the end.
And an active author/publisher who still posts corrections, or on older book. Maths books tend to have errors in the solutions.
I really like Arthur Benjamin's work on mental mathematics. I'm not savant-level, doing division in the thousands or huge floating points in my head yet but I sure am a lot sharper than I was coming out of high school from studying his work, and I guarantee you will just have fun with expanding your capability to think about numbers. [1]

I got a copy of this book from the 1920s which is really cool because it teaches you math lessons you have to actually go out and physically do stuff with like pegs and strings in a field, from the perspective of the history of mathematics where people were limited to such devices in order to do stuff like trigonometry. Very very different approach, probably not for everyone, but for me I just think it's pretty cool. It definitely was written in the 1920s though so you better get used to that particular writing style if you plan on digesting it like a course. It's designed that way, though, and it's got great reviews. Just keep in mind maybe some of the history is subject to have changed over the years. [2]

Ultimately I've self-taught myself a lot more than I ever learned in school for sure but a wide variety of sources is probably more what you're after in terms of getting a grip on what's interesting enough to pursue further for your own means and ends. I think exploring what fascinates you the most and then just going and finding things from that point is a pretty good start as long as you've got elementary understandings up to a point where the fascination actually happens.

[1] https://www.goodreads.com/book/show/83585.Secrets_of_Mental_...

[2] https://www.goodreads.com/book/show/66355.Mathematics_for_th...

I had two acceptable maths teachers in my life. Good teachers seem so rare.
No. Books can't understand how you think and learn and help guide you. There is no substitute for a good teacher. Is there a best book for that student? Probably, but who is that student? We don't know!

Beyond just learning mathematics, seeing the art and beauty in it is also best taught by someone who knows the subject and the student. Without the beauty, it's just what could be in a textbook, assuming you found the right book.

If you're asking if you can find a good book to teach someone, that depends on your style and theirs . . .

We will almost certainly have good student-focused AI teachers during our lifetimes for things like math and languages. Will AI be able to show us the art? I can't say for sure, but I bet so . . .

> There is no substitute for a good teacher.

Note that this does not mean one cannot fully learn a subject from books or that the average teacher is better than books or that a good teacher dealing with 30 other kids will be better than a good book.

Of course you’re not guaranteed a good teacher by studying the conventional way. Studying from books is a great skill to develop regardless.
Nothing can replace a teacher, much less a good one, but I recently stumbled upon an introduction book on calculus from 1910 (yes last century) that had a really nice way to explain things.

I just read the beginning so I don't know how far it goes but anyway here the link https://calculusmadeeasy.org/

What type of mathematics? What level?

I taught myself all of A-level maths and further maths here in the UK from the standard text books at the time (Bostock and Chandler) before I started in the sixth form and then maybe about half of the first year university material before I went. Still better when you have somebody to teach you but not impossible. I did have access to somebody whom I could ask questions but didn't really use that.

It's going to very much depend on which field you are interested in. The subject is huge. If you want to study statistics or number theory, you're going to be looking at very different skills and knowledge, with only basically high school maths in common.
That's not quite true, as there are overlaps. To name just one big topic: Pseudorandom number generators. Here you have a (number) theory, including things like finite fields, to generate and understand deterministic numeric sequences, but also a lot of statistical methods tonassess whether these "look" random. Knuth has an entire chspter on these.
I'd honestly spend my energy on finding a platform which gets you closest to a mentor. Books are good, but in no way, shape, or form a substitute for a mentor. Mentoring is two-way communication, a books is almost always a one-way.

Forums, chat groups/channels, pen pal, whatever it takes.

One reason for a good tutor is having some direction. I feel this is like asking for a "serious computer book". Do you want a book on C++ or Dummies guide to Windows or regular expressions or Turing machines or chip design or functional programming or hash functions or ...?
Semi-serious question; how come you never hear of a 10 year old math prodigy where someone stopped them at age 4 and said "ah, ah, ah I refuse to tell you anything about math until you've decided what you want to learn specifically and what you are going to use it for", yet it's everywhere on the internet, multiple times in every thread on all kinds of topics?

Yes, I know why people think it's a great comment, what I want to know is given how many of us learned about computers as children before we had any idea about functional programming or Turing machines, how do we get recommendation threads to move away from this?

Honestly, with a child I would give them a book on a topic I think they would enjoy. If an adult said "suggest a book I might like", I'd also give them something on a topic I thought was interesting.

Saying you want a "serious maths book" really is just too large a range. I could suggest an excellent book on computational group theory (my prefered area of maths), but honestly for most people that wouldn't be something they'd want to read.

- Linear Algebra: Gilbert Strang videos + book

- Calculus and basics of Analysis: Spivak

- Group Theory: Visual Group Theory

- Complex Analysis: Visual Complex Analysis

- Writing proofs: Proofs: A Long Form Textbook by Jay Cummings

- Abstract Algebra: Dummit-Foote book. First few chapters if you like.

___

Pop-Sci:

1. The Joy of X by Steven Strogatz

2. Fermat's Enigma by Simon Singh

(All this books except AA are very highly recommended by me)

After Strang I recommend working through Hoffman and Kunze
Hoffman/Kunze have zero as an eigenvector for reasons I can’t quite fathom.
Agreed, that is weird. I like the rest of the book though.
If someone has a nonstandard analysis text that they endorse as really good, I would love to hear about that. It would be a shame to tell a kid about deltas and epsilons.
The best book for me was "Vector Calculus, Linear Algebra, and Differential Forms" by Hubbard and Hubbard.

It is the only book (I know of) that brings you from absolute basics to an integrated development of the subjects from its title. And that integrated developments actually leads to a good didactic method.

The search function will show you many comments on HN recommending it.

I hear they have an interesting take on the implicit function theorem.
Buy from matrixeditions.com for $98.
> Mathematics is best learned under the guidance of a mentor.

I disagree. The most important thing is curiosity.

For math, you need to think, question, and play with the concepts. It is much more active than just reading. Develop an intuition by seeing what happens when you maximize or minimize a parameter. Think about the consequences if something were different. Try to derive relations for fun.

This also applies to music and code.

If you figure out how things work on your own, you will retain it far better than going through the motions mechanically.

> For math, you need to think, question, and play with the concepts.

And having someone to answer questions and validate hypotheses makes the learning process a lot smoother and helps correct mistakes before they solidify.

> The most important thing is curiosity.

I mean yeah, but a good teacher can point out your mistakes that are not obvious to you at the entry level. When it comes to i.e. music, such mistakes can make a bad habit that impacts the following development (wrong hand position, wrong fingering, etc). I don't know if that extends to math though.

Dummit & Foote is fantastic for algebra. I also really liked Carrothers for analysis. None of the other books I used really stood out, other than Lang's algebra book, but that one was for the opposite reason - once, he said "the following is obvious" and I thought, "oh yes, it is!" and it was one of my proudest moments as a math student. Would not recommend for self-teaching.

More important than a good teacher, imo, is collaborators you're learning with who are around your same level. Sometimes they should figure out exercises faster than you, sometimes you should figure things out faster than them, often together. If that's not happening, it's a lot harder (and more frustrating) to learn. This becomes more important the more advanced you get. I never found lectures useful really but if I had no one to collaborate with I was almost certainly going to drop the class.

Haven’t read this one, but has been recommended highly to those trying to pick up rigorous math starting from non-quant training:

https://longformmath.com/analysis-home

>This book is the first of a series of textbooks which I am calling “long-form textbooks.” Rather than the typical definition-theorem-proof-repeat style, this text includes much more commentary, motivation and explanation. The proofs are not terse, and aim for understanding over economy. Furthermore, dozens of proofs are preceded by "scratch work" or a proof sketch to give students a big-picture view and an explanation of how they would come up with it on their own. Examples often drive the narrative and challenge the intuition of the reader. The text also aims to make the ideas visible, and contains over 200 illustrations.

>The writing is relaxed and includes interesting historical notes, periodic attempts at humor, and occasional diversions into other interesting areas of mathematics.

Have read this one, a book on advanced high-school math with a ‘problem-solving’ bent from a passionate teacher.

https://www.amazon.in/Educative-Jee-mathematics-PB-Joshi/dp/...

Let me ask and answer some questions here that doesn't answer your question directly.

What makes a good book and why there are tons of them on the same subjects?

The best book for you is the one that speaks to your technical preparation and perspective. A few hits the sweet spot for a broad audience - perhaps because they are good at drawing analogies with common experiences - but even some obscure books can be good if it aligns with what your background.

How can a mentor help and can you do without one?

A mentor can help lay out the roadmap to build from simple topics to more difficult ones. Maybe more critically, provide rapid feedback on your understanding. They can also explain things in more than one way. Some textbooks do lay out the roadmap reasonably well, provided that it starts from concepts that you are already familiar with (again, you need to find the right book for you). Problem sets in textbooks are meant to provide feedback on your understanding, but it often fails to provide smaller hints if you can't solve the problem outright. You could get a set of solutions for the problems and that could partially help. Grabbing multiple textbook on the same subject can also help understand the most commonly covered (and by implication most essential) elements on the subject, and also give you multiple explanations of the same concept (though not always).

Takeaway message?

You could potentially try to pick up multiple books on the same subject and try to learn this way. Follow the one that speaks to your background most closely, but the others are also likely to help.

You could potentially try to pick up multiple books on the same subject and try to learn this way. Follow the one that speaks to your background most closely, but the others are also likely to help.

FWIW, one of my favorite maths Youtubers, the "Math Sorcerer"[1], highly encourages this approach. Don't fixate on one particular book, but buy many books on a given topic, and allow yourself to experience different presentations of the material. And as he often points out, if you're willing to accept used books, and older editions (which is often OK if you're not buying a book for a specific class), then you can quite often get copies really cheap from Alibris, bookfinder, etc.

[1]: https://www.youtube.com/c/TheMathSorcerer

I have a stack of statistics books that I will flip through if I need explanation or illustration of a new concept. Usually one will help me out better than the others, but the combination of the different explanations usually improves my understanding.

I suppose this is really a parallelization of the "third textbook" model (i.e., if the third textbook you try when learning something new seems "much better" than the first two, it might because you actually did learn some things from the first two).

(comment deleted)