Should I Become a Mathematician?

[dkotschessaa]

VCrakeV discusses and asks:You would not enjoy any major or minor in Mathematics. Look at the program requirements for a minor concentration at your school and decide if you believe you would or would not want/be interested in earning minor concentration or a degree in Math.

I'm a little perplexed by this response. The poster listed a lot of mathematical subjects he found enjoyable and a few he did not. You certainly don't have to like every subject to enjoy studying math. You study the stuff you like and the stuff you don't like but which you have to study anyway to be well rounded. One may also find tastes change over time, either because of something personal or finding a class or professor that makes the subject click for you.

-Dave K

 

[symbolipoint]

I'm a little perplexed by this response. The poster listed a lot of mathematical subjects he found enjoyable and a few he did not. You certainly don't have to like every subject to enjoy studying math. You study the stuff you like and the stuff you don't like but which you have to study anyway to be well rounded. One may also find tastes change over time, either because of something personal or finding a class or professor that makes the subject click for you.

-Dave K

Maybe the portion of his quote that I took would help see how my response would fit:

VCrakeV discusses and asks:

I want to know if I would enjoy a minor or major. I enjoy Algebra, derivative Calculus, integral Calculus, complex numbers, and concepts of infinity. But matrix Algebra and matrices in general bore me to tears. I also dislike 3D graphing, and 3D visuals in general. A major would also require some computer science, which I find quite a bore.

Checking what he said he dislikes helped me decide what to tell him. Not a bad opinion, just MY opinion, and other opinions from other members are still possible.
VCrakeV must decide what he wants to do and what efforts he will make to do it.

 

[dkotschessaa]

Maybe the portion of his quote that I took would help see how my response would fit:

VCrakeV discusses and asks:

Checking what he said he dislikes helped me decide what to tell him. Not a bad opinion, just MY opinion, and other opinions from other members are still possible.
VCrakeV must decide what he wants to do and what efforts he will make to do it.

Point taken. It's just that your statement "you would not enjoy..." seemed kind of definitive to me.

I think it is super important to know that when you choose to accomplish something or learn something there is a big chance you will not enjoy a lot of the process, and that doesn't always mean its not for you. I actually am not big fan of calculus/analysis etc. and at one point as an undergrad I felt that maybe math wasn't for me. I talked to a grad student at the time who told me "nah, there's other kinds of math." Turns out I am just better and more at home with more "discrete" math.

-Dave K

 

[Logical Dog]

one can develop his 3d spatial skills by:
1. playing with functions in microsoft mathematics (a free software)2. creating things in wings 3d (another free software):

I understand the importance, but it doesn't matter if I won't enjoy it. I can always just study whatever math interests me on my own time, in case university programs have too much math I don't like. But might I like such a program?

 

[symbolipoint]

Point taken. It's just that your statement "you would not enjoy..." seemed kind of definitive to me.

I think it is super important to know that when you choose to accomplish something or learn something there is a big chance you will not enjoy a lot of the process, and that doesn't always mean its not for you. I actually am not big fan of calculus/analysis etc. and at one point as an undergrad I felt that maybe math wasn't for me. I talked to a grad student at the time who told me "nah, there's other kinds of math." Turns out I am just better and more at home with more "discrete" math.

-Dave K

Hard to know exactly what to tell the VCrakeV member. If he wants a degree in Math or Phys, then some things he must study no matter how is likes or dislikes. Maybe he will prefer something in Engineering that will use much Mathematics and not need an undergraduate degree-or-more in Mathematics in order to achieve something in Engineering or whatever related.

 

[dkotschessaa]

I want to be a mathematician but I am probably too stupid, my favourite things so far have been mostly in discrete mathematics...set theory and logic, loved proofs and mathematical induction when I came across them . I also like vectors and am trying to study analysis on the side, calculus and algebra are my weakest backgrounds, I have very basic knowledge in them and Geometry too, but most of all I have an obsession with numbers and how almost everything else, evaluates to one or can be constructed through them, somehow this always amazes me (dont know why). I only started loving math a few months ago, i ask some very silly questions about it sometimes.. I know it is not a phase.

The question for me is not should, but if I can. @mathwonk great and vast thread..I will be coming here more often.

We've crossed on a couple of threads, and of course I've clarified that I'm not a mathematician, but I am a fairly slow person who is getting a master's later in life than the average.

I'm not going to tell you that you aren't stupid, because I am fairly stupid and so I'd rather just we talk stupid to stupid.

The thing to decide is not if you are smart enough, but if you are willing to work as hard as you need to and sacrifice as much as you need to in order to overcome any deficiencies you might have (real or perceived). If you aren't one of these "I got advanced placement in 10th grade and won math competitions every year" kids then whatever. I stopped worrying about "smart" during undergrad. There was a point where I believed I'd never even get my bachelor's. I thought it would be the most amazing thing ever if I did. I'd never done anything like it. It would be so amazing and so hard and oh my god.

When I asked myself whether I could get the degree my brain kept insisting that this was probably impossible. When I asked myself if I'd ever be "good at math," the answer came up, "no." Am I really super smart? "no."

So the different question I asked was "Can I study for the tests, do the homework, and pass the courses?"

hmm.

Turns out I was able to do that. One class, one homework set, one test, one subject at a time. Turns out there was no class called "The class where you find out if you are really a mathematician and intelligent or just a fraud." There were just these individual classes that if you took them, they added up to a degree. Weird right?

-Dave K

 

[symbolipoint]

dkotschessaa,
Great posting, #3716.

What do you think of the kind of person who studies one course (of Mathematics) at a time, but find he does not pass and needs to repeat every single one of them, such as the typical three-course series of Calculus 1,2,3? The person studies very hard EACH time, does not pass, then REPEATS the course passing successfully. Very slow progress. For this person, the courses, each of them, is just too difficult to pass when going though the semester one time. Learning happens, but the learning did not happen during the first time-through for each of these courses. Worked HARD! Studied long time! Did ALL the homework! Asked for some instructor help! Nobody has the answer to what person is doing wrong; maybe this person just needs MORE TIME to understand and to learn. Could this type of person go back to college or university and study to earn a degree in Mathematics? Probably not, because as he goes through each new course, the same trouble would happen - not pass, need to repeat course, then pass it, and so on. The person will finally know some good Math concepts and skills but getting a undergraduate degree in MATH will just take too long. Is the person stupid? Probably not. Anyone who works through this tough stuff is at least, very hard-working, but something in his head is just not letting him succeed in Mathematics more efficiently. He will very likely find that actually USING some of his learned mathematical concepts and skills in applied fields or situations works well for him; and then other people will say to him, "Wow, you are such a great mathematician", without realizing how much he struggled in his Math courses.

 

[IGU]

What do you think of the kind of person who studies one course (of Mathematics) at a time, but find he does not pass and needs to repeat every single one of them...

You didn't ask me, but I'll stick in my opinion anyway (hope you don't mind!).

I think people learn best in different ways, highly dependent on the individual. Such a person as you describe is clearly not learning well from lectures. Indeed lectures are not a particularly good way to learn anything for most people, so this person is not unusual. This person should spend some serious time trying to figure out how it is they learn best. Then they should use that method to learn whatever it is they want to learn. The classes should be taken only after they have learned the material (or at earliest simultaneously), merely to record a grade (and perhaps review). In this way classes will never need to be taken more than once.

Note that at top universities it is often the case that the best students rarely attend lectures. That's because they don't want to waste their time. Part of what makes them the best students is that they learned early what is the best way for them to learn (and it wasn't lectures). The key insight here is that learning methods are almost entirely orthogonal to the content to be learned. What is most important for any student is to figure out what their best learning methods are -- after that everything is much, much easier.

 

[symbolipoint]

IGU,
I just wonder, how many of the really successful Mathematics students actually do that/this?:

The classes should be taken only after they have learned the material (or at earliest simultaneously), merely to record a grade (and perhaps review). In this way classes will never need to be taken more than once.

That runs right along with, "having studied the course before".

 

[IGU]

I just wonder, how many of the really successful Mathematics students actually do that/this?

From my experience at Caltech (many years ago): many. From what I've read on the TSR forums from Cambridge mathematics students, many there too.

That runs right along with, "having studied the course before".

Sure. It's always best to study material at your leisure in your preferred way. Lectures are a terrible way to encounter material for the first time. It is pretty much inevitable that for difficult mathematical material a student will be looking at it multiple times. Gaining control over that process is one thing that good students do. And often (but not always) it turns out that the lectures are superfluous. Everybody has to find their own way. Certainly for the person under discussion, who needs to take classes twice, a better way must be possible.

 

[dkotschessaa]

dkotschessaa,
Great posting, #3716.

What do you think of the kind of person who studies one course (of Mathematics) at a time, but find he does not pass and needs to repeat every single one of them, such as the typical three-course series of Calculus 1,2,3? The person studies very hard EACH time, does not pass, then REPEATS the course passing successfully. Very slow progress. For this person, the courses, each of them, is just too difficult to pass when going though the semester one time. Learning happens, but the learning did not happen during the first time-through for each of these courses. Worked HARD! Studied long time! Did ALL the homework! Asked for some instructor help! Nobody has the answer to what person is doing wrong; maybe this person just needs MORE TIME to understand and to learn. Could this type of person go back to college or university and study to earn a degree in Mathematics? Probably not, because as he goes through each new course, the same trouble would happen - not pass, need to repeat course, then pass it, and so on. The person will finally know some good Math concepts and skills but getting a undergraduate degree in MATH will just take too long. Is the person stupid? Probably not. Anyone who works through this tough stuff is at least, very hard-working, but something in his head is just not letting him succeed in Mathematics more efficiently. He will very likely find that actually USING some of his learned mathematical concepts and skills in applied fields or situations works well for him; and then other people will say to him, "Wow, you are such a great mathematician", without realizing how much he struggled in his Math courses.

I know you are being hypothetical, but most of the time when I've seen a student fail a class multiple times the problem was in the approach, or they weren't giving it enough time in the *right* way, or they had other classes occupying their grey matter, or some attitude towards mathematics they picked up from somewhere, or some otherwise emotional blockage.

But OK, instead of going down that road I'll let the rest of your scenario play out, because I don't think that was the point.

The following was posted by Field's Medalist Timothy Gowers on Google+ about a year ago:

"What is it like to do maths?

About 99% of the time it's like this. "

My wife, who supported me through my degree, and who (rightfully) was keen on me not wasting my time, would often ask me how my homework/study was going.

I would often respond with something like, "Well, I didn't get any of the problems done, and I don't feel like I really understand that much yet. But I spent a lot of good time thinking about the concepts."

It took a bit of explaining to assure her that this was actually, in my opinion, productive time. Obviously I had no evidence of this fact unless ultimately I ended up producing something (like a completed homework, or a decent test grade).

So, I think that if you are at least giving a sincere effort to do mathematics, there will always be some benefit, even if you don't seem to have much to show for it. I did terribly in Graduate Algebra, but having gone through the class made other classes and subjects seem a lot easier.

-Dave K

 

[dkotschessaa]

Sure. It's always best to study material at your leisure in your preferred way. Lectures are a terrible way to encounter material for the first time. It is pretty much inevitable that for difficult mathematical material a student will be looking at it multiple times. Gaining control over that process is one thing that good students do. And often (but not always) it turns out that the lectures are superfluous. Everybody has to find their own way. Certainly for the person under discussion, who needs to take classes twice, a better way must be possible.

I attended good lectures where I enjoyed the professor's teaching style and found it elucidating. If the professor was not very good, I would often not attend unless I was stuck at a certain point, then I would go in so that I could hear him discuss it and ask questions about the parts I didn't understand.

-Dave K

 

[symbolipoint]

dkotschessaa,

What I described was the pattern of repeated fail/non-passage of ONE COURSE AT A TIME. The finer detail is

1. Take course
2. NOT pass course
3. Repeat course one time
4. Pass course successfully
5. Next Course of a sequence of courses --- Repeat at #1 with the incremented course

Meaning, each course passed but always needed to be repeated ONE TIME.

 

[dkotschessaa]

dkotschessaa,

What I described was the pattern of repeated fail/non-passage of ONE COURSE AT A TIME. The finer detail is

1. Take course
2. NOT pass course
3. Repeat course one time
4. Pass course successfully
5. Next Course of a sequence of courses --- Repeat at #1 with the incremented course

Meaning, each course passed but always needed to be repeated ONE TIME.

Oh.

In that case, I can't even say. It seems like a pretty unlikely scenario. Most people who fail courses fail one or two and drop out if they fail a whole bunch more. Sometimes they leave and come back much later. Strictly speaking the time it took me to get my bachelors was 18 years, if you count the first time I tried to go to college in 1996. I also repeated pre-calculus and two semesters of calculus, though this was due to time passed rather than failure.

-Dave K

 

[symbolipoint]

Oh.

In that case, I can't even say. It seems like a pretty unlikely scenario. Most people who fail courses fail one or two and drop out if they fail a whole bunch more. Sometimes they leave and come back much later. Strictly speaking the time it took me to get my bachelors was 18 years, if you count the first time I tried to go to college in 1996. I also repeated pre-calculus and two semesters of calculus, though this was due to time passed rather than failure.

-Dave K

Many people will give up when they fail a couple of Math (even Calculus) courses, and then pick or find something much less mathematics-intensive. Then there are some, maybe only a few, who will persist and keeping working at the needed mathematics courses until passing because these people stay dedicated to whatever math-intensive field they have chosen. Not sure which is the smarter way to go. Fail a couple of courses and change direction; or keep at it until passing each of the needed mathematics courses.

Something worth knowing is that if a student really works hard to learn a course the first time, does not pass it, and then repeats the course and again REALLY WORKS HARD the second time too, the course really does become easier to learn and understand.

 

[dkotschessaa]

Many people will give up when they fail a couple of Math (even Calculus) courses, and then pick or find something much less mathematics-intensive. Then there are some, maybe only a few, who will persist and keeping working at the needed mathematics courses until passing because these people stay dedicated to whatever math-intensive field they have chosen. Not sure which is the smarter way to go. Fail a couple of courses and change direction; or keep at it until passing each of the needed mathematics courses.

Something worth knowing is that if a student really works hard to learn a course the first time, does not pass it, and then repeats the course and again REALLY WORKS HARD the second time too, the course really does become easier to learn and understand.

Might need a separate thread,but I'll make 2 quick points:
1) I think mathematics has something to offer anyone at any level
2) we may need to discuss what we mean by working really really hard

 

[mathwonk]

Recently I rewrote my linear algebra notes. Hoping to achieve a result that someone could actually learn from, I expanded the explanation from under 15 pages to over 125 pages., Also I felt as if I actually understood the topic at last. The summary of what I learned is in the new 2 page introduction. If anyone enjoys and/or benefits from these, I am happy. One fun thing I learned from writing these is a cell decomposition of the grassmannian is given by the row reduced echelon form! who knew? (probably many of you, but not me.) (Would this qualify as an insight article? If so feel free to post it as one.)

http://alpha.math.uga.edu/%7Eroy/laprimexp.pdf

 

[symbolipoint]

Might need a separate thread,but I'll make 2 quick points:
1) I think mathematics has something to offer anyone at any level
2) we may need to discuss what we mean by working really really hard

Point #2 seems like it is not too precise, but honestly, WE KNOW WHAT IT MEANS! We do not need overly academic psycho games here.

Truly, some people do not know how to study too well, but other people do. One can typically not expect to study algebra 1 or 2 for just an hour per day, three times per week, and earn at least a C. Not enough effort. Not enough focus. Not enough study session length for good development. Without my trying to describe the details, filling the effort upward from that description, the study effort comes increasingly closer to "STUDYING REALLY HARD".

 

[mathwonk]

In relation to the discussion of how hard one needs to work to understand something, I would observe that the linear algebra notes I just posted have been rewritten many times over several years, and I only lately feel I understood the topic. That is also in addition to having taught the subject at various levels, from undergrad to graduate algebra many times over several decades, and to have written a graduate level algebra book including a thorough treatment of linear algebra. I have also read treatments by Lang, E. Artin, M. Artin, Hoffman and Kunze, etc etc... and watched videos by Strang.

Among the newer insights I have acquired is the fact that such topics as jordan normal form and diagonalization are usually emphasized, although as I recently appreciated, they are completely impractical for actual calculations in any even mildly general problem. This is due to the fact, usually ignored in calculus and other courses, that one generally cannot actually factor a randomly given polynomial into irreducible factors, say over the rationals, much less the reals. Hence all problems that we give students in calculus and linear algebra are carefully rigged to be easily doable, and there is no guarantee at all that the problems they encounter in real life practice, or even make up for themselves, will be even remotely doable by hand. As a professor, working from a book with cooked problem sets already included, I had the luxury of ignoring this inconvenient truth, and remained unaware of how useless the skills were that I was offering my students.

relatively few books explain to students the one actually feasible technique that they can always use in actual calculations, namely diagonalization of the characteristic matrix, by row and column operations within the ring of polynomials. This always yields the determinant and the invariant factors, hence also the rational canonical form. Then in those rare cases where these factors can be split further into irreducibles, it also may be refined to the jordan form.

most books also ignore explaining the geometric meaning of the reduced row echelon form, including the nice fact that it allows one to put coordinates on the grassman manifold of subspaces of a given vector space, and even gives a nice cell decomposition that easily yields the homology of the grassmannian. uniqueness of this reduced row echelon form is also usually omitted although there are many rather elementary and easy, as well as enlightening arguments. It has taken me years to appreciate all of these things, so I would tell any young or new student that math repays lots and lots of careful and repeated consideration. It also helps to try to explain it to someone else, which is my main reason for writing so many math essays and books or booklets. Of course you always hope someone else will benefit but they seldom seem to attract many readers. In this last case though I can say that envisioning a particular audience helped me focus my explanation by constantly aiming it at what I thought would be clear to that audience. So it helps in writing to imagine who you are writing for.

 

[mathwonk]

To summarize what I have learned, the basic tool in finite dimensional linear algebra and matrix theory is row and column operations. This is a computational way of changing your space by isomorphisms until the structure of your operator or matrix has been simplified enough to be visibly understandable.

In particular, two matrices are row equivalent, or left equivalent, if and only if the linear operators they represent become equal after changing the target space by an isomorphism, if and only if they have the same kernel, if and only if the matrices have the same row space, iff one can be transformed into the other by row operations, or iff they become equal after left multiplication by an invertible matrix, and finally iff they have the same row reduced echelon form.

The reduced row echelon form of a matrix is a matrix whose rows consist of a particularly nice basis for the common row space of matrices in the equivalence class; this is the unique basis that projects to the standard basis of the coordinate subspace spanned by the “pivot” coordinates. This provides a canonical representative for the left equivalence class. Finding solutions to the equation AX=0, i.e. finding a basis of the kernel of A, is easily done by row reducing A, since the reduced form has the same kernel, and one which is more easily found.

Row reduced r by n echelon matrices of rank r allow one to decompose the Grassmannian manifold of all r dimensional subspaces of k^n into “n choose r” cells, where each cell corresponds to the location of the r pivot columns. The largest cell is the one with the first r columns as pivots, hence the manifold has dimension r.(n-r), the number of free entries in such a row reduced matrix. They also allow one to put local coordinate charts on this manifold if we relax the definition of reduced to allow each of the n-r non pivot columns to have r entries, even if they are not the last n-r columns. Unlike the cells, these charts of course overlap.

Two matrices are right equivalent iff they have the same column space, iff the linear operators they represent have the same image, iff they can be transformed into one another by column operations, iff they become equal after right multiplication by an invertible matrix, i.e. they become equal as linear operators after an isomorphism of the source space. A canonical representative of this class is obtained by row reducing the transpose and then transposing it back. This has as columns a natural basis of the column space, analogous to the case above.

Two matrices A,B are (2 - sided) equivalent iff they can be transformed into one another by a combination of row and column operations, iff they become equal after (possibly) different isomorphisms of both source and target space, iff B = QAP where Q,P are invertible, iff A,B have the same rank r. A canonical representative for this class is the diagonal matrix whose first r diagonal entries are ones and the rest zeroes.

Two square matrices A,B are similar iff they become the same after performing conjugation by some invertible matrix, i.e. iff B = CAC^(-1) for some invertible C, i.e. they become equal as operators after performing a single isomorphism of the common source and target space. This equivalence can be determined by row and column operations performed on the associated “characteristic matrix”. If A is a square matrix, its associated characteristic matrix is the matrix [Id.X-A] with polynomial entries. This matrix can be diagonalized by row and column operations in the ring of polynomials, using the Euclidean algorithm. This can be done in a unique way so that the diagonal entries successively divide one another. Two square matrices A,B of the same size, are similar iff their characteristic matrices have the same diagonalized form. The non constant entries on the diagonal, which characterize the similarity class, are called the invariant factors of the (similarity class of the) matrix. Thus two n by n matrices are similar iff they have the same invariant factors.

If the invariant factors of A are f1,...,fn, then the linear operator represented by the original matrix is similar to the operation of multiplication by X on the product space k[X]/(f1) x ... x k[X]/(fn). The matrix of that multiplication operator, in the standard bases {1, X, X^2,...} for these factor spaces, is called the rational canonical form of the original matrix A.

Since multiplication by X satisfies the minimal polynomial f on the factor space k[X]/(f), it follows that the largest of the invariant factors of A is the minimal polynomial of the matrix A. In case one can factor this polynomial into irreducible factors over the field k, one can decompose the product decomposition further into a product of space of form k[X]/(g) where each polynomial g is a power of an irreducible factor of the minimal polynomial. This decomposition then gives rise to the jordan canonical form, after a slight tweak of the usual choice of basis. Since multiplication by X carries each basis vector in the standard basis {1,X,X^2,...} into the next one, except for the last, a decomposition into a product of spaces like k[X]/(f) is called a “cyclic” decomposition. The rational canonical decomposition is the cyclic decomposition with the fewest number of factors, while the Jordan decomposition is the one with the largest number of factors.

The nicest jordan form occurs when the irreducible factors of the minimal polynomial are all linear, and all occur to the first power in the minimal polynomial. Then the jordan form is diagonal. Even though one may not be able to compute this diagonal form, when working over the real number field this case always occurs when the original matrix A equals its transpose. Moreover in this nice case, the basis vectors making the matrix diagonal can even be chosen as mutually orthogonal, which is nice for doing geometry.

One can deduce from all this that the characteristic polynomial of A, which equals det[Id.X-A], is the product of the invariant factors of A, and its constant term is the determinant of A, and that this term is non zero if and only if A is invertible. One can actually compute the inverse of A by row reducing the matrix [A , Id].

that’s all folks. I guess the main difference between my old and my new point of view is that I like to focus now more on actually computable techniques, rather than the ideally simplest types of matrices (diagonal) which are impractical to compute,

 

[jedishrfu]

Why not put this together as an insight article on linear algebra?

Also there's an interesting set of videos on youtube by 3blue1brown called the Essence of Linear Algebra which are quite good.

 

[mathwonk]

can they just download my notes (linked in post 3727) from my webpage as an insight article? or do i need to write a new one (maybe ≈ 125 pages is too long?). I am not quite up to doing that right this second, having finished this project to my own satisfaction, over many years. (I first posted or linked to the 15 page version here over 12 years ago.)

 

[jedishrfu]

Insight articles are a page or two. Check the PF insights tab to see examples.

 

[PetSounds]

I'm a prospective math major. While I see myself likely going to grad school, I'd like to have the option of earning a living straight out of undergrad—you never know how circumstances and interests will change in four years.

So, how can I manouever myself into a position where I can either continue into higher education or land a well-paying job? I plan on taking a few courses in computer science, economics, and physics. Is it wise to cultivate knowledge in a field where I can apply math skills?

 

[symbolipoint]

Yes computer science and programming, good, important things to include. Engineering courses can add to your value too.

 

[mathwonk]

here is a link to my latest version of a condensed, advanced, undergraduate/graduate linear algebra book. It is aimed at someone who has already had at least one course in the topic and wants a beeline treatment of the main classification results on linear transformations up to various equivalences such as similarity. In particular both rational canonical form and Jordan form are discussed to some extent, plus spectral theorems. If anyone has the stamina to read some of it, I would benefit from some feedback.

I am aware it takes time to read something like this, since I myself have been reading Mumford's redbook of algebraic geometry since June and am only up to page 58. A reader of my notes might be someone who possibly already understands linear algebra, and is interested in seeing whether this summary agrees with his/her understanding. Thanks in advance.

http://alpha.math.uga.edu/%7Eroy/laprimexp.pdf

 

[mathwonk]

By the way, even for those who do not wish to spend a lot of time reading those notes, there is a 3 page introduction summarizing the whole document which I recommend to anyone who wants to know briefly what the subject is about. Indeed even reading the first page of this introduction will tell you exactly what a general linear transformation on a finite dimensional space looks like, up to isomorphism ("similarity").

The reason I recommend the introduction is that I wrote it only after completely finishing the notes, i.e. after I myself actually felt I understood the topic as well as I ever would.

Basically the answer is that every linear operator breaks up into a sum of "cyclic" operators. These cyclic operators can have different degrees. The simplest cyclic operator, of degree one, just takes a basis vector u to a scalar multiple of itself, like au. A cyclic operator of degree 2, operates on a sequence of 2 basis vectors, u,v, and takes u to v, and takes v to some linear combination of u and v, say au+bv. A cyclic operator of degree 3 operates on a sequence of 3 basis vectors u,v,w, and takes u to v, takes v to w, and takes w to some linear combination of u,v,w, say au+bv+cw. And so on. Thus to understand a cyclic operator you just need to know its degree and the coefficients of the last linear combination, which can be conveniently expressed as the coefficients of a polynomial, the "minimal polynomial" of the cyclic operator.

An operator which is a sum of operators all of degree one, is called diagonalizable, and most books emphasize this case. Unfortunately they often do not tell you that in practice it is usually impossible to decide whether this happens, and even when it does it is usually impossible to find those degree one operators. It is always possible however to decompose any operator into cyclic ones of higher degree, by finding their minimal polynomials, and hence this is emphasized first in my notes, before discussing the diagonalizable special case. The point is that although it is usually impossible in practice to diagonalize a given scalar matrix, it is always possible to diagonalize the associated "characteristic matrix" of polynomials, and this let's you find the minimal polynomials of a sequence of cyclic operators that decompose your original matrix.

I see I am repeating myself somewhat, but at least this version is more elementary than the ones above, and hopefully thus more clear.

 

[qspeechc]

I'm about to bore young students with some simple advice, but sometimes we need to repeat the simple stuff.

If you want to excel at anything you have to go above and beyond what is expected of you. Doing things like, working through problems that weren't assigned in class, and going over old exam papers, things like that I assume are standard practice. You need to be doing more than that.

When I was a young student, you're so impressed or even overwhelmed by your surroundings, your fellow studends, the professors, etc., that you take the education you're given as the one true gospel. Well there are many short-comings with your education system, that people have been thinking about since the beginning of time. For example, the topics you cover and the forms and types of the courses you take will be standard, and have been for several decades now (in Maths anyway), when they may to be the best anymore; also, they try to cram a lot into a 4 year degree, but there's still a lot they have to leave out; your lecturers are probably trying to balance a research career with teaching, etc.

The one time I came top of a class was in Real Analysis. Some of the students I beat in that course went on to get PhDs at top international universities. The only reason I came first that was because in the vacation before the course, I got a book on analysis out of the library and went through the first few chapters of it myself, because I felt this was a subject I was weak in, but I was intrigued by it nonetheless.

I think it goes without saying, you have to be using your vacations to do some additional studying; whether it's revising old coursework, trying to solve problems you couldn't do before, preparing for new courses, or studying some other topic on your own.

But besides that, you should be studying outside of what assigned to you by your lecturers. If your course only covers 7 chapters in a 10 chapter book, make a plan to study the others at some other time. Go to the library, find other books on the subject, and if they catch your fancy, work through that as well. Don't take what's given to you in your coursework as the be-all and end-all of your education. Only you can really educate yourself, and you have to take control of your education. Go to the library and try to read old papers on the subject you're studying. If they're too advanced, try to find older papers and try again. Study topics outside those you will do in your degree. For example, often Number Theory isn't taught in universities, but you may want to study that on your own. Etc.

There's a lot of advice on PF and elsewhere on the web on how to take control of your own mathematics education. Don't be afraid to go further than what's set out for you in courses.

 

[Tukhara]

This will be glossed over, but I feel an interesting piece of advice is: getting things wrong is more important than getting things right in mathematics.

 

[symbolipoint]

This will be glossed over, but I feel an interesting piece of advice is: getting things wrong is more important than getting things right in mathematics.

Notable comment. It reminds me of some saying from some obscure source which remarked, ",...wisdom is the result of bad experience."

 

[mathwonk]

Another piece of basic math advice: I have been struggling at length to make a proof today and have just realized I have not yet used the hypothesis!

the advice of qspeechc reminds me of a course I took in which the lecturer omitted a few basic proofs which I learned on my own. On the final I gave those proofs and it seemed to make an impression. This is quite useful when requesting a letter of recommendation.

 

[qspeechc]

How important is Graph Theory and Combinatorics to a mathematician? I'm sure it depends on the field, but let's say the main fields, algebra, analysis, geometry, number theory, etc.?

 

[mathwonk]

I have myself never used either in my career as algebraic geometer. Of course I don't know anything about them either which makes it hard to use them. On the other hand one of my more knowledgeable friends, did use graph theory in his proof that surfaces in P^3 can have a certain number of singular points.

Beauville, A. ``Sur le nombre maximum de points doubles d'une surface dans P^3.'' Journées de géométrie algébrique d'Angers (1979). Sijthoff & Noordhoff, pp. 207-215, 1980

I was amazed and impressed at the time, but not enough to study graph theory. I tend to study things not so much because they may be useful but because they appeal to me.

edit: I seem to be confusing graph theory with coding theory, but they may be related, or at least one may be used in the other. I seem to recall my friend utilized the concept of a "Hamming code" in his proof, but i no longer have the reference in my library.

 

[jedishrfu]

Here’s something on Hamming codes

https://en.m.wikipedia.org/wiki/Hamming_code

I didn’t see a reference to graph theory in the article but I remember a hamming code representation as the corners of a cube which would be a 3D graph.

https://en.m.wikipedia.org/wiki/Hamming_distance

 

[mathwonk]

i had to google "hamming codes/graph theory" to get a connection.