Is Vector Calculus Needed for Statistics?

[laz0r]

I've taken a multi-variable calculus course already that covers infinite sequences and series, Taylor's theorem, quadratics surfaces, double and triple integration etc.

I'm looking to get a Master's Degree in statistics two years from now, is there any point of me taking a class that involves multi-variable vector calculus (Green's Theorem, gradient, divergence/curl etc...)? From the material that I've seen online it seems that it is not used in statistics at all, but I'm looking for more opinions here.

I'd like to point out that I did well in the first multi-variable calculus course, but the "pure math" portion of the course did not appeal to me. I much prefer applied math courses (hence the statistics degree).

Thanks in advance!

 

[TheKracken]

What is your current major?

 

[homeomorphic]

Can't comment too much on statistics, but I'm pretty sure SOME statistician somewhere would use it. Green's theorem, div, grad, curl stuff is hardly "pure" math. It's very applied-sounding physics/engineering type stuff, although there's a role for it in some pure math, too.

I have a sneaky suspicion you might not be too happy with graduate-level material. My understanding is that they want you to take stuff like real analysis.

 

[laz0r]

Well this course isn't even required for my degree, I'm just wondering if it's worth taking it.

My current major is statistics.

 

[MarneMath]

Two comments:

After your first course in mathematical statistics, you'll face greater and greater abstraction. I would be rather surprised if your course in mathematical statistics was not proof based. Knowing how to do statistics is easy, what is hard is knowing when to use it and how to implement it. That requires a lot of abstraction for it to be generally useful.

Secondly, vector calculus isn't essential. I wouldn't say it's useless. I would simply say that if you knew how to to work with vector calculus, a good deal of deviations for estimates for parameters can be heavily simplified. Furthermore, there does exist a sub-group of people who work exclusively with probabilistic models where finding a minimum of maximum value is essential. Many of the techniques you need to solve these problems relate heavily to gradient descent or conjugate search methods. To understand those methods, it takes an understanding of vector calculus.

Nevertheless, I personally haven't found a need for it outside of directly solving for a proof. Furthermore, linear algebra will infinitely be more important to you and programming.

 

[laz0r]

Yeah, I'm just finishing a mathematical statistics course right now (the first of two) and it is somewhat centered around proofs. That being said, I found it to be an alright class to take.

I'll probably go ahead and take more linear algebra instead of vector calculus, as I have room for it.

 

[AlephZero]

Green's theorem, div, grad, curl stuff is hardly "pure" math.

Hmm... Green's theorem is just a special case of Stokes's theorem (which was actually discovered by Kelvin, not Stokes).

The modern statement of Stokes's theorem as general result about integrating differential forms on manifolds sounds like "pure" math to me, even if applied mathematicians had found some practical uses for special cases before the pure guys hijacked it.

 

[IGU]

I'm looking to get a Master's Degree in statistics two years from now...

What you should need for statistics is some basic analysis (point set topology and metric spaces) leading to a course in measure theory. When you understand all that you're ready to understand probability theory, the foundation of statistics. See, for example, Stanford Math 230ABC.

 

[StatGuy2000]

To the OP:

Obviously this would depend on the specific requirements of the Masters program in statistics you are referring to, but if your goal is to finish with a terminal Masters degree, you would probably not require further courses in vector calculus or advanced real analysis (although it wouldn't hurt to take them either). I agree with other posters that a basic understanding of analysis (at least to understand the definition of measure, Hilbert spaces, etc.) will help you tremendously in understanding probability theory.

If your intent is to pursue further graduate studies in statistics, then having more advanced math courses will help in understanding, say, mathematical or theoretical statistics.

 

[laz0r]

Thanks for all the replies, I'm actually not planning on pursuing anything past a Master's degree in statistics, I'd like to get into the industry as soon as possible.

I'll definitely look into expanding my knowledge of linear algebra, as I know it's used quite a lot in statistics. I'll also look into a real analysis courses or simply learn it on my own if it will assist me in the future.

 

[homeomorphic]

Hmm... Green's theorem is just a special case of Stokes's theorem (which was actually discovered by Kelvin, not Stokes).

The modern statement of Stokes's theorem as general result about integrating differential forms on manifolds sounds like "pure" math to me, even if applied mathematicians had found some practical uses for special cases before the pure guys hijacked it.

The course he was describing didn't sound like differential forms.

 

[member2357]

At my university only calculus 2, linear algebra, real analysis, probability and statistics and prerequisite for the statistics major.