27 comments

[ 4.5 ms ] story [ 68.5 ms ] thread
Maxwell's equations as most of us know it are the Heaviside-Gibbs reforumulation. Good article though.
And the one we should know is the Geometric Algebra formulation, where it ends up as Maxwell’s Equation, singular.

Gibbs/Heaviside vectors are cumbersome and confusing, and the faster we can dump them as a society the better.

Can you link to an explanation/derivation of the geometric formulation?
Yep. I'm real glad Hestenes has been championing VA/VC so hard. That first text is a perfect supplementary reading that any undergraduate should read. Who was the author?

I skipped the a few and went straight to the last PDF (the Survey) and coming from a stronger foundational math background, I felt right at home. (I'd imagine those who went down the physics route would be thrown off by the notation though, so I understand the author's preface). He mentions physics and CS applications, but I really think this has structural engineering applications as well -- e.g. R^4 seems like a perfect domain to explore bidirectional tensile and shear strength of materials (especially composite materials/aggregates), then perform some FEA with all of the simple operations within that come inherent to operating within the bounds of the 4-vector form. I mean that's a real trivial example (and I obviously don't know enough about structural engineering to make a good analogy) but it seems like one of those cases where you move the problem set lifted into another domain, perform your manipulation then finally lower back into your original (co-)domain. (Frequently done in algebraic topology, most famously by Perelman on Poincare)

I think there is a very good reason why structural engineers never see this type of thing, and it's because most structural engineers don't like math. They like clicking buttons in their CAD/FEA/BIM/<insert-latest-TLA> computer system.

The people who write those systems on the other hand enjoy this math very much, and tend to use Galerkin as a verb and will casually point out just how wonderful the Lax-Milgram theorem is.

The vector calculus formulation is not cumbersome and confusing, and we should not replace it. Many problems are simplest in vector calculus. Also, vector calculus formulation gives some insights that are not evident in other formulations. It's also extremely enlightening from a pedagogic perspective when you are learning physics. Of course if you're a mathematician and you don't want to learn physics, than vector calculus is just a distraction because it's idiosyncratic and doesn't connect with anything else.

The differential forms formulation, which in many other cases (but not all!) is more useful to work with than the vector calculus formulation should probably be replaced with geometric algebra though. They are basically the same thing, but differential forms are more idiosyncratic, and some symmetries are not apparent. Neither are useful as an introduction to the aspiring physicist or engineer though.

Also credit to this guy: https://en.wikipedia.org/wiki/Oliver_Heaviside

... Who took the original 20 equations and simplified them to the 4 we normally see.

That same year he patented, in England, the coaxial cable.

That's a fairly major footnote, on an interesting career. He seems to be one of those people who quietly had a significant impact on the future.

If you ever used Laplace transforms to solve differential equations, that is Heavisides work too.

Good book: http://www.amazon.com/Oliver-Heaviside-Electrical-Genius-Vic...

I've read that book, it is very good. A fantastic story/subject, though probably not for a general audience, just those interested in a detailed history of Heaviside and his work. It's nice to see a well-written 'technical' history book - so often such books are written by someone who doesn't really understand the subject - this is an exception.

You are mistaken about Laplace Transforms - Heaviside used a different method which was rather ad-hoc - sometimes it works, sometimes not. For the same problems, we now use Laplace Transforms instead, because they are better/cleaner.

Heaviside was somewhat written out of history because he was considered a crank during his lifetime. His reputation is basically rehabilitated these days, but it's only physicists/mathematicians/engineers who have heard of him.
His name was all over the physics textbooks, probably even more frequently than Maxwell, so he is not entirely forgotten.
Bill Bryson mentions Heaviside in his latest book The Road to Little Dribbling.
In reading up on this the other day, when that Guardian article appeared, I was intrigued to find that there is a conspiracy theory types who advance the idea that Heaviside was the bad guy because he removed important parts of Maxwell's work that explains, well...X-Files type stuff. Read on: http://www.cheniere.org/references/maxwell.htm
Man. What a pile of nonsense.
> This work is all documented in the hard French scientific literature.

Be nice to have some citations.

From the Wikipedia article:

"From 1882 to 1902, except for three years, he contributed regular articles to the trade paper The Electrician, which wished to improve its standing, for which he was paid £40 per year. This was hardly enough to live on, but his demands were very small and he was doing what he most wanted to. "

The prototypical hacker?

Where are all the gradient and curl operators in the picture of the original manuscript on the page? Or was that notation developed later?
Of course, like every article published by a newspaper or magazine, its only purpose is to carry links to other articles published by this newspaper, and omit any reference to the original source.

So here's the original publication: http://rstl.royalsocietypublishing.org/content/155.toc

Philosophical Transactions of the Royal Society of London, For the year ⅯⅮⅭⅭⅭⅬⅩⅤ, Vol. 155

http://rstl.royalsocietypublishing.org/content/155/local/fro... (Title page and Contents) http://rstl.royalsocietypublishing.org/content/155/459.full.... (Article)

Some HN readers might be interested to know that Maxwell's equations can be written in a single expression (with one auxiliary condition).

The first step is to define the Faraday tensor F, which can be written as a 4x4 antisymmetric matrix. Such a matrix has 16 elements, but the diagonal is all zeroes (by antisymmetry: x = -x => x = 0), and the 6 off-diagonal elements on one side of the diagonal are simply the negatives of their mirrored values. This leaves 16 - 4 - 6 = 6 independent elements—exactly enough to contain the 6 elements of the electric and magnetic field vectors (3 each for E and B), which in fact are exactly the elements of the Faraday tensor.

In terms of F, the Maxwell equations for the divergence of B and the curl of E can be combined in the single equation

    dF = 0
where d is the exterior derivative. As with ordinary potential theory, where a vector field with zero curl can be written as the gradient of a scalar function, a tensor whose exterior derivative is zero can be derived from another tensor, called in this case the 4-potential A (which contains both the ordinary electric potential and the magnetic vector potential):

    F = dA      (1)
(This is the auxiliary condition alluded to above.) The second expression for F, corresponding to the divergence of E and the curl of B, combines the exterior derivative and the Hodge star operator ⭑ [1]:

    d⭑F = J     (2)
where J is the 4-current density (which contains both the charge density and the 3-dimensional current density) [2]. Substituting (1) for F in (2) then yields Maxwell's equations in a single expression:

    d⭑dA = J    Maxwell's equation(s)
This equation is perhaps a more compact and elegant choice for those T-shirts: "God said: d⭑dA = J—and there was light!"

[1]: The ⭑ character may not display in some browsers. In this case, you can use an asterisk instead.

[2]: Eq. (2) is written in "God's units", equivalent to SI with μ₀ = ε₀ = 1 => c = 1.