Why do u,v components in Cauchy-Riemann conditions are irrotational?

224 Views Asked by At

It's very strange to me! When we decompose a complex function to a real part and an immaginary part, we have

$f(z) = u(x,y) + j v(x,y)$

following the conditions of analyticity we can derive the Cauchy-Riemann conditions where

  1. $u_x$ = $v_y$
  2. $v_x$ = -$u_y$

I see that my text book says that these components are irrotational while I see that it depends. When I calculate the curl I see:

Curl = 0 + 0 + ($v_x$ - $u_y$) = 2 * $v_x$

I see that if we consider u the second and v the first component it will be irrotational but in the default order the thing that I'd expected to work the curl doesn't appear to be 0. What is my error?

1

There are 1 best solutions below

5
On BEST ANSWER

The vector field $\vec f(x,y)$ that corresponds to the complex function $f(z)$ has a sign change on the $y$-component. That is,

$$\vec f(x,y) = u(x,y) \hat x - v(x,y) \hat y$$

The curl and divergence of this vector field are zero as a consequence of the Cauchy-Riemann conditions. In vector fields that are divergenceless and curlless* are fully determined by their values on some bounding surface (or a hypersurface beyond 3d), just as 2d holomorphic functions are fully determined by their values on some closed curve.

*Curl doesn't exist beyond 3d, but there are ways to generalize it, and it's not a crime to call these generalizations "curl" also.


Edit: now why should there be a sign change? Well, you could identify it by inspection, but that's a pretty unsatisfying answer. A better (though more complicated) one is to construct a correspondence between complex functions and vector fields. You can do that if they're both algebraic elements of the same algebra--a "clifford" algebra.

Clifford algebra works with a noncommutative product called the geometric product. A 2d clifford algebra can be built from four basis elements: call them $1, e_1, e_2, e_1 e_2$. Basically, the multiplication works like so: $e_1 e_1 = e_2 e_2 = 1$ (so it captures the dot product), and $e_1 e_2 = -e_2 e_1$ (so for orthogonal vectors, it anticommutes like the cross product), and $e_1 e_1 e_2 = (e_1 e_1) e_2 = e_2$, so it's associative.

Now, how does this all connect to complex numbers? As it turns out, the object $e_1 e_2$ has a special property:

$$(e_1 e_2)^2 = e_1 e_2 e_1 e_2 = -e_1 e_1 e_2 e_2 = -1$$

So algebraically, $e_1 e_2$ behaves exactly like the imaginary unit, $j$! The algebra of linear combinations $a + b e_1 e_2$ for real numbers $a, b$ is exactly the algebra of complex numbers.

But! We have the extra benefit of still having $e_1, e_2$ around as basis vectors. Vectors coexist alongside these complex numbers. Indeed, you can take a position vector $x e_1 + y e_2$ and convert it to a complex number by multiplying on the left by $e_1$:

$$e_1 (x e_1 + y e_2) = x + y e_1 e_2$$

So, this demands the question why we can't do this for vector fields. The answer comes from differentiation, from the use of $\nabla$. Consider a function $f(x,y) = u(x,y) + e_1 e_2 v(x,y)$, like your complex function, and let's take a derivative with $\nabla$:

$$\nabla f = e_1 (\partial_x u - \partial_y v) + e_2 (\partial_y u + \partial_x v)$$

The components are those expressions in the Cauchy-Riemann condition, so if $\nabla f = 0$, then the C-R condition holds.

How can we build up a vector field from $f$, though? We can't do it by multiplying by $e_1$ on the left: $\nabla$ is in the way, and inserting an $e_1$ between $\nabla$ and $f$ isn't something you can just do whenever you like.

But you can multiply by $1 = e_1 e_1$ on the right. This gives us the expression $[\nabla (f e_1)] e_1$. That $f e_1$ evaluates to

$$f e_1 = u e_1 + e_1 e_2 e_1 v = u e_1 - v e_2$$

I have, perhaps, gone a bit too far in trying to give evidence for this mathematical curiosity, but I hope it's piqued your interest.