Gradient

Required math: vectors, calculus

Required physics: none

Given a scalar field defined over three-dimensional space (for example, the temperature in a room or the density of some substance over its volume), we would like to be able to determine the rate of change of the field at a given point as we move along a given direction. For example, if we know the temperature in a room we might like to know how fast the temperature changes as we move straight up towards the ceiling.

To make things precise, we’ll consider a unit vector that points in the direction in which we wish to find the rate of change of the scalar field. In rectangular coordinates we can write this unit vector as

\displaystyle   \mbox{\ensuremath{\hat{\mathbf{s}}}} \displaystyle  = \displaystyle  u_{x}\hat{\mathbf{i}}+u_{y}\hat{\mathbf{j}}+u_{z}\hat{\mathbf{k}}
\displaystyle  s_{x}^{2}+s_{y}^{2}+s_{z}^{2} \displaystyle  = \displaystyle  1

From linear algebra, we know that a line {\mathbf{l}=(x,y,z)} parallel to this vector can be written in vector form as

\displaystyle  \mathbf{l}=\mathbf{r}_{0}+t\hat{\mathbf{s}}\ \ \ \ \ (1)
where {t} is a parameter that ranges from {-\infty} to {+\infty} and {\mathbf{r}_{0}=(x_{0,}y_{0},z_{0})} is a point on the line. In components, we have

\displaystyle   x \displaystyle  = \displaystyle  x_{0}+ts_{x}
\displaystyle  y \displaystyle  = \displaystyle  y_{0}+ts_{y}
\displaystyle  z \displaystyle  = \displaystyle  z_{0}+ts_{z}

Now if we have some scalar field given by {a=a(x,y,z)} we can find its directional derivative in a given direction by finding the total derivative of {f} with respect to the parameter {t}. From the chain rule for functions of several variables, we have, since each of the components of {\mathbf{l}} depends on {t}:

\displaystyle  \frac{da}{dt}=\frac{\partial a}{\partial x}\frac{dx}{dt}+\frac{\partial a}{\partial y}\frac{dy}{dt}+\frac{\partial a}{\partial z}\frac{dz}{dt}

From the above equations, we can work out the deriviatives with respect to {t} and get

\displaystyle  \frac{da}{dt}=\frac{\partial a}{\partial x}s_{x}+\frac{\partial a}{\partial y}s_{y}+\frac{\partial a}{\partial z}s_{z} If we define a vector called the gradient of {f} by

\displaystyle  \nabla a\equiv\frac{\partial a}{\partial x}\hat{\mathbf{i}}+\frac{\partial a}{\partial y}\hat{\mathbf{j}}+\frac{\partial a}{\partial z}\hat{\mathbf{k}}
we can write the directional derivative as

\displaystyle  \frac{da}{dt}=\nabla a\cdot\hat{\mathbf{s}}

Since {\hat{\mathbf{s}}} is a unit vector, we can write this as

\displaystyle  \frac{da}{dt}=|\nabla a|\cos\theta
where {\theta} is the angle between the gradient and the unit vector {\hat{\mathbf{s}}}. From this we see that {da/dt} has its maximum value when {\theta=0}, or when the gradient is parallel to the direction of the derivative. In other words, the gradient of a scalar field points in the direction of greatest rate of change of the field at a given point.

In general curvilinear coordinates, we have three coordinates {u}, {v} and {w} whose unit vectors are mutually perpendicular. We can define the line {\mathbf{l}} parametrically in terms of these coordinates in the same way as with rectangular coordinates above, so that the rate of change of the scalar field {a} is

\displaystyle  \frac{da}{dt}=\frac{\partial a}{\partial u}\frac{du}{dt}+\frac{\partial a}{\partial v}\frac{dv}{dt}+\frac{\partial a}{\partial w}\frac{dw}{dt}\ \ \ \ \ (2)

In these general coordinates, however, a line element has the form

\displaystyle  d\mathbf{l}=f\; du\hat{\mathbf{u}}+g\; dv\hat{\mathbf{v}}+h\; dw\hat{\mathbf{w}}
where {f}, {g} and {h} are functions of the three coordinates. Dividing through by {dt} and taking the limit to get a derivative, we have

\displaystyle  \frac{d\mathbf{l}}{dt}=f\;\frac{du}{dt}\hat{\mathbf{u}}+g\;\frac{dv}{dt}\hat{\mathbf{v}}+h\;\frac{dw}{dt}\hat{\mathbf{w}}

From this, we can write 2 as a dot product if we define the gradient in curvilinear coordinates as

\displaystyle  \nabla a=\frac{1}{f}\frac{\partial a}{\partial u}\hat{\mathbf{u}}+\frac{1}{g}\frac{\partial a}{\partial v}\hat{\mathbf{v}}+\frac{1}{h}\frac{\partial a}{\partial w}\hat{\mathbf{w}}

Then we get

\displaystyle  \frac{da}{dt}=\nabla a\cdot\frac{d\mathbf{l}}{dt}

We can still write the equation for the line {\mathbf{l}} in the form 1, except now the components of the unit vector {\hat{\mathbf{s}}} would be written in terms of the three basis vectors for whatever coordinate system we are using. Thus the analysis above is still valid for general coordinate systems, and the gradient still represents the direction of maximum increase of the scalar field.

About these ads
Post a comment or leave a trackback: Trackback URL.

Trackbacks

  • By Curl and Stokes’s theorem « Physics tutorials on Thursday, 6 October 2011 at 16:05

    [...] are several theorems involving the curl, but probably the most important one is that the curl of a gradient of a scalar field is always zero. That [...]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Follow

Get every new post delivered to your Inbox.

Join 367 other followers

%d bloggers like this: