Things that bother me about the log function.
I was just going to
focus on "What base do people mean?" until I realized that there's a lot
of ambiguous things about this function.
The first is..."What base do people mean?" Generally, you have to
take it by context of the person you're talking to. If I'm talking to a
computer scientist, it's base 2. If I'm talking to an engineer, it's
base 10. If I'm talking to a mathematician, it's base e. We all just
say 'log' though.
Next up is, "What's the branch cut?" Generally, people mean the
$-\pi$ branch cut, where log of negative numbers is undefined. However,
over the complex plane, we could mean different possible branch cuts.
Moreover, depending on the surface you're considering the log function
over, different things happen, which brings me to...
"What's the domain?" If it's over $\mathbb{C}$, it makes sense to
even ask the previous question. If it's over $\Re$, then _really_
you're considering the domain $\Re^+$, the positive real axis.
Although, you _could_ define a log function over the negative real axis
and leave the positive real axis undefined. Moreover, there's a
particular kind of Riemann surface (in fact, it's _constructed_ so that
the following happens) where the log function over _it_ is defined
_everywhere_. Moreover, the question of which domain you're considering
is important to answer the question...
"What's the derivative?" I remember hearing this story of a Physics
professor docking points off of a student for drawing the graph of the
log function's derivative as being $1/x$, but only on the _positive_
part of the axis. Technically, the student was 100% correct. The
derivative of $log(x)$ is $1/x$ only on the positive real axis because
$log(x)$ itself is only _defined_ on the positive real axis. Because of
this, depending on where the branch cut, domain, and even _base_ you're
considering, the derivative will be different!
However, we usually don't clarify these things. All of this
information about the log function is usually easily taken up according
to the context of the discussion.
No comments:
Post a Comment