On 23/05/23 21:30, Jerry Friedman wrote:
> On Monday, May 22, 2023 at 8:20:57 PM UTC-6, Peter Moylan wrote:
>> On 23/05/23 03:15, Stefan Ram wrote:
>>> Paul Epstein <
peps...@gmail.com> writes:
>>>> the word "vertex" was used to describe the extremum of a
>>>> parabola
>>>
>>> |vertex, n. ... |1. a. ... |the point in a curve or surface at
>>> which the axis meets it ... |1715 J. T. Desaguliers tr. N. Gauger
>>> Fires Improv'd 13 |Two half Parabolas's whose Vertex's are C c.
>>
>> That's very confusing. A parabola meets the vertical axis at one
>> point, and the horizontal axis at two points. (Or sometimes none,
>> and rarely one.) Usually none of those three points is at the
>> extremum of the parabola.
>>
>> Ah. Google now tells me that the axis in question is the line of
>> symmetry of the parabola. Fair enough, but the word "axis" tout
>> court is highly ambiguous.
>>
>> Until this thread I had never heard of the vertex of a parabola,
>> and parabolae were certainly covered in my Form 3 mathematics. Is
>> this a recently introduced term?
> ...
>
> Stefan Ram gave the 1715 quotation from the OED.
I noticed that, but somehow failed to notice that that rules out
"recently introduced". Blame an ageing brain.
> "Vertex" is the only term I know for that point. What did you call
> it? (Also, "axis" is the term I know for the line of symmetry.)
I don't think we had names for these things, other than minimum and
maximum. I can see that "line of symmetry" would be a useful term for
the case where that line is tilted with respect to one's coordinate
axes, but we never covered that case in high school, and beyond high
school the topic hardly ever came up, except for parabolic antennae.
The equation describing a parabola did come up, of course, for example
in least-squares fitting of data to a quadratic [1], but in that sort of
application one uses ad hoc terminology.
[1] That very topic arose in a job I did a few years ago. Given data for
sound intensity versus angle, the customer wanted a way to express this
as a sum of two or three sound sources at unknown angles. I decided to
tackle this by fitting the data to a sum of five Gaussian functions. How
this is related to quadratic least squares is left as an exercise for
the reader.
A few years earlier, I was asked to produce a real-time calculation
method for gas concentration versus detector output, for a carbon
monoxide detector, based on calibration data at a few fixed points. In
that case, though, I ended up deciding that least squares approximation
by a cubic was a better solution, with the actual cubic changing over
the range. My boss originally asked for a spline solution, but I was
able to convince him that for that application area splines were not the
best approach.