(1) Select three points (uniformly) randomly in the interior of a unit square. What is the expected area of the triangle formed by those 3
points?
I think it would be useful if I could answer the simpler question: given
two points (uniformly) randomly placed on a unit line. What is the
expected distance between them? I'll have a think about that.
David Entwistle <qnivq.ragjvfgyr@ogvagrearg.pbz> wrote or quoted:
I think it would be useful if I could answer the simpler question: given
two points (uniformly) randomly placed on a unit line. What is the
expected distance between them? I'll have a think about that.
Possible spoiler for this:
Tvira bar cbvag cynprq havsbezyl enaqbzyl ba n havg yvar sebz mreb
gb bar, jung vf vgf rkcrpgrq k inyhr? Jryy, V guvax vg vf whfg
gur vagrteny bs k bire gur havg yvar, nf jr whfg unir gb nirentr
(fhz) gur cbffvoyr k inyhrf jvgu gur fnzr jrvtug (ol havsbezvgl)
rireljurer!
Fb sbe gjb cbvagf cynprq havsbezyl enaqbzyl, bar fubhyq hfr
gur qbhoyr vagrteny bs gurve qvfgnapr bire /gjb/ havg yvarf.
Gur pungobg gbyq zr guvf vagrteny vf bar guveq, juvpu znxr
fbzr frafr tvira gung gur frpbaq cbvag unf yrff fcnpr gb zbir
njnl sebz gur svefg cbvag pbzcnerq gb gur svefg pnfr jurer gur
svefg cbvag jnf ng gur bevtva naq gur erfhyg jnf bar unys.
N fznyy Clguba cebtenz nyfb cevagf inyhrf arne bar guveq.
On 27/07/2026 18:39, Stefan Ram wrote:
David Entwistle <qnivq.ragjvfgyr@ogvagrearg.pbz> wrote or quoted:
I think it would be useful if I could answer the simpler question: given >> two points (uniformly) randomly placed on a unit line. What is the
expected distance between them? I'll have a think about that.
Possible spoiler for this:
Tvira bar cbvag cynprq havsbezyl enaqbzyl ba n havg yvar sebz mreb
gb bar, jung vf vgf rkcrpgrq k inyhr? Jryy, V guvax vg vf whfg
gur vagrteny bs k bire gur havg yvar, nf jr whfg unir gb nirentr
(fhz) gur cbffvoyr k inyhrf jvgu gur fnzr jrvtug (ol havsbezvgl)
rireljurer!
Fb sbe gjb cbvagf cynprq havsbezyl enaqbzyl, bar fubhyq hfr
gur qbhoyr vagrteny bs gurve qvfgnapr bire /gjb/ havg yvarf.
Gur pungobg gbyq zr guvf vagrteny vf bar guveq, juvpu znxr
fbzr frafr tvira gung gur frpbaq cbvag unf yrff fcnpr gb zbir
njnl sebz gur svefg cbvag pbzcnerq gb gur svefg pnfr jurer gur
svefg cbvag jnf ng gur bevtva naq gur erfhyg jnf bar unys.
N fznyy Clguba cebtenz nyfb cevagf inyhrf arne bar guveq.
Stefan's thinking is all correct, and his chatbot has given him the correct value for the double
integral. But there is a "trick" to the problem that makes the answer easy! I'd even go so far as
to say /obvious/... I'm not rot13ing my post because I'm not going to directly specify the trick,
but only talk about why such a trick might work.
So first, why is Stefan's double-integral a bit of a pain to compute by hand? It's because the
distance between points x,y on the unit line is |x-y|, which is either (x-y) or (y-x) depending on
whether x<y or not. Either expression easy to integrate by hand (e.g. with respect to y if we're
doing y first), but to use these expressions we need to know whether x<y or not. So this means
breaking up the ranges we integrate over into smaller subranges where we definitely know either x<y
or x>=y. Not really that hard (I imagine - I'm sure I've done this at some point, but can't
remember the experience :)), but certainly fiddly with subcases etc.
But suppose instead of two points on a unit line we had 7? The "trick" works as well for 7 points,
giving the expected distance between adjacent points. It can do this by utilising a symmetry in the
problem where we can argue that the value we get for adjacent points x_1,x_2 applies equally for
x_2,x_3 etc., and we can add all the (equal) expectations together to get an obvious total. Well,
this doesn't quite work with David's problem exactly as given, because the end-points of the
interval [0,1] mess things up a bit, and the "trick" is seeing how to look at the problem slightly
differently (but equivalently) in a more symmetrical manner where the endpoints are no longer special!
That's enough of a hint (I'd like to think! but happy to spell out more details if asked) just
knowing that the more general case of n points is "easy" with the trick focusses the mind I think,
and surprisingly may be easier than just thinking about 2 points.
[Someone asked for such cases where thinking of an apparently harder version of a problem might
suggest solutions missed when focussing on a simpler case - maybe this would count as an example?
Hmm, it was when we were talking about arrangements of 4 points in a plane having just two different
distances between any pairs of the points, and one of those arrangements is hard to see, but easy to
see if we're looking for 5 points.]
Mike.
BUT that 2nd cup often leads to heart palpitations.
(Sometimes I try Decaf for its placebo effect.)
(1) Select three points (uniformly) randomly in the interior of a unit square. What is the expected area of the triangle formed by those 3 points?
(2) What if "unit square" is replaced with "a parallelogram of unit area"?
There are at least three ways to solve #1:
(a) a SEXtuple definite integral with Hero's formula inside.
This may be a popular 21st-century approach with powerful math
software at one's fingertips.
(b) Monte Carlo simulation.
(c) The nice step-by-step approach used by J.J. Sylvester in the 19th century.
Problem 1 had a triangle inside a quadrilateral. Let's reverse this
for Problem 3.
(3) Select four points (uniformly) randomly in the interior of an
equilateral triangle. What is the probability that the 4 points can be ordered such that they form a convex quadrilateral?
(And what if "equilateral" is struck from the problem statement?)
(4) ...
Cheers,--- Synchronet 3.22a-Linux NewsLink 1.2
James
Well,
this doesn't quite work with David's problem exactly as given, because
the end-points of the interval [0,1] mess things up a bit, and the
"trick" is seeing how to look at the problem slightly differently (but equivalently) in a more symmetrical manner where the endpoints are no
longer special!
I think it would be useful if I could answer the simpler question: given
two points (uniformly) randomly placed on a unit line.
David Entwistle <qnivq.ragjvfgyr@ogvagrearg.pbz> wrote or quoted:
I think it would be useful if I could answer the simpler question: given >two points (uniformly) randomly placed on a unit line.
I think "unit interval" is more common (for [0,1]).
BTW: In my previous post, I removed the ROT13 from a post of
James, and then quoted it thinking the ROT13 in this group is
used for spoilers, and what I quoted was no spoiler. However,
in hindsight, I think I should, as a general rule, never remove
ROT13 from someone else's post and then quote it.
I think it would be useful if I could answer the simpler question: givenWhat is the
two points (uniformly) randomly placed on a unit line (unit interval).
expected distance between them? I'll have a think about that.
Mo problem. I indulged in the 2nd cuppa today with no palpitations.
Yay!! (I had a headache; either coffee or Tylenol seems to cure that.) >However each cup was only a single dose of Nescafe 3-in-1.
Good coffee-shop coffee is -- what? -- 2-+ times the strength
of a Nescafe dose?
(I don't think these classics have appeared in r.p before.)
| Sysop: | Amessyroom |
|---|---|
| Location: | Fayetteville, NC |
| Users: | 76 |
| Nodes: | 6 (0 / 6) |
| Uptime: | 55:36:41 |
| Calls: | 1,265 |
| Calls today: | 1 |
| Files: | 1,362 |
| D/L today: |
3 files (2,684K bytes) |
| Messages: | 285,684 |