Thread Rating:

Poll

21 votes (45.65%)
14 votes (30.43%)
6 votes (13.04%)
3 votes (6.52%)
12 votes (26.08%)
3 votes (6.52%)
6 votes (13.04%)
5 votes (10.86%)
12 votes (26.08%)
10 votes (21.73%)

46 members have voted

EdCollins
EdCollins
  • Threads: 20
  • Posts: 1739
Joined: Oct 21, 2011
February 15th, 2021 at 4:37:40 PM permalink
Quote: Ace2

Suppose you roll a fair die until some face has appeared six times. For instance, your rolls could be 42132253261242 (for six 2's).On average, how many rolls would it take?

19.74
Wizard
Administrator
Wizard
  • Threads: 1493
  • Posts: 26504
Joined: Oct 14, 2009
February 15th, 2021 at 8:02:19 PM permalink
Quote: Ace2

Suppose you roll a fair die until some face has appeared six times. For instance, your rolls could be 42132253261242 (for six 2's).

On average, how many rolls would it take?




Integrate the following from 0 to infinity:

x*(exp(-x/6)*(1+x/6+x^2/72+x^3/1296+x^4/31104+x^5/933120))^5*exp(-x/6)*(x^5/933120)

The answer is 2597868106693535971 / 131621703842267136 = Approximation: 19.73738396371749
"For with much wisdom comes much sorrow." -- Ecclesiastes 1:18 (NIV)
Ace2
Ace2
  • Threads: 32
  • Posts: 2672
Joined: Oct 2, 2017
February 16th, 2021 at 7:11:16 AM permalink
Quote: EdCollins

19.74

Please show your calculation
It’s all about making that GTA
ThatDonGuy
ThatDonGuy
  • Threads: 117
  • Posts: 6274
Joined: Jun 22, 2011
February 16th, 2021 at 7:25:06 AM permalink
Quote: Ace2

Suppose you roll a fair die until some face has appeared six times. For instance, your rolls could be 42132253261242 (for six 2's).

On average, how many rolls would it take?



I get 2,597,868,106,693,535,971 / 131,621,703,842,267,136 = about 19.737384

This is a brute-force Markov chain problem which can be worked backwards.
Let (a,b,c,d,e,f) be the state with a 1s, b 2s, ..., f 6s
If E(a,b,c,d,e,f) is the expected number of rolls needed from state (a,b,c,d,e,f):
E(a,b,c,d,e,f) = 1 + 1/6 (E(a+1,b,c,d,e,f) + E(a,b+1,c,d,e,f) + E(a,b,c+1,d,e,f) + E(a,b,c,d+1,e,f) + E(a,b,c,d,e+1,f) + E(a,b,c,d,e,f+1))
Note E(a,b,c,d,e,f) = 0 if any of the variables = 6
If n = 7776 a + 1296 b + 216 c + 36 d + 6 e + f, and F(n) = E(a,b,c,d,e,f), then calculate F(46,655), then F(46,654), and so on down to F(0).

EdCollins
EdCollins
  • Threads: 20
  • Posts: 1739
Joined: Oct 21, 2011
February 16th, 2021 at 7:32:57 AM permalink
Quote: Ace2

Please show your calculation



Sorry. No "calculation." My answer was based upon a quick & dirty simulation I wrote via Excel.

Sub Dice_Rolls()
max_sims = 10000000: Randomize Timer:
For x = 1 To max_sims
For xx = 1 To 36
r_num1 = Int((6 - 1 + 1) * Rnd + 1)

If r_num1 = 1 Then one = one + 1
If r_num1 = 2 Then two = two + 1
If r_num1 = 3 Then three = three + 1
If r_num1 = 4 Then four = four + 1
If r_num1 = 5 Then five = five + 1
If r_num1 = 6 Then six = six + 1

If one = 6 Then
y = y + xx: Exit For
ElseIf two = 6 Then
y = y + xx: Exit For
ElseIf three = 6 Then
y = y + xx: Exit For
ElseIf four = 6 Then
y = y + xx: Exit For
ElseIf five = 6 Then
y = y + xx: Exit For
ElseIf six = 6 Then
y = y + xx: Exit For
End If
Next xx
one = 0: two = 0: three = 0: four = 0: five = 0: six = 0
Next x

MsgBox (y / max_sims)
End Sub
Ace2
Ace2
  • Threads: 32
  • Posts: 2672
Joined: Oct 2, 2017
February 16th, 2021 at 8:44:59 AM permalink
Quote: Wizard


Integrate the following from 0 to infinity:

x*(exp(-x/6)*(1+x/6+x^2/72+x^3/1296+x^4/31104+x^5/933120))^5*exp(-x/6)*(x^5/933120)

The answer is 2597868106693535971 / 131621703842267136 = Approximation: 19.73738396371749

That is the correct answer! I knew the Wizard would get this one
It’s all about making that GTA
Ace2
Ace2
  • Threads: 32
  • Posts: 2672
Joined: Oct 2, 2017
February 16th, 2021 at 8:47:23 AM permalink
Quote: EdCollins

Sorry. No "calculation." My answer was based upon a quick & dirty simulation I wrote via Excel.

A simulation is great for verifying a calculated solution, but I don't consider it a solution. Most problems are easy to simulate provided your computer has enough horsepower
Last edited by: Ace2 on Feb 16, 2021
It’s all about making that GTA
Wizard
Administrator
Wizard
  • Threads: 1493
  • Posts: 26504
Joined: Oct 14, 2009
February 16th, 2021 at 10:01:53 AM permalink
Here is my solution in more detail (PDF).

Recommended integral calculator.
"For with much wisdom comes much sorrow." -- Ecclesiastes 1:18 (NIV)
Ace2
Ace2
  • Threads: 32
  • Posts: 2672
Joined: Oct 2, 2017
February 16th, 2021 at 2:37:54 PM permalink
Quote: Wizard

Here is my solution in more detail (PDF).

Recommended integral calculator.

There is a more efficient approach to this problem. The expected waiting time to arrive at state S is the sum of the probabilities of not being in state S at all times (just like a geometric series). Using that logic, you get to the formula in one step and integrate over all time:

[{ (x/6)^5/5! + (x/6)^4/4! + (x/6)^3/3! + (x/6)^2/2 + (x/6) + 1} / e^(x/6) ]^6 dx
It’s all about making that GTA
Wizard
Administrator
Wizard
  • Threads: 1493
  • Posts: 26504
Joined: Oct 14, 2009
February 17th, 2021 at 6:59:36 AM permalink
Quote: Ace2

Recommended integral calculator.

There is a more efficient approach to this problem. The expected waiting time to arrive at state S is the sum of the probabilities of not being in state S at all times (just like a geometric series). Using that logic, you get to the formula in one step and integrate over all time:

[{ (x/6)^5/5! + (x/6)^4/4! + (x/6)^3/3! + (x/6)^2/2 + (x/6) + 1} / e^(x/6) ]^6 dx



Hmmm. Thank you. I guess I should have thought of that.

Here is a new and improved solution.
Last edited by: Wizard on Feb 17, 2021
"For with much wisdom comes much sorrow." -- Ecclesiastes 1:18 (NIV)
Gialmere
Gialmere
  • Threads: 44
  • Posts: 2942
Joined: Nov 26, 2018
February 18th, 2021 at 7:27:44 AM permalink


A white rook and a black bishop are randomly placed on an empty chessboard.

What is the probability that one of the pieces is attacking the other?
Have you tried 22 tonight? I said 22.
Ace2
Ace2
  • Threads: 32
  • Posts: 2672
Joined: Oct 2, 2017
February 18th, 2021 at 8:15:12 AM permalink
Quote: Gialmere



A white rook and a black bishop are randomly placed on an empty chessboard.

What is the probability that one of the pieces is attacking the other?

Any chance of a problem for people who don't play chess?
It’s all about making that GTA
chevy
chevy
  • Threads: 3
  • Posts: 146
Joined: Apr 15, 2011
Thanked by
Gialmere
February 18th, 2021 at 8:33:05 AM permalink


I get 1456/4032 = 36.111%

for each possible position of rook, it can attack 7 vertical + 7 horizontal = 14

for each possible position of rook, it can be attacked by bishop.....depending on location of rook
-outer "ring" of squares (8x8 border only = 28 squares) attacked by bishop from 7 positions
-next "ring" of squares (6x6 border only = 20 squares) attacked by bishop from 9 positions
-next "ring" of squares (4x4 border only = 12 squares) attacked by bishop from 11 positions
-inside block of squares (2x2 = 4 squares) attacked by bishop from 13 positions

[ (64*14)+(28*7+20*9+12*11+4*13) ] / [64*63] = 1456/4032

ThatDonGuy
ThatDonGuy
  • Threads: 117
  • Posts: 6274
Joined: Jun 22, 2011
Thanked by
Gialmere
February 18th, 2021 at 9:00:42 AM permalink
Quote: Gialmere

A white rook and a black bishop are randomly placed on an empty chessboard.

What is the probability that one of the pieces is attacking the other?




Select the bishop's space first
Each of the 28 spaces on the board edge has 7 spaces that can attack the rook
Each of the 20 spaces one space away from the board edge spaces has 9 spaces that can attack the rook
Each of the 12 spaces two spaces away has 11 spaces
The 4 spaces in the center each has 13 spaces
No matter where the bishop is placed, there are 14 spaces from where the rook can attack it, none of which are on spaces where the bishop can attack the rook.

There are 64 squares on which the bishop can be placed; for each one, there are 63 on which the rook can be placed.
The probability is (28 x (7 + 14) + 20 x (9 + 14) + 12 x (11 + 14) + 4 x (13 + 14)) / (64 x 63) = 13/36

charliepatrick
charliepatrick
  • Threads: 39
  • Posts: 2946
Joined: Jun 17, 2011
Thanked by
Gialmere
February 18th, 2021 at 9:22:27 AM permalink
Assume for simplicity the Rook is placed in one of the quarters, say NW. Then the other quarters would create identical chances.
Wherever the rook is, it can take the bishop on 7 squares on the same vertical column and 7 on the same row.
If the rook is in on the edge, i.e. A, then there are 7 places (on the diagonal) the bishop can be where it could take the rook.
B means 9 places, C means 11, D means 13.

There are 7 A's, 5 B's, 3 C's and 1 D. Thus the average number of squares where the bishop could be taken is
(7*21+5*23+3*21+1*23)/16 = 22.75.
So the chances of being taken is 22.75/63 = 91/252 (x4) = 13/36 (/7).
A A A A
A B B B
A B C C
A B C D
Gialmere
Gialmere
  • Threads: 44
  • Posts: 2942
Joined: Nov 26, 2018
February 18th, 2021 at 5:59:02 PM permalink
Quote: chevy



I get 1456/4032 = 36.111%

for each possible position of rook, it can attack 7 vertical + 7 horizontal = 14

for each possible position of rook, it can be attacked by bishop.....depending on location of rook
-outer "ring" of squares (8x8 border only = 28 squares) attacked by bishop from 7 positions
-next "ring" of squares (6x6 border only = 20 squares) attacked by bishop from 9 positions
-next "ring" of squares (4x4 border only = 12 squares) attacked by bishop from 11 positions
-inside block of squares (2x2 = 4 squares) attacked by bishop from 13 positions

[ (64*14)+(28*7+20*9+12*11+4*13) ] / [64*63] = 1456/4032


Quote: ThatDonGuy




Select the bishop's space first
Each of the 28 spaces on the board edge has 7 spaces that can attack the rook
Each of the 20 spaces one space away from the board edge spaces has 9 spaces that can attack the rook
Each of the 12 spaces two spaces away has 11 spaces
The 4 spaces in the center each has 13 spaces
No matter where the bishop is placed, there are 14 spaces from where the rook can attack it, none of which are on spaces where the bishop can attack the rook.

There are 64 squares on which the bishop can be placed; for each one, there are 63 on which the rook can be placed.
The probability is (28 x (7 + 14) + 20 x (9 + 14) + 12 x (11 + 14) + 4 x (13 + 14)) / (64 x 63) = 13/36


Quote: charliepatrick

Assume for simplicity the Rook is placed in one of the quarters, say NW. Then the other quarters would create identical chances.
Wherever the rook is, it can take the bishop on 7 squares on the same vertical column and 7 on the same row.
If the rook is in on the edge, i.e. A, then there are 7 places (on the diagonal) the bishop can be where it could take the rook.
B means 9 places, C means 11, D means 13.

There are 7 A's, 5 B's, 3 C's and 1 D. Thus the average number of squares where the bishop could be taken is
(7*21+5*23+3*21+1*23)/16 = 22.75.
So the chances of being taken is 22.75/63 = 91/252 (x4) = 13/36 (/7).
A A A A
A B B B
A B C C
A B C D


Correct!
------------------------------



---------------------------------

Quote: Ace2

Any chance of a problem for people who don't play chess?


Yes. I have been chess heavy lately but, I'm pretty much out of puzzles. I have a dice one on tap for tomorrow though.
Have you tried 22 tonight? I said 22.
Gialmere
Gialmere
  • Threads: 44
  • Posts: 2942
Joined: Nov 26, 2018
February 19th, 2021 at 7:27:04 AM permalink


A single die is rolled until a run of six different faces appears. For example, one might roll the sequence
535463261536435344151612534 with only the last six rolls all distinct.

What is the expected number of rolls?
Have you tried 22 tonight? I said 22.
EdCollins
EdCollins
  • Threads: 20
  • Posts: 1739
Joined: Oct 21, 2011
February 19th, 2021 at 9:59:49 AM permalink
Quote: Gialmere



A single die is rolled until a run of six different faces appears. For example, one might roll the sequence
535463261536435344151612534 with only the last six rolls all distinct.

What is the expected number of rolls?

My answer is...

64.8 rolls
Gialmere
Gialmere
  • Threads: 44
  • Posts: 2942
Joined: Nov 26, 2018
February 19th, 2021 at 10:44:40 AM permalink
Quote: EdCollins

My answer is...

64.8 rolls


Incorrect, but you're on the right track.
Have you tried 22 tonight? I said 22.
ThatDonGuy
ThatDonGuy
  • Threads: 117
  • Posts: 6274
Joined: Jun 22, 2011
February 19th, 2021 at 11:24:51 AM permalink
Quote: Gialmere

A single die is rolled until a run of six different faces appears. For example, one might roll the sequence
535463261536435344151612534 with only the last six rolls all distinct.

What is the expected number of rolls?



I'm not entirely sure, as I am having a little problem simulating it to confirm the answer...

Let E(N) be the expected number of rolls needed if the N most recent are all different
Suppose the last 5 rolls are 1, 2, 3, 4, 5, in order.
If the next roll is 1, then you have 2, 3, 4, 5, 1; if it is 2, you have 3, 4, 5, 2; if it is 3, you have 4, 5, 3; if it is 4, you have 5, 4; if it is 5, you have 5; if it is 6, you are done.
E(5) = 1 + 1/6 E(1) + 1/6 E(2)+ 1/6 E(3) + 1/6 E(4) + 1/6 E(5) + 1/6 x 0
Similarly, if the last 4 are 1, 2, 3, 4, then a 5 or 6 results in a "run" of 5:
E(4) = 1 + 1/6 E(1) + 1/6 E(2)+ 1/6 E(3) + 1/6 E(4) + 1/3 E(5)
E(3) = 1 + 1/6 E(1) + 1/6 E(2)+ 1/6 E(3) + 1/2 E(4)
E(2) = 1 + 1/6 E(1) + 1/6 E(2)+ 2/3 E(3)
E(1) = 1 + 1/6 E(1) + 5/6 E(2)
E(0) = 1 + E(1)
The first five can be written as five equations in five unknowns:
1/6 E(1) + 1/6 E(2) + 1/6 E(3) + 1/6 E(4) - 5/6 E(5) = -1
1/6 E(1) + 1/6 E(2) + 1/6 E(3) - 5/6 E(4) + 1/3 E(5) = -1
1/6 E(1) + 1/6 E(2) - 5/6 E(3) + 1/2 E(4) = -1
1/6 E(1) - 5/6 E(2) + 2/3 E(3) = -1
-5/6 E(1) + 5/6 E(2) = -1
Solving results in E(1) = 52.8, so the expected number = E(0) = E(1) + 1 = 53.8

Gialmere
Gialmere
  • Threads: 44
  • Posts: 2942
Joined: Nov 26, 2018
February 19th, 2021 at 12:05:23 PM permalink
Quote: ThatDonGuy

I'm not entirely sure, as I am having a little problem simulating it to confirm the answer...


Let E(N) be the expected number of rolls needed if the N most recent are all different
Suppose the last 5 rolls are 1, 2, 3, 4, 5, in order.
If the next roll is 1, then you have 2, 3, 4, 5, 1; if it is 2, you have 3, 4, 5, 2; if it is 3, you have 4, 5, 3; if it is 4, you have 5, 4; if it is 5, you have 5; if it is 6, you are done.
E(5) = 1 + 1/6 E(1) + 1/6 E(2)+ 1/6 E(3) + 1/6 E(4) + 1/6 E(5) + 1/6 x 0
Similarly, if the last 4 are 1, 2, 3, 4, then a 5 or 6 results in a "run" of 5:
E(4) = 1 + 1/6 E(1) + 1/6 E(2)+ 1/6 E(3) + 1/6 E(4) + 1/3 E(5)
E(3) = 1 + 1/6 E(1) + 1/6 E(2)+ 1/6 E(3) + 1/2 E(4)
E(2) = 1 + 1/6 E(1) + 1/6 E(2)+ 2/3 E(3)
E(1) = 1 + 1/6 E(1) + 5/6 E(2)
E(0) = 1 + E(1)
The first five can be written as five equations in five unknowns:
1/6 E(1) + 1/6 E(2) + 1/6 E(3) + 1/6 E(4) - 5/6 E(5) = -1
1/6 E(1) + 1/6 E(2) + 1/6 E(3) - 5/6 E(4) + 1/3 E(5) = -1
1/6 E(1) + 1/6 E(2) - 5/6 E(3) + 1/2 E(4) = -1
1/6 E(1) - 5/6 E(2) + 2/3 E(3) = -1
-5/6 E(1) + 5/6 E(2) = -1
Solving results in E(1) = 52.8, so the expected number = E(0) = E(1) + 1 = 53.8


Also incorrect, but also hot on the trail.
Have you tried 22 tonight? I said 22.
EdCollins
EdCollins
  • Threads: 20
  • Posts: 1739
Joined: Oct 21, 2011
February 19th, 2021 at 12:24:21 PM permalink
Ah, yes, I messed up earlier. Thank you. My second answer is approximately 77.2 (I wrote a VBA simulation to help determine the answer, since I don't know how to solve it otherwise, and I forgot to clear out some of my holding variables after each sim. That skewed the sim results completely.)


Assuming this answer is correct (or close) I find it very interesting that the number of rolls is this high. Intuitively, I'm sure I would have guessed something much lower.
Last edited by: EdCollins on Feb 19, 2021
ThatDonGuy
ThatDonGuy
  • Threads: 117
  • Posts: 6274
Joined: Jun 22, 2011
Thanked by
Gialmere
February 19th, 2021 at 12:58:16 PM permalink
Quote: Gialmere

Also incorrect, but also hot on the trail.


I think there's a bug in my equation solver - when I do it by hand, I get:


416/5 = 83.2



Found the problem in my equation solver: it panicked if, when trying to normalize row N, column N was already zero.
Gialmere
Gialmere
  • Threads: 44
  • Posts: 2942
Joined: Nov 26, 2018
Thanked by
unJon
February 19th, 2021 at 5:27:17 PM permalink
Quote: ThatDonGuy

I think there's a bug in my equation solver - when I do it by hand, I get:


416/5 = 83.2



Found the problem in my equation solver: it panicked if, when trying to normalize row N, column N was already zero.


Correct!

It's interesting you had to solve it by hand. I guess that, even in this high tech age, John Henry will still occasionally beat the steam drill.


---------------------------------------

Have you tried 22 tonight? I said 22.
teliot
teliot
  • Threads: 43
  • Posts: 2871
Joined: Oct 19, 2009
February 21st, 2021 at 6:05:13 AM permalink
For this question, watching this video is absolutely encouraged!



So here's the question: Is there a value x=t so that if Gabriel's Horn goes from x=1 to x=t, then the volume and surface area of Gabriel's Horn are equal? If so, can you solve for this value?
Climate Casino: https://climatecasino.net/climate-casino/
unJon
unJon
  • Threads: 14
  • Posts: 4603
Joined: Jul 1, 2018
February 21st, 2021 at 7:41:17 AM permalink
Quote: teliot

For this question, watching this video is absolutely encouraged!



So here's the question: Is there a value x=t so that if Gabriel's Horn goes from x=1 to x=t, then the volume and surface area of Gabriel's Horn are equal? If so, can you solve for this value?



Cool video. Though it left my questioning why you can get away with using dx to integrate the volume instead of needing to use ds. Likewise they made a big deal of needing to use ds to integrate for surface area yet they could have ignored that complicating factor and used dx and still shown it was infinite. In fact that’s what they did at the end by showing integers of 1 to infinity of 2piR was infinite.

The paradox further messes with me. If pi units of paint fills the horn, then it must coat all the interior surfaces of the horn. Yet the interior surfaces of the horn have area of infinity. But we painted them!

As to your question, need coffee and paper and pencil before attempting. The video gives the necessary equations so shouldn’t be difficult to attempt to set them equal to one another.
The race is not always to the swift, nor the battle to the strong; but that is the way to bet.
teliot
teliot
  • Threads: 43
  • Posts: 2871
Joined: Oct 19, 2009
February 21st, 2021 at 8:12:14 AM permalink
Quote: unJon

Cool video. Though it left my questioning why you can get away with using dx to integrate the volume instead of needing to use ds.

I completely agree with this observation. That part of the video was lame for the infinite horn. On the other hand, maybe you need that full formula for the finite case? Come to think of it, maybe the whole problem is much tougher than I originally thought ... maybe showing that such an x=t exists is all you can do.

And yes, filling the volume with paint should include the paint touching the surface, ergo painting it by default. That's what makes this a paradox.
Last edited by: teliot on Feb 21, 2021
Climate Casino: https://climatecasino.net/climate-casino/
ThatDonGuy
ThatDonGuy
  • Threads: 117
  • Posts: 6274
Joined: Jun 22, 2011
Thanked by
teliot
February 21st, 2021 at 8:23:51 AM permalink
Quote: teliot

So here's the question: Is there a value x=t so that if Gabriel's Horn goes from x=1 to x=t, then the volume and surface area of Gabriel's Horn are equal? If so, can you solve for this value?



The only value where they are equal is t = 1 (i.e. both volume and surface area are zero), and even this assumes that the disc of radius one at t = 1 is not included in the area.

The surface area = INTEGRAL[1, t] (2 PI / t) dt = 2 PI INTEGRAL[1, t] (1/t) dt = 2 PI (ln t - ln 1) = 2 PI ln t.
The volume = INTEGRAL[1, t] (PI / t^2) dt = PI INTEGRAL[1, t] (1/t^2) dt = PI (-1/t + 1/1) = PI (1 - 1/t).
These are equal when 2 ln t = 1 - 1/t.
2 ln 1 = 1 - 1/1, so they are equal when t = 1.
If t > 1, then they are equal when 2 ln t + 1/t - 1 = 0.
Let f(t) = 2 ln t + 1/t - 1
df/dt = 2/t + 1/t^2; since this is positive for all t > 1, f(t) is strictly increasing for all t > 1, which means f(t) > f(1).

If the disc is included, then the surface area = 2 ln t PI (2 ln t + 1), and they are equal when ln t = -1/t, but for t > 1, ln t > 0 and -1/t < 0.


Another thing that makes it paradoxical: if you have another horn that is y = 2/x rotated around the x-axis, the volume between the two is finite, so you can "paint the outside" of the original horn by filling the gap between the two.
Last edited by: ThatDonGuy on Feb 21, 2021
teliot
teliot
  • Threads: 43
  • Posts: 2871
Joined: Oct 19, 2009
February 21st, 2021 at 8:34:37 AM permalink
Quote: ThatDonGuy


The only value where they are equal is t = 1 (i.e. both volume and surface area are zero), and even this assumes that the disc of radius one at t = 1 is not included in the area.

The surface area = INTEGRAL[1, t] (2 PI / t) dt = 2 PI INTEGRAL[1, t] (1/t) dt = 2 PI (ln t - ln 1) = 2 PI ln t.
The volume = INTEGRAL[1, t] (PI / t^2) dt = PI INTEGRAL[1, t] (1/t^2) dt = PI (-1/t + 1/1) = PI (1 - 1/t).
These are equal when 2 ln t = 1 - 1/t.
2 ln 1 = 1 - 1/1, so they are equal when t = 1.
If t > 1, then they are equal when 2 ln t + 1/t - 1 = 0.
Let f(t) = 2 ln t + 1/t - 1
df/dt = 2/t + 1/t^2; since this is positive for all t > 1, f(t) is strictly increasing for all t > 1, which means f(t) > f(1).

If the disc is included, then the surface area = 2 ln t, and they are equal when ln t = -1/t, but for t > 1, ln t > 0 and -1/t < 0.


Your solution was my original thought on this. But now, I disagree about the surface area calculation you've given ... thinking out loud, if you replace your formula with "The Surface Area > ..." and the rest of your argument works.
Climate Casino: https://climatecasino.net/climate-casino/
chevy
chevy
  • Threads: 3
  • Posts: 146
Joined: Apr 15, 2011
Thanked by
unJon
February 21st, 2021 at 10:17:10 AM permalink


I guess I have always disagreed with this as a "paradox". The problem I always had was with equating surface area to volume. Gabriel's Horn hides this in a calculation where we concentrate on one integral being finite but another is infinite. But consider the x-y plane. The surface area of that is infinite, but the volume of paint required to "paint" it is 0. In the case of Gabriel's Horn, you go out and buy your can of paint with volume, pi "units", you paint the inside (or outside) using up 0 "units", and you still have pi "units" of paint left to fill the horn with.

As to Teliot's original question, I agree with the calculations Teliot and ThatDonGuy discussed...but if I go back to comparing surface area to volume, does equating them have some physical or geometric meaning I am missing?

unJon
unJon
  • Threads: 14
  • Posts: 4603
Joined: Jul 1, 2018
February 21st, 2021 at 10:22:05 AM permalink
Quote: chevy



I guess I have always disagreed with this as a "paradox". The problem I always had was with equating surface area to volume. Gabriel's Horn hides this in a calculation where we concentrate on one integral being finite but another is infinite. But consider the x-y plane. The surface area of that is infinite, but the volume of paint required to "paint" it is 0. In the case of Gabriel's Horn, you go out and buy your can of paint with volume, pi "units", you paint the inside (or outside) using up 0 "units", and you still have pi "units" of paint left to fill the horn with.

As to Teliot's original question, I agree with the calculations Teliot and ThatDonGuy discussed...but if I go back to comparing surface area to volume, does equating them have some physical or geometric meaning I am missing?



These are good points. But there is a natural relationship between surface area and volume. And, for me, there is some impact to the paradox. Like this: here is an infinite amount of paper for you to take and use all of it but create a shape with a finite amount of volume. You could not take that paper and make a sphere with finite volume. But you could make a Gabriel’s Horn. To your point you could also just lie it flat and make a plane (though no volume really “enclosed” there).
The race is not always to the swift, nor the battle to the strong; but that is the way to bet.
ThatDonGuy
ThatDonGuy
  • Threads: 117
  • Posts: 6274
Joined: Jun 22, 2011
February 21st, 2021 at 10:47:09 AM permalink
Quote: teliot

Quote: ThatDonGuy


The only value where they are equal is t = 1 (i.e. both volume and surface area are zero), and even this assumes that the disc of radius one at t = 1 is not included in the area.

The surface area = INTEGRAL[1, t] (2 PI / t) dt = 2 PI INTEGRAL[1, t] (1/t) dt = 2 PI (ln t - ln 1) = 2 PI ln t.
The volume = INTEGRAL[1, t] (PI / t^2) dt = PI INTEGRAL[1, t] (1/t^2) dt = PI (-1/t + 1/1) = PI (1 - 1/t).
These are equal when 2 ln t = 1 - 1/t.
2 ln 1 = 1 - 1/1, so they are equal when t = 1.
If t > 1, then they are equal when 2 ln t + 1/t - 1 = 0.
Let f(t) = 2 ln t + 1/t - 1
df/dt = 2/t + 1/t^2; since this is positive for all t > 1, f(t) is strictly increasing for all t > 1, which means f(t) > f(1).

If the disc is included, then the surface area = 2 ln t, and they are equal when ln t = -1/t, but for t > 1, ln t > 0 and -1/t < 0.


Your solution was my original thought on this. But now, I disagree about the surface area calculation you've given ... thinking out loud, if you replace your formula with "The Surface Area > ..." and the rest of your argument works.



I can see where the surface area in "if the disc is included" is wrong; it should be PI (2 ln t + 1). Is that what you were referring to?
chevy
chevy
  • Threads: 3
  • Posts: 146
Joined: Apr 15, 2011
February 21st, 2021 at 10:55:24 AM permalink
unJon:

Interesting comparison of using infinite paper to create a finite volume Gabriel's Horn, but not a sphere. To that end I concede, there are intrinsic relationships between surface area and volume.

I think I like your implied style "Can you make an object with infinite surface area and finite (non-zero) volume?" as a better wording than the video's "paint it" paradox.

My objection was 2D vs 3D, and not really an infinity issue. My objection still holds for a 1x1x1 cube....takes zero volume of paint to coat the inside(surface area), but 1 volume unit of paint to fill it. We wouldn't compare 6 "units" to coat it with 1 "unit" to fill it.
teliot
teliot
  • Threads: 43
  • Posts: 2871
Joined: Oct 19, 2009
February 21st, 2021 at 11:02:56 AM permalink
Quote: ThatDonGuy

Quote: teliot

Quote: ThatDonGuy


The only value where they are equal is t = 1 (i.e. both volume and surface area are zero), and even this assumes that the disc of radius one at t = 1 is not included in the area.

The surface area = INTEGRAL[1, t] (2 PI / t) dt = 2 PI INTEGRAL[1, t] (1/t) dt = 2 PI (ln t - ln 1) = 2 PI ln t.
The volume = INTEGRAL[1, t] (PI / t^2) dt = PI INTEGRAL[1, t] (1/t^2) dt = PI (-1/t + 1/1) = PI (1 - 1/t).
These are equal when 2 ln t = 1 - 1/t.
2 ln 1 = 1 - 1/1, so they are equal when t = 1.
If t > 1, then they are equal when 2 ln t + 1/t - 1 = 0.
Let f(t) = 2 ln t + 1/t - 1
df/dt = 2/t + 1/t^2; since this is positive for all t > 1, f(t) is strictly increasing for all t > 1, which means f(t) > f(1).

If the disc is included, then the surface area = 2 ln t, and they are equal when ln t = -1/t, but for t > 1, ln t > 0 and -1/t < 0.


Your solution was my original thought on this. But now, I disagree about the surface area calculation you've given ... thinking out loud, if you replace your formula with "The Surface Area > ..." and the rest of your argument works.



I can see where the surface area in "if the disc is included" is wrong; it should be PI (2 ln t + 1). Is that what you were referring to?

if you want an exact value for the surface area over a finite interval, you cannot disregard the term in the surface area integral that is disregarded in the video in the infinite upper limit case.

For example, find the limit x = t that gives a surface area equal to twice the volume. I have no idea how to do this.
Climate Casino: https://climatecasino.net/climate-casino/
unJon
unJon
  • Threads: 14
  • Posts: 4603
Joined: Jul 1, 2018
February 21st, 2021 at 11:15:58 AM permalink
Quote: chevy


My objection was 2D vs 3D, and not really an infinity issue. My objection still holds for a 1x1x1 cube....takes zero volume of paint to coat the inside(surface area), but 1 volume unit of paint to fill it. We wouldn't compare 6 "units" to coat it with 1 "unit" to fill it.



Yes agree with your point here.
The race is not always to the swift, nor the battle to the strong; but that is the way to bet.
teliot
teliot
  • Threads: 43
  • Posts: 2871
Joined: Oct 19, 2009
February 21st, 2021 at 11:16:43 AM permalink
Quote: chevy

unJon:

Interesting comparison of using infinite paper to create a finite volume Gabriel's Horn, but not a sphere. To that end I concede, there are intrinsic relationships between surface area and volume.

I think I like your implied style "Can you make an object with infinite surface area and finite (non-zero) volume?" as a better wording than the video's "paint it" paradox.

My objection was 2D vs 3D, and not really an infinity issue. My objection still holds for a 1x1x1 cube....takes zero volume of paint to coat the inside(surface area), but 1 volume unit of paint to fill it. We wouldn't compare 6 "units" to coat it with 1 "unit" to fill it.

Then space filling curves should be equally objectionable. A one dimensional line filling two-dimensional space.
Climate Casino: https://climatecasino.net/climate-casino/
ThatDonGuy
ThatDonGuy
  • Threads: 117
  • Posts: 6274
Joined: Jun 22, 2011
February 21st, 2021 at 12:53:35 PM permalink
Quote: teliot

if you want an exact value for the surface area over a finite interval, you cannot disregard the term in the surface area integral that is disregarded in the video in the infinite upper limit case.

For example, find the limit x = t that gives a surface area equal to twice the volume. I have no idea how to do this.


Maybe I am calculating it wrong, but I thought the surface area = the integral that sums the circumferences of the circles from x = 1 to x = t.
The circumference of the circle at x is 2 PI (1/x), whose integral with respect to x is 2 PI ln (x), so the integral over x = 1 to t is 2 PI ln t - 2 PI ln 1 = 2 PI (ln t - 0) = 2 PI ln t.
Then again, that method doesn't seem to work when I try it with calculating the surface area of a sphere...
teliot
teliot
  • Threads: 43
  • Posts: 2871
Joined: Oct 19, 2009
February 21st, 2021 at 1:51:16 PM permalink
Quote: ThatDonGuy

Maybe I am calculating it wrong, but I thought the surface area = the integral that sums the circumferences of the circles from x = 1 to x = t.
The circumference of the circle at x is 2 PI (1/x), whose integral with respect to x is 2 PI ln (x), so the integral over x = 1 to t is 2 PI ln t - 2 PI ln 1 = 2 PI (ln t - 0) = 2 PI ln t.
Then again, that method doesn't seem to work when I try it with calculating the surface area of a sphere...

watch the video @13:30 and you will see what's going wrong with your calculation method.
Climate Casino: https://climatecasino.net/climate-casino/
charliepatrick
charliepatrick
  • Threads: 39
  • Posts: 2946
Joined: Jun 17, 2011
February 21st, 2021 at 2:20:24 PM permalink
Occasionally The [London] Times publishes a maths puzzle. This one is NOT easy (although easier than some they have had) and usually you need to find the first key and just work your way through (rather like suduko). Using programming/excel is optional, but I know people who can solve it just using a calculator. (Typically they have clues where you have primes, squares etc and know some of the digits. By pure fluke there's only one answer that fits. Luckily this one isn't quite that devious.)

I have found this link with a PDF

(If this doesn't work the link is here https://www.crosswordsolver.org/forum/799536/listener-4647/ but the page might have spoilers lower down as it's a forum.)
DogHand
DogHand
  • Threads: 2
  • Posts: 1526
Joined: Sep 24, 2011
February 21st, 2021 at 2:47:54 PM permalink
Quote: teliot

watch the video @13:30 and you will see what's going wrong with your calculation method.



The surface area for a Gabriel's Horn of length t is the integral from 1 to t of 2*pi*sqrt(1-1/x^4)/x.

If for simplicity we let a == sqrt(1+1/t^4) and b == sqrt(2) then this integral is

Area = (pi/2)*(2*(b-a) + ln((1+a)*(b-1)/((-1+a)*(b+1))))

As shown in the video,

Volume = pi*(1-1/t)

Set Area =Volume and calculate t.

As ThatDonGuy indicated earlier in the thread, the only value of t that makes them equal is t=1

Dog Hand
chevy
chevy
  • Threads: 3
  • Posts: 146
Joined: Apr 15, 2011
February 21st, 2021 at 5:19:58 PM permalink
Quote: teliot

Then space filling curves should be equally objectionable. A one dimensional line filling two-dimensional space.



Perhaps it should. Certainly counter-intuitive, but somehow comparing cardinality of two infinite sets (number of points on unit interval to number of points in unit square (for example)) doesn't bother me as much.

I honestly don't know much about space-filling curves beyond a quick wikipedia read. Maybe my mind is limited to differentiable cases.

(From Wikipedia)

Space-filling curves are special cases of fractal curves. No differentiable space-filling curve can exist. Roughly speaking, differentiability puts a bound on how fast the curve can turn.
gordonm888
Administrator
gordonm888
  • Threads: 60
  • Posts: 5055
Joined: Feb 18, 2015
February 21st, 2021 at 11:06:46 PM permalink
How can you fill the horn with paint without painting the interior surface?

Well, there is a parameter that we are missing and assuming is everywhere equal - the thickness of the coat of paint.

As the flare of the horn stretches out to infinity, the finite thickness of a coat of paint will eventually place some of the coat outside of the "defined volume" of the interior of the horn. Even if the thickness of the coat is one molecule in width, there is an infinite length of horn's flare that is closer to the "length" of the horn than one molecular width.

The problem is really in saying that we fill the horn with paint -because the paint volume becomes infinitely thin as the horn flares out to infinity. Thus, the paint itself must progressively become infinitely thin as the horn flares to infinity. It is this requirement for infinite thinning of the paint layer on the interior of horn that allows a finite volume of paint to cover an infinite surface area.
So many better men, a few of them friends, are dead. And a thousand thousand slimy things live on, and so do I.
marcel55
marcel55
  • Threads: 0
  • Posts: 18
Joined: Feb 21, 2021
February 22nd, 2021 at 5:01:50 AM permalink
Correct man but they are other ways as other people are doing it with other methods also. You can have a view on them also.
Gialmere
Gialmere
  • Threads: 44
  • Posts: 2942
Joined: Nov 26, 2018
February 22nd, 2021 at 7:46:43 AM permalink
Hmm. I'm not sure what the status of this thread is but, here is a classic math puzzle (which I hope hasn't been posted before)...



There is a building of 100 floors
-If an egg drops from the Nth floor or above it will break.
-If it's dropped from any floor below, it will not break.
You're given 2 eggs.

How do you find N in the minimum number of drops?
Have you tried 22 tonight? I said 22.
teliot
teliot
  • Threads: 43
  • Posts: 2871
Joined: Oct 19, 2009
February 22nd, 2021 at 8:33:31 AM permalink
Quote: gordonm888

How can you fill the horn with paint without painting the interior surface?

Well, there is a parameter that we are missing and assuming is everywhere equal - the thickness of the coat of paint.

As the flare of the horn stretches out to infinity, the finite thickness of a coat of paint will eventually place some of the coat outside of the "defined volume" of the interior of the horn. Even if the thickness of the coat is one molecule in width, there is an infinite length of horn's flare that is closer to the "length" of the horn than one molecular width.

The problem is really in saying that we fill the horn with paint -because the paint volume becomes infinitely thin as the horn flares out to infinity. Thus, the paint itself must progressively become infinitely thin as the horn flares to infinity. It is this requirement for infinite thinning of the paint layer on the interior of horn that allows a finite volume of paint to cover an infinite surface area.

I'm trying to imagine the thickness of the paint out near x=Tree(3) ...

No, really, infinite surface area and finite volume have nothing to do with painting, but it's the painting that makes it conceptually difficult to believe.
Climate Casino: https://climatecasino.net/climate-casino/
ThatDonGuy
ThatDonGuy
  • Threads: 117
  • Posts: 6274
Joined: Jun 22, 2011
February 22nd, 2021 at 8:52:36 AM permalink
Quote: Gialmere

Hmm. I'm not sure what the status of this thread is but, here is a classic math puzzle (which I hope hasn't been posted before)...
There is a building of 100 floors
-If an egg drops from the Nth floor or above it will break.
-If it's dropped from any floor below, it will not break.
You're given 2 eggs.

How do you find N in the minimum number of drops?


I think it has been posted before - in any event, I'm pretty sure I've heard of it, so I'll post what I think is a solution, although I don't have a definitive proof of it yet.

Drop the first egg from the 1st, 3rd, 6th, 10th, ..., floors until either it breaks or you drop it from the 91st floor without breaking.
Go back to 1 floor above the highest one from which you dropped the first egg without breaking, and drop the second egg from each floor until it does.
For example, if it breaks on the 45th floor, the last successful floor was 36, so drop the second egg from the 37th, 38th, 39th,... flooors until it breaks.

Joeman
Joeman
  • Threads: 36
  • Posts: 2414
Joined: Feb 21, 2014
February 22nd, 2021 at 9:01:43 AM permalink
The best way to find N is to drop the first egg from floors at regular intervals until it breaks. Then, with the second egg, start with the the floor above the highest floor from which the first egg did not break, and go up by 1 until it breaks. The floor from which the second egg breaks will be floor 'N.'

The (worst case) number of drops required for this can be represented by the function:

H/x + (x-1)

Where x is the chosen interval, and H is the height of the building in terms of floors.

To find the minimum, set the first derivative of the above function equal to zero:

-H/x2 +1 = 0, which simplifies to x = H1/2

Given H = 100, x must be 10. So, we would then drop the first egg from every 10th floor (starting at floor 10) until it breaks. Then, drop the second egg from the floor directly above where the penultimate drop took place, and go up one floor at a time for subsequent drops until it breaks, indicating floor 'N.' This will take a maximum of 19 drops (when N=99).
"Dealer has 'rock'... Pay 'paper!'"
charliepatrick
charliepatrick
  • Threads: 39
  • Posts: 2946
Joined: Jun 17, 2011
Thanked by
Gialmere
February 22nd, 2021 at 9:16:56 AM permalink
Here are various plans - starting with the worst!

Plan A: It seems reasonable that with one egg you could drop from floor 1, floor 2... and just work your way up until you find N. Average number of drops about 50.

With two eggs Plan B, not optimal, is to narrow that range to either 0-50 or 51-100 by dropping one from floor 50, and repeating as above. Average number of drops about 25 + 1. A similar plan is to go up in even numbers (dropping from floor 2,4,6 etc and then when the egg breaks see if it would have dropped from one floor lower. Similarly 25+1.

Plan C, not optimal, is to drop the egg from floor 33. If it breaks go from 0-32. If it doesn't try floor 66. This has a lower average.

Plan D, not optimal, is to drop egg from 10,20,30 etc then hone in on the last digit with the second egg. While this is quicker for low numbers, it takes quite a while with the higher ones.

Plan E, is the best I have found so far (although I don't know how to prove whether it's a minimum).

Suppose you try to ensure the worst case scenario is you needed 14 drops, then that might be better. So for instance if your first drop was on floor 14, then it could take a worst case of 14 to find the floor (14,1,2,3...13). However If you had more drops before the first egg broke, then you might want the gaps smaller. So one plan is drop an egg at 14,14+13=27,27+12=39 etc. and the work your way up. 14 27 39 50 60 69 77 84 90 95 98 seems a good plan and ensures the worst case is 14, I think this gives an average of about 10.35. (I've also ignored that once 99 doesn't break you know it's 100, so the average would be 10.34.)
Wizard
Administrator
Wizard
  • Threads: 1493
  • Posts: 26504
Joined: Oct 14, 2009
Thanked by
Gialmere
February 22nd, 2021 at 9:33:17 AM permalink
Quote: Gialmere

How do you find N in the minimum number of drops?



Are you asking for the minimum mean N or the minimum maximum N?

If it's the maximum N, then I get the same answer as Don. Looks like the maximum number of drops needed is ...

14


It's probably the same answer as for the mean N as well.
"For with much wisdom comes much sorrow." -- Ecclesiastes 1:18 (NIV)
Gialmere
Gialmere
  • Threads: 44
  • Posts: 2942
Joined: Nov 26, 2018
February 22nd, 2021 at 9:51:50 AM permalink
Quote: charliepatrick

Here are various plans - starting with the worst!

Plan A: It seems reasonable that with one egg you could drop from floor 1, floor 2... and just work your way up until you find N. Average number of drops about 50.

With two eggs Plan B, not optimal, is to narrow that range to either 0-50 or 51-100 by dropping one from floor 50, and repeating as above. Average number of drops about 25 + 1. A similar plan is to go up in even numbers (dropping from floor 2,4,6 etc and then when the egg breaks see if it would have dropped from one floor lower. Similarly 25+1.

Plan C, not optimal, is to drop the egg from floor 33. If it breaks go from 0-32. If it doesn't try floor 66. This has a lower average.

Plan D, not optimal, is to drop egg from 10,20,30 etc then hone in on the last digit with the second egg. While this is quicker for low numbers, it takes quite a while with the higher ones.

Plan E, is the best I have found so far (although I don't know how to prove whether it's a minimum).

Suppose you try to ensure the worst case scenario is you needed 14 drops, then that might be better. So for instance if your first drop was on floor 14, then it could take a worst case of 14 to find the floor (14,1,2,3...13). However If you had more drops before the first egg broke, then you might want the gaps smaller. So one plan is drop an egg at 14,14+13=27,27+12=39 etc. and the work your way up. 14 27 39 50 60 69 77 84 90 95 98 seems a good plan and ensures the worst case is 14, I think this gives an average of about 10.35. (I've also ignored that once 99 doesn't break you know it's 100, so the average would be 10.34.)


Quote: Wizard

Are you asking for the minimum mean N or the minimum maximum N?

If it's the maximum N, then I get the same answer as Don. Looks like the maximum number of drops needed is ...

14


It's probably the same answer as for the mean N as well.



Correct!

We're looking for the maximum and 14 drops it is. This problem is evidently one of those Google interview questions.

----------------------------------

Saw a sign at a farm that said, "duck, eggs."

I was contemplating the use of the comma when it hit me.
Have you tried 22 tonight? I said 22.
  • Jump to: