• Log InLog In
  • Register
Liquid`
Team Liquid Liquipedia
EDT 16:25
CET 21:25
KST 05:25
  • Home
  • Forum
  • Calendar
  • Streams
  • Liquipedia
  • Features
  • Store
  • EPT
  • TL+
  • StarCraft 2
  • Brood War
  • Smash
  • Heroes
  • Counter-Strike
  • Overwatch
  • Liquibet
  • Fantasy StarCraft
  • TLPD
  • StarCraft 2
  • Brood War
  • Blogs
Forum Sidebar
Events/Features
News
Featured News
[ASL22] Finals Preview: Sweet Victory!12[ASL22] Ro4 Preview: Mirror Mirror6[ASL22] Ro8 Preview: Within Reach5[ASL22] Ro8 Preview: In A Tizzy11[ASL22] Ro16 Preview: Holy Diver5
Community News
Weekly Cups (October 19-25): Clem triples first in 5.0.174Evil Geniuses - shutting down20ISSL (IntoTheTV x SOOP SC2 League): Premier5Weekly Cups (October 12-18): 5.0.17 era begins1[$1250] International ShieldBattery League92
StarCraft 2
General
How do you feel about the mass reverts in the 5.0.17 PTR? Weekly Cups (October 19-25): Clem triples first in 5.0.17 Evil Geniuses - shutting down REPORT: StarCraft 2: Remastered In Development No more SC2 streamers on team liquid?
Tourneys
Stellar Fest TWO the Moon (Dec 16-20) RSL 2026 Offline Finals LCQ Sign-Ups Sparkling Tuna Cup - Weekly Open Tournament 2026 KungFu Cup Announcement RSL goes to London! 2026 Offline Finals Nov 21-22
Strategy
[H] ZvP Mid-Late Game: Stalkers Collossi HT
Custom Maps
Simple Questions/Answers [G] How to playtest arcade maps [A] Nemrods 1/4 players
External Content
Mutation # 549 Endless Sparkles The PondCast: SC2 News & Results Mutation # 548 Shining Bright Mutation # 547 Temple Of Terror
Brood War
General
Sagi.gg Launcher Released SC4ALL II Brood War Complete Invited Player Lineup BW General Discussion Starcraft player cRuZa Passed away. BGH Auto Balance -> http://bghmmr.eu/
Tourneys
[IPSL] IPSL Winter 26-27 Ro24 is HERE! [Megathread] Daily Proleagues KCM Race Survival 2026 Season 4 [BSL23] RO24 Group D - Sunday 21:00 CET
Strategy
50,000-Word Terran Builds Memo 2026-10-22 Cliff Jump Revisited (1 in a 1000 strategy) Simple Questions, Simple Answers
Other Games
General Games
Nintendo Switch Thread Stormgate/Frost Giant Megathread Mobile Legends: Bang Bang Armies of Exigo - YesYes? General RTS Discussion Thread
Dota 2
Dota 2 Champions League Season 3 Begins April 25!
League of Legends
[TL LoL EUW IHs] Teemo shall perish
Heroes of the Storm
Heroes of the Storm 2.0
Hearthstone
Deck construction bug
TL Mafia
TL Mafia Community Thread
Community
General
R I P Patrick Steward US Politics Mega-thread European Politico-economics QA Mega-thread Things Aren’t Peaceful in Palestine Russo-Ukrainian War Thread
Fan Clubs
Reynor Fan Club The Flash Fan Club Serral Fan Club
Media & Entertainment
Anime Discussion Thread Movie Discussion! [Manga] One Piece
Sports
Football (Soccer) Thread TeamLiquid Health and Fitness Initiative For 2023 MLB/Baseball 2023
World Cup 2022
Tech Support
Computer Build, Upgrade & Buying Resource Thread
TL Community
Recent Gifted Posts
Blogs
Innovate or Game Over?
TrAiDoS
ShieldBattery on Mac
Kraekkling
[ASL22] Ro4 Day1 Ticket Giv…
bITt.mAN
Can Bots Beat Pros?? Starcr…
namkraft
Customize Sidebar...

Website Feedback

Closed Threads



Active: 10622 users

100 Prisoners Problem - Page 2

Blogs > Slithe
Post a Reply
Prev 1 2 3 Next All
fanatacist
Profile Blog Joined August 2007
10319 Posts
January 30 2008 19:42 GMT
#21
On January 31 2008 04:39 Cascade wrote:
hmm, I'm struggling with this one... I cant really find any better ideas than the ones mention. Trying a brute force method now, but I saw the 20% spoiler and I think I can prove that that kinid of survival probabilys are impossible by using a best case scenario:

Fist guys probability is 50%. No way around that.

Second guy's best shot (for himself alone) is clearly to take the other 50 boxes, in which case he will get 50/99.

Best possible case for third guy is if he knows that the two first guys names NOT were in his boxes. This is clearly not possible to combine with maximising second guys chances, but this is an upper bound. So third guys chances are AT BEST 50/98.

Similarly forth guy will have at best 50/97 until 50:th guy that at best know that the previous 49 guys names are in the other boxes, and he'll get 50/51. guy number 51 and abova can all be guaranteed (in best case for them alone) to pick the right boxes.

So total, if we take maximum survival probability for each person to survive, which is clearly not accievable, we get:

50/100 * 50/99 * 50/98 * ... 50/51 = 50^50 * 50! / 100! = 2.9.. * 10^(-9)

So either my best case calculation is flawed, or the estimate of 20% is way of.

I'll go work on the brute force a bit more now. :/ let you know if i find anything.

I'm thinking there is a way to eliminate more than 1 box as wrong for each prisoner's specific name for each 1 entrant into the room. In other words, perhaps there is a method that would make the second guy's probability not 50/99 but 50/80, just to give a random number.

And someone else said it's over 30%.
Peace~
Lemonwalrus
Profile Blog Joined August 2006
United States5465 Posts
January 30 2008 19:45 GMT
#22
Here is the actual answer, please don't read unless you totally give up. Think outside of the box. (sorry, I couldn't resist. )
+ Show Spoiler +
Last chance, if you turn back now, nobody has to know you were here.
+ Show Spoiler +
Link
Cascade
Profile Blog Joined March 2006
Australia5405 Posts
Last Edited: 2008-01-30 19:56:30
January 30 2008 19:54 GMT
#23
On January 31 2008 04:42 fanatacist wrote:
Show nested quote +
On January 31 2008 04:39 Cascade wrote:
hmm, I'm struggling with this one... I cant really find any better ideas than the ones mention. Trying a brute force method now, but I saw the 20% spoiler and I think I can prove that that kinid of survival probabilys are impossible by using a best case scenario:

Fist guys probability is 50%. No way around that.

Second guy's best shot (for himself alone) is clearly to take the other 50 boxes, in which case he will get 50/99.

Best possible case for third guy is if he knows that the two first guys names NOT were in his boxes. This is clearly not possible to combine with maximising second guys chances, but this is an upper bound. So third guys chances are AT BEST 50/98.

Similarly forth guy will have at best 50/97 until 50:th guy that at best know that the previous 49 guys names are in the other boxes, and he'll get 50/51. guy number 51 and abova can all be guaranteed (in best case for them alone) to pick the right boxes.

So total, if we take maximum survival probability for each person to survive, which is clearly not accievable, we get:

50/100 * 50/99 * 50/98 * ... 50/51 = 50^50 * 50! / 100! = 2.9.. * 10^(-9)

So either my best case calculation is flawed, or the estimate of 20% is way of.

I'll go work on the brute force a bit more now. :/ let you know if i find anything.

I'm thinking there is a way to eliminate more than 1 box as wrong for each prisoner's specific name for each 1 entrant into the room. In other words, perhaps there is a method that would make the second guy's probability not 50/99 but 50/80, just to give a random number.

And someone else said it's over 30%.


Yeah i must ahve missed something if you get to 30%.

edit2: also better spoiler this, since it may contain a hint...
+ Show Spoiler +

hmm, probably it is the fact that they do not choose all boxes at once, but they open one at a time. Then they can use the name in the first box they open, to choose which box to open afterwards. yeah, that's must be it... brb.


EDIT: I'm NOT READING THAT SPOILER!
fanatacist
Profile Blog Joined August 2007
10319 Posts
January 30 2008 19:56 GMT
#24
If they do not know who is going in before/after them, do they at least know if they are first, second, third, etc.?
Peace~
Lemonwalrus
Profile Blog Joined August 2006
United States5465 Posts
Last Edited: 2008-01-30 20:02:18
January 30 2008 20:01 GMT
#25
On January 31 2008 04:56 fanatacist wrote:
If they do not know who is going in before/after them, do they at least know if they are first, second, third, etc.?


Don't worry, this is only in a spoiler because it kinda gives a hint.
+ Show Spoiler +
I am not sure of the answer to that question, but the best I can tell you is that the answer, whether it be yes or no, is irrelevant.
fanatacist
Profile Blog Joined August 2007
10319 Posts
January 30 2008 20:06 GMT
#26
On January 31 2008 05:01 Lemonwalrus wrote:
Show nested quote +
On January 31 2008 04:56 fanatacist wrote:
If they do not know who is going in before/after them, do they at least know if they are first, second, third, etc.?


Don't worry, this is only in a spoiler because it kinda gives a hint.
+ Show Spoiler +
I am not sure of the answer to that question, but the best I can tell you is that the answer, whether it be yes or no, is irrelevant.

Thanks [:

Very manner problem solvers here! n_n
Peace~
Lemonwalrus
Profile Blog Joined August 2006
United States5465 Posts
January 30 2008 20:10 GMT
#27
I just know that it feels really good to solve something like this on your own without help.
FieryBalrog
Profile Blog Joined July 2007
United States1381 Posts
January 30 2008 20:25 GMT
#28
OK, I looked at the solution. Couldn't resist :p

This is only in spoiler in case it gives a hint:

+ Show Spoiler +
I would have never have got it, I didn't understand the mathematics behind the explanation, I'll believe it though
I will eat you alive
Cascade
Profile Blog Joined March 2006
Australia5405 Posts
Last Edited: 2008-01-30 21:09:56
January 30 2008 21:06 GMT
#29
ok, I think I have found the solution.

+ Show Spoiler +

Prisoner number a starts a box number a. He will there find the name of prisoner number b, and he will next open box number b, and so on.

This will work as long as there are no "loops" longer than 50 among the boxes. A loop is when you follow boxes with the algorithm above, you will eventually come back to where you started. Which is your number! So as long as you finish one lap in your loop within 50 steps, you're fine. If the loop is longer than 50, everyone with numbers in that loop will fail, which is what correlates the success of the prissoners. and since this is everyone or noone thingy, you want as much correlation as possible.

Next thing would be to calculate the probability of not having loops longer than 50. But I wont do that. If this is correct solution, the calculation will be in walrus' spoiler. if im wrong, it wont make snece to calculate it. But 30% sounds plausible at first sight.

Then proving that there are no other better strategies... no idea right now. Maybe you can make som smart argument, or it will be very complicated.
I'll have a look at the spoiler anyway.


yay, it was right! ^^
+ Show Spoiler +

And they didnt prove that it was the best strategy, or calculate the probability exactly, which suggest that both are complicated. Which probably forgives my omission of those two...
Lemonwalrus
Profile Blog Joined August 2006
United States5465 Posts
Last Edited: 2008-01-30 21:10:16
January 30 2008 21:08 GMT
#30
On January 31 2008 06:06 Cascade wrote:
ok, I think I have found the solution.

+ Show Spoiler +

Prisoner number a starts a box number a. He will there find the name of prisoner number b, and he will next open box number b, and so on.

This will work as long as there are no "loops" longer than 50 among the boxes. A loop is when you follow boxes with the algorithm above, you will eventually come back to where you started. Which is your number! So as long as you finish one lap in your loop within 50 steps, you're fine. If the loop is longer than 50, everyone with numbers in that loop will fail, which is what correlates the success of the prissoners. and since this is everyone or noone thingy, you want as much correlation as possible.

Next thing would be to calculate the probability of not having loops longer than 50. But I wont do that. If this is correct solution, the calculation will be in walrus' spoiler. if im wrong, it wont make snece to calculate it. But 30% sounds plausible at first sight.

Then proving that there are no other better strategies... no idea right now. Maybe you can make som smart argument, or it will be very complicated.
I'll have a look at the spoiler anyway.

Don't readify my spoiler if you haven't solved it yet, gives away whether he was right or not.
+ Show Spoiler +
You are my hero.
fanatacist
Profile Blog Joined August 2007
10319 Posts
Last Edited: 2008-01-30 21:13:20
January 30 2008 21:11 GMT
#31
If you knew whether you were second or first, this would be my guess (expanded inductively, obv.):
+ Show Spoiler +
Assume 4 boxes. Prisoners 1, 2, 3, 4. Boxes a, b, c ,d.

1. 1 opens a+b.
Chance of living = 50%.

2. 2 opens a. Sees either the name of 1, of himself, or of another prisoner.
- The chance it is 1 is 50%, because he survived the first time by opening a+b. If it is 1, then 2 is in either b, c, or d. That's 33% chance of survival. Let's assume he would choose b.
- The chance it is 2 is 16.67% [P(a not=1)/3 = 50/3], because then 1 would be in b and the spots a c d are divided between 2 3 4. If it is 2, then there is a 100% chance survival.
- The chance it is 3 or 4 is 33.33%. This means 1 is in b, and 2 is in either c or d. That is 50% chance of survival. Let's assume he would choose c.
Chance of living = .5x.33 + .1667x1 + .33x.5 = 50%.

3. 3 opens a. Sees either the name of 1, of himself, or of another prisoner.
- The chance it is 1 is 50%, because he survived the first time by opening a+b. If it is 1, then 2 is in b, because he survived. That means 3 is in c or d. That's 50% chance of survival. Let's assume he would choose c.
- The chance it is 2 is 16.67%. Then 3 knows he is in c or d. That's 50% chance of survival. Let's assume he would choose c.
- The chance it is 3 or 4 is 33.33%. Then 3 knows he is in a or d, because 2 is in c, because he survived. If he is in a, he lives. If he is not in a, then he knows he is in d and lives. That is 100% chance of survival.
Chance of living = .5x.5 + .1667x.5 + .33x1 = 66.67%.

4.4 opens a. Sees either the name of 1, of himself, or of another prisoner.
- The chance it is 1 is 50%, because he survived the first time by opening a+b. If it is 1, then 2 is in b, and 3 is in c, because they survived. That means 4 is in d. That's 100% chance of survival.
- The chance it is 2 is 16.67%. Then 3 is in c. Then 4 is in d. 100% chance of survival.
- The chance it is 3 or 4 is 33.33%. This means 1 is in b, and 2 is in c, and 3 is in d. That is 100% chance of survival.
Chance of living = 100%

Total chance = .5 x .5 x .66 x 1



Something like that. I think the method, whatever it is, should have inductive reasoning in it.


EDIT: My idea was ok, but it gets raped by that. I lose. I should post my problem later.
Peace~
Slithe
Profile Blog Joined February 2007
United States985 Posts
Last Edited: 2008-01-30 21:41:40
January 30 2008 21:38 GMT
#32
Wow Cascade congratulations you got it. Very impressive.

Here's a link from wikipedia for the calculations
[link]

For those with clarification questions, you can think of the problem this way:

After the prisoners have discussed their strategy, they are taken to their own individual jail cells, far away from the other prisoners. Each prisoner goes into the room one at a time, looks at the boxes, and leaves without changing anything. Because they're in separate cells far away from eachother, they don't know who has gone when, and they cannot in anyway communicate with each other because they are isolated.

Also, all prisoners die simultaneously if any of the prisoners failed to find their name.
trickser
Profile Joined October 2006
Germany139 Posts
January 30 2008 22:02 GMT
#33
Sorry but the solutions presented in the spoilers make no sense to me.
+ Show Spoiler +

First of all i dont understand how they get the 30% probability. Second isn't there a high probability that a box has the name written on it in it? What is the prisoner supposed to do then?
Heart Catch Pre-Cure. Saa Mina De! Heart Catch Pre-Cure Hanasaka Seyo!
Aepplet
Profile Joined December 2003
Sweden2908 Posts
January 30 2008 22:49 GMT
#34
On January 31 2008 07:02 trickser wrote:
Sorry but the solutions presented in the spoilers make no sense to me.
+ Show Spoiler +

First of all i dont understand how they get the 30% probability. Second isn't there a high probability that a box has the name written on it in it? What is the prisoner supposed to do then?
+ Show Spoiler +
the probability of it having the name in it is 1%. If he finds it on the first try, that's just good =)
Cascade
Profile Blog Joined March 2006
Australia5405 Posts
January 30 2008 23:06 GMT
#35
On January 31 2008 07:49 Aepplet wrote:
Show nested quote +
On January 31 2008 07:02 trickser wrote:
Sorry but the solutions presented in the spoilers make no sense to me.
+ Show Spoiler +

First of all i dont understand how they get the 30% probability. Second isn't there a high probability that a box has the name written on it in it? What is the prisoner supposed to do then?
+ Show Spoiler +
the probability of it having the name in it is 1%. If he finds it on the first try, that's just good =)


+ Show Spoiler +
And he wont find it on the second try or later, because to get there he would have to pick the name which is in the box he should go to. If name number 5 is in box number 5, then he cant get to box number 5 by picking name 5 from another box.

The 30% isnt obvious at all. :/ If you understand the picture with the loops it will seem a bit more reasonable that it is in the range 10%-50% instead of 10^-9 though.
Jonoman92
Profile Blog Joined September 2006
United States9110 Posts
January 30 2008 23:20 GMT
#36
ah that's cool, I hope those prisoners got a good memory to memorize which of the 100 boxes belongs to each guy though.
Slithe
Profile Blog Joined February 2007
United States985 Posts
January 30 2008 23:34 GMT
#37
On January 31 2008 08:20 Jonoman92 wrote:
ah that's cool, I hope those prisoners got a good memory to memorize which of the 100 boxes belongs to each guy though.


Don't worry, the prisoners in these problems are typically very intelligent. In fact, I'm pretty sure they're in prison for being too smart for their own good.
EmeraldSparks
Profile Blog Joined January 2008
United States1451 Posts
Last Edited: 2008-01-30 23:37:11
January 30 2008 23:36 GMT
#38
So, I'm pretty stupid as I'm definitely wrong, but I'm just going to make things easier on the prisoners and the answer still comes out less than 25% for me.

+ Show Spoiler +
If all of the prisoners succeed in finding their own name, then they all get to live. However, if even one person fails to find their name, it's the death penalty for all of them.

Actually, we're going to designate two prisoners. If they both find their name, then everybody lives. Otherwise, everybody dies; if there were a 30% chance of everybody finding in any one situation each of those 30%s would satisfy these requirements too, so the chance of living is greater than or equal to 30%. We might as well remove all the boxes with the names of other prisoners, and all the other prisoners don't need to open any boxes since it has no effect in this scenario as opening a name of another prisoner has no effect and with zero information transfer somebody whose choices do not directly affect the chance of instadeath has no effect on whether the two designated prisoners find their name. As the boxes are effectively indistinguishable from the start, there is only one variable involved, how many chosen boxes overlap for designated prisoner A and designated prisoner B. We may as well rearrange the boxes in our minds so that there are first X blocks that A chooses only, 50-X that A and B choose, X that B chooses only, and 50-X that neither choose. This number X is a constant for any given runthrough. Now, lets examine where A's name is - there is a 50% chance of it lying in the first 50, and in fact, a X% chance of lying in a box B is not choosing and a 50-X% of lying in a box A and B are choosing. Examine, now, where B's name is - in the X% chance case that A's name is not in the fifty boxes that B is choosing, there are 50 boxes that say life with 99 to choose from - a (X/100)(50/99) chance that A's name is in a box he chooses but not in a box that B has chosen and that B's name is in a box that B has chosen. Consider the other case that spells life - that A's name is in the 50-X that B has also chosen. Then there are 49 unoccupied boxes that spell life for B, and 99 that spell death, giving us a ((50-X)/100)(49/99) chance that A's name is in a box that both A and B have chosen and that B's name is in a box he has chosen. This covers all cases where the names of A and B are in the fifty they have chosen, giving us a total of (X/100)(50/99) + ((50-X)/100)(49/99) chance that they have survived, or (50X + (50*49 - 49X)/(99*100) = (X + 49*50)/(99*100). But since X must be between zero and fifty, (X + 49*50)/(99*100) <= (50 + 49*50)/(99*100) = 50/198 ~= 25.2525% < 30%.


I know I must have made a small assumption somewhere in there but I can't seem to find it at all, probably in all my indistinguishability and throwing-stuff-away arguments.
But why?
Night[Mare
Profile Blog Joined December 2004
Mexico4793 Posts
January 31 2008 00:03 GMT
#39
hahaha nerdy thread.
I wish i had a clue how to solve something like this. Even though im studing some ingenieering, im clueless in this problems. GJ for those who've solved it already
Teamliquidian townie
Tynuji
Profile Blog Joined June 2007
127 Posts
February 06 2008 04:14 GMT
#40
100%, the prisoners all have the same name.
Prev 1 2 3 Next All
Please log in or register to reply.
Live Events Refresh
Next event in 15h 35m
[ Submit Event ]
Live Streams
Refresh
StarCraft 2
Nina 9
StarCraft: Brood War
Shuttle 488
ProTech153
White-Ra 150
sSak 50
910 28
Sexy 24
Dota 2
Gorgc15452
syndereN17
League of Legends
JimRising 240
Counter-Strike
edward32
kRYSTAL_3
Heroes of the Storm
Liquid`Hasu573
XaKoH 282
Trikslyr72
Other Games
Beastyqt988
FrodaN855
B2W.Neo823
ceh9627
mouzStarbuck338
ToD182
KnowMe84
QueenE48
Dewaltoss45
fpsfer 0
[ Show 18 non-featured ]
StarCraft 2
• Hupsaiya 40
• Adnapsc2 19
• Reevou 5
• AfreecaTV YouTube
• intothetv
• Kozan
• IndyKCrew
• Migwel
StarCraft: Brood War
• 80smullet 20
• FirePhoenix13
• ChaeLyn 7
• BSLYoutube
• STPLYoutube
• ZZZeroYoutube
League of Legends
• Nemesis3189
• Jankos3062
Other Games
• imaqtpie634
• Shiphtur269
Upcoming Events
GSL
15h 35m
Epic.LAN
16h 35m
Big Brain Bouts
20h 35m
OSC
1d 1h
Replay Cast
1d 2h
Sparkling Tuna Cup
1d 13h
WardiTV Invitational
1d 15h
Epic.LAN
1d 16h
IPSL
1d 19h
eOnzErG vs McRibbed
LancerX vs StRyKeR
AI Arena Tournament
1d 21h
[ Show More ]
OSC
2 days
Yamato Cup
2 days
IPSL
2 days
Artosis vs RiceHat
Kyrie vs Sony
Map Test Tournament
3 days
WardiTV Weekly
3 days
Ultimate Battle
3 days
Rush vs BeSt
Replay Cast
4 days
Rogue vs Classic
Patches Events
4 days
PiGosaur Cup
5 days
Kung Fu Cup
5 days
The PondCast
6 days
KCM Race Survival
6 days
Liquipedia Results

Completed

Proleague 2026-10-28
Blizzard Classic Cup 2026
HCC Season 3

Ongoing

Acropolis #5
KCM Race Survival 2026 Season 4
International ShieldBattery League
BSL Season 23
PGL Masters Bucharest 2026
Stake Ranked Episode 5
1win Private Club #2
Thunderpick World Champ. '26
ESL Pro League Season 24
Stake Ranked Episode 4
1win Private Club #1
Logitech G Play Connect 2026
SL StarSeries Fall 2026
FISSURE Playground #3
BLAST Open Fall 2026
Esports World Cup 2026

Upcoming

IPSL Winter 2026-27
SC4ALL II: Brood War
BSL 23: Non-Korean Championship
HSC XXX
Stellar Fest 2: Lunar Cup
SC4ALL II: StarCraft II
Kung Fu Cup 2026 Grand Finals
RSL Offline Finals
eXTREMESLAND 2026
PGL Major Singapore 2026
Stake Ranked Episode 6
BLAST Rivals Fall 2026
IEM Beijing 2026
TLPD

1. ByuN
2. TY
3. Dark
4. Solar
5. Stats
6. Nerchio
7. sOs
8. soO
9. INnoVation
10. Elazer
1. Rain
2. Flash
3. EffOrt
4. Last
5. Bisu
6. Soulkey
7. Mini
8. Sharp
Sidebar Settings...

Advertising | Privacy Policy | Terms Of Use | Contact Us

Original banner artwork: Jim Warren
The contents of this webpage are copyright © 2026 TLnet. All Rights Reserved.