12
$\begingroup$

A team of $11$ cyclists begin a race. If each cyclist rides individually then their respective race times will be $10, 11, \ldots, 20$ minutes. However, if three cyclists with individual times $(a, b, c)$ ride as a triplet, they will finish the race in $(a+b+c)/3$ minutes - due to a clever energy preserving rotational strategy. What is the least amount of time needed for every cyclist to finish the race?

$\endgroup$
1
  • 1
    $\begingroup$ If teams can change during the race, mean of 10-20 is 15, suggesting 15 minutes is the shortest possible time (it might not be achievable). You can easily do better than the best fixed teams. $\endgroup$ Commented Jan 25, 2023 at 14:45

4 Answers 4

17
$\begingroup$

Via mixed integer linear programming with a binary decision variable for each of the $\binom{11}{1}+\binom{11}{3}=176$ possible teams, I have found that the minimum is

15 minutes and 20 seconds,

attainable via teams with times

 {10,16,19}, average 15
 {11,17,18}, average 15 + 1/3
 {12,14,20}, average 15 + 1/3
 {13}, average 13
 {15}, average 15

Here's an alternative optimal solution, with only one "slow" team:

 {10,17,18}, average 15
 {11,16,19}, average 15 + 1/3
 {12,13,20}, average 15
 {14}, average 14
 {15}, average 15


Here's the formulation I used. For each team $T$ (subset of cardinality $1$ or $3$), let $a_T$ be the average time and let binary decision variable $x_T$ indicate whether that team is used. Let decision variable $z$ represent $\max_T a_T x_T$. The problem is to minimize $z$ subject to linear constraints \begin{align} \sum_{T: c \in T} x_T &= 1 &&\text{for all cyclists $c$} \tag1\label1\\ a_T x_T &\le z &&\text{for all teams $T$} \tag2\label2 \end{align} Constraint \eqref{1} assigns each cyclist to exactly one team. Constraint \eqref{2} enforces the minimax objective.

$\endgroup$
11
  • 1
    $\begingroup$ This is correct! Actually there are many equivalent solutions. $\endgroup$ Commented Jan 25, 2023 at 3:04
  • 2
    $\begingroup$ Yes, 13 optimal solutions. $\endgroup$
    – RobPratt
    Commented Jan 25, 2023 at 3:07
  • $\begingroup$ What is the (11 1) for in the total count of combinations? $\endgroup$
    – justhalf
    Commented Jan 25, 2023 at 5:11
  • 2
    $\begingroup$ I mean, for the purpose of counting different combinations, it's double counting some, right? (it doesn't affect the optimization result of course) $\endgroup$
    – justhalf
    Commented Jan 26, 2023 at 7:37
  • 2
    $\begingroup$ @JacobRaihle The number of solutions to constraint (1) is $$1+\binom{11}{3}+\binom{11}{3}\binom{8}{3}/2!+\binom{11}{3}\binom{8}{3}\binom{5}{3}/3! = 20186,$$ but the MILP solver does not explicitly generate them all. $\endgroup$
    – RobPratt
    Commented Jan 26, 2023 at 14:55
27
$\begingroup$

Computer brute forcing seems somewhat excessive for puzzles like this ("axe to the head" is how my native language might describe it), so here's a (post-accept) solution to balance things out.

By the puzzle's definition, the completion time for a triplet is the average of the three individual times. This means there's a quantity that will always be preserved, regardless of grouping:

The average individual completion time of the entire bunch never changes.

That quantity is easily calculated to be 15 minutes, and it is of course a hard limit for the overall completion time, too.

Now, if a grouping exists so that the 15 min overall time is achievable, then every single group (of 1 or 3 persons) will have to complete the race at exactly 15 minutes: If any group were faster, then there would have to be another group running slower than 15 minutes in order to restore the global average of individual times.

However, there are 11 cyclists, so all the groups cannot run at exactly 15 minutes: there are at least two individual riders, and only one of them can be the guy that completes the race in exactly 15 minutes. The other individual guy must be either slower (dragging the total down by at least a minute) or faster (causing another group to be slower by at least 1/3 of a minute.)

To see if the minimal slowdown is achievable, it's easiest to put the 14 and 15 minute guys as the two individual riders, and then form groups totalling 45, 45, and 46 minutes from the rest of the guys. Turns out this is easy to do, for example like this:

 14 (14 minutes)
 15 (15 minutes)
 12, 13, and 20 (-> 45/3 = 15 minutes)
 11, 16, and 18 (-> 45/3 = 15 minutes)

..and we don't even have to check the final group of the three other guys to know they must finish in exactly

15 min 20s,

or else the overall average of individual completion times would have changed, which is impossible.

$\endgroup$
0
$\begingroup$

Assuming that

combinations must be exactly three cyclists, and must stay together for the entire race

then the best we can do is

17 minutes, because once we've sped up the ones who take 20 and 19 and 18 minutes, we can't form any more triplets.

Let's demonstrate that this is achievable:

(10, 11, 20) ride together and all finish in 41/3 = 13.6+ minutes.

(12, 13, 19) ride together and all finish in 44/3 = 14.6+ minutes.

(14, 15, 18) ride together and all finish in 47/3 = 15.6+ minutes.

16 and 17 ride individually.

$\endgroup$
4
  • 1
    $\begingroup$ This is a good start, but you can do better. $\endgroup$ Commented Jan 25, 2023 at 2:33
  • 2
    $\begingroup$ "because once we've sped up the ones who take 20 and 19 and 18 minutes, we can't form any more triplets." why can't you make 17 go to form another triple? how about 16? $\endgroup$
    – justhalf
    Commented Jan 25, 2023 at 5:12
  • 8
    $\begingroup$ @justhalf Because 11 people can form at most 3 teams of triplets. The error lies in assuming that 17 can't be in the same triplet as one of 18,19,20 and still finish in less than 17 minutes. $\endgroup$ Commented Jan 25, 2023 at 8:21
  • $\begingroup$ The problem is that you are speeding the 20, 19 and 18 by too much, ending up with 16 and 17 riding individually. You should be able to redistribute the groups such that it are faster cyclists that ride individually. $\endgroup$ Commented Jan 26, 2023 at 16:12
-2
$\begingroup$

The group will be fastest when they ride

in one group of 11. With most of the time or all the time the 3 fastest in front.

The worst time will be at most

13 minutes 40 seconds

Proof

We know that

there is a clever energy preserving rotational strategy that works for 3 cyclists with time a<b<c such that the slowest time c is reduced to (a+b+c)/3

and common knowledge dictates that

larger groups should be able to preserve the same or even more energy as a result of energy preserving rotational strategies.

Therefore

It must work at least as good for 11 cyclists.

We might go for the time by the top three cyclists

They could finish alone in 11 minutes.

But, we can not know

at which pace the slowest cyclists will be able to follow the pack.

Yet, we do know that they can finish in

at least in 13 minutes and 40 seconds, because the slowest cyclist is able to follow at that speed in a group of three, so certainly as well in a bigger group.

$\endgroup$
9
  • 3
    $\begingroup$ I was tempted to write something like this too, but puzzling.SE is about the puzzle, not about modelling reality. The boost from riding in a large bunch is unbeliveably astounding, but the puzzle presented here doesn't go that far. Please don't be put off by the downvotes here - for real riding there's bicycles.stackexchange.com $\endgroup$
    – Criggie
    Commented Jan 27, 2023 at 8:54
  • 1
    $\begingroup$ @Criggie it is always nice to see different interpretations and the puzzle presented is not clear about that limitation (is it a riddle or a straightforward math puzzle?). The puzzle provides information about an energy saving trick and does not say that this works strictly only for a group of 3. It is not too weird to interpret the question as allowing for bigger groups. With this answer I was very minimal in making further assumptions, it is not like I am extrapolating for further speed improvements when they drive in larger groups (they might even go faster than 10 minutes) $\endgroup$ Commented Jan 27, 2023 at 9:45
  • 1
    $\begingroup$ A better way to pose the question, to avoid this ambiguity, might be, "find a combination to divide the group of 11 cyclist into groups of size 3 and/or 1 that minimises the amount of time needed to have every cyclist finish the race." $\endgroup$ Commented Jan 27, 2023 at 10:26
  • 3
    $\begingroup$ Even for lateral-thinking puzzles—which this is NOT—the solver is not given free license to invent their own rules or scenarios. Especially for puzzles not tagged "lateral thinking", the right answer to a puzzle will be the one that uses what the puzzle gave you or hinted at, without inventing facts, rules, or interpretations out of thin air to make a "solution" work. Puzzle posters can't close every loophole… and shouldn't have to. $\endgroup$
    – Rubio
    Commented Jan 27, 2023 at 11:33
  • 1
    $\begingroup$ @Rubio I get that, but there's also a gray area. Not closing all loopholes is different from leaving all open. I don't consider this so much as lateral-thinking. I am using the information and am not 'making up information'. Assuming that the cyclists can ride as a single group of 11 is not more assumption than that they can ride as independent groups of 3. (If this would be track cycling, then the different groups will be interfering with each other in the bends) Anyway, I won't mind if my answer is classified as 'clever, but not what the question intended', but more than that is exaggerated. $\endgroup$ Commented Jan 27, 2023 at 11:47

Not the answer you're looking for? Browse other questions tagged or ask your own question.