1

I am having hard time with this question .I have not understood what is point and why is sequence $a_1 + 21$ , $a_2 + 21 $... has been taken in second picture .Please help me understand the question

Thanks

enter image description here

enter image description here

Taylor Ted
  • 3,508

3 Answers3

4

This is an application of the Pigeonhole Principle. I'll try to explain the given answer in another way. Hopefully it will help your understanding of the solution.

We want to show that there is a consecutive sequence of days (we don't really know how many days) during which exactly 21 games are played. The strategy is to find two days ($1\leq j < i \leq 77$) such that $a_i = a_j + 21$; i.e. the total number of games played up to day $i$ is exactly 21 more than the total number of games played up to day $j$.

The first series of inequalities follows from the fact that at least 1 game is played per day, and the total number of games is limited by the number of weeks (11) and the maximum number of games per week (12). Thus, we can write:

$$1 \leq a_1 < a_2 < \cdots < a_{77} \leq 132$$

Now, if I interpret your question correctly, you are wondering why we have introduced the $a_i + 21$ terms. The reason for doing so is that we can apply the Pigeonhole Principle. Also, from the restatement of the question above, you can see that what we are after are two days $i$ and $j$ such that the total number of games played up to those points differ by exactly 21 (i.e. $a_j + 21 = a_i$). It's fairly easy to see (simply by adding 21 to the sequence of inequalities above), that:

$$22 \leq a_1 + 21 < a_2 + 21 < \cdots < a_{77} + 21 \leq 153$$

So, we know that all of the numbers

$$a_1, a_2, \cdots, a_{77}, a_1 + 21, a_2 + 21, \cdots , a_{77} + 21$$

are in between 1 and 153. We know that there are 154 items in the list, and only 153 possible values they can take, so we know that there must be two which are equal, by the pigeonhole principle.

Also, since the terms in each sequence are strictly increasing, we know that:

$$a_i \neq a_j \qquad \text{whenever $i \neq j$}$$

and also

$$a_i + 21 \neq a_j + 21 \qquad \text{whenever $i \neq j$}$$

Hence, the two which are equal must be from 'different' lists. That is, one is from the list $a_1, \cdots , a_{77}$, and the other is from the list $a_{1} + 21, \cdots, a_{77} + 21$.

Hence there exists a $j < i$ such that $a_j + 21 = a_i$, meaning that exactly 21 games were played during the days $j+1, j+2, \cdots, i - 1, i$.

MacCab
  • 199
  • I have still not understood why you have added 21 to previous sequence of numbers ? – Taylor Ted Jul 08 '15 at 15:01
  • The appearance of the number 21 in the solution comes from the question itself. As Ross Millikan pointed out below, "If we were asked to prove there was a series of days on which he played 19 games, we would have used $a_1+19,a_2+19… $ and the same argument would have gone through." – MacCab Jul 08 '15 at 18:26
3

Alternative solution (it proves a stronger result)

Here is a general theorem: Given a list of $n$ integers $a_1,a_2\dots a_n$ there is a list of consecutive integers that have a sum that is a multiple of $n$.

Proof:

Consider

$a_1$

$a_1+a_2$

$a_1+a_2+a_3$

$\dots$

$a_1+a_2+\dots+a_n$

If any of these sums is a multiple of $n$ we are done. Otherwise, each of them is congruent to one of the $n-1$ congruence classes $\bmod n$ that is not zero. So by the pigeonhole principle there is $j<k$ so that $a_1+a_2\dots a_j\equiv a_1+a_2+\dots a_k\bmod n$. Therefore $a_{j+1}+a_{j+2}+\dots a_k$ is a multiple of of $n$. This ends the proof.

We can use this theorem to solve your problem.

Consider the $21$ days $8,9,10,11\dots 28$, there is a sequence of consecutive days in which a multiple of $21$ games was played. This can be $21$ or $42$. If it is $42$ then that is because these $21$ days contain four weeks. two full weeks and two other weeks of which we have $7$ days in total. At least $42-24=18$ games were played in these $7$ days. Meaning, at least $7$ games were played in the other $7$ days of these two weeks. So at least $25$ games were played in these two weeks, this means at least $13$ games were played in one of these two weeks, a contradiction. So it is not $42$ and must therefore be $21$, so indeed there is a period of days in which $21$ games were played and we only needed days $1,2,3\dots 35$ to prove this.

Asinomás
  • 107,565
  • Don't know about congruent class modulo concept – Taylor Ted Jul 08 '15 at 14:49
  • You should read up on it, it is very intuitive and nice. Basically its just the remainder a number leaves when dividing by $n$. – Asinomás Jul 08 '15 at 14:53
  • can you explain this line " Otherwise each of them is congruent to one of the n−1 congruence classes modn that are not zero." – Taylor Ted Jul 08 '15 at 14:57
  • well, if none of them is a multiple of $n$ it is because each of them leaves a remainder different from zero, this remainder can be any number between $1$ and $n-1$, so there are $n-1$ possibilities. – Asinomás Jul 08 '15 at 15:02
  • Even stronger result: The first 3 consecutive calendar weeks (containing 21 days), ending before day 28, have a max of 36 games and therefore must have a run of days with 21 games, since 0 and 42 are impossible. The well known "standard" solution looks a bit silly, using a more complicated proof to prove a far weaker result. In fact, the standard solution assumes that 11 weeks have 132 games, not 144 games, which means it is assuming that day 1 is beginning of a week. Under this condition, the first 21 days are guaranteed to have a run of days with 21 games! – misterbee Jun 14 '24 at 02:27
  • You can also observe that a span of 21 days is 4 weeks minus 7 days, and so has a max of 4x12 - 7=41 games, which is a tidier way to show that the only multiple of 21 in the valid range is 21 itself. – misterbee Jun 14 '24 at 02:35
2

The sequence was chosen that way because we were asked to prove there was a series of days on which he played $21$ games. If we were asked to prove there was a series of days on which he played $19$ games, we would have used $a_1+19, a_2+19 \dots $ and the same argument would have gone through.

Ross Millikan
  • 383,099