A national math contest consisted of 1 1 multiple choice questions, each having 1 1 possible choices, of which only 1 of the choices is correct. A student's score for the contest is the total number of correct answers.
Suppose that 1 1 1 students actually wrote the exam, answered every single question, and no two students have more than one answer in common. The highest possible average score for the students can be expressed as b a where a , b are coprime positive integers. What is the value of a + b ?
Details and assumptions
Each correct answer is worth 1 points.
This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try
refreshing the page, (b) enabling javascript if it is disabled on your browser and,
finally, (c)
loading the
non-javascript version of this page
. We're sorry about the hassle.
Each question can be answered correctly on at most 1 1 papers. This is because any two students who answer it correctly cannot have the same answers for any other given question, and that other question can have at most 1 1 different answers. So the maximum total number of correct answers on all the papers is 1 1 ⋅ 1 1 , which would give an average score of 1 1 ⋅ 1 1 / 1 1 1 = 1 2 1 / 1 1 1 .
To show that this maximum average is attained, we exhibit a collection of papers that satisfies the condition. Label each student with an ordered pair ( a , b ) in F 1 1 2 , where F 1 1 is the field with 1 1 elements (the integers mod 1 1 ). Label the questions 0 , 1 , … , 1 0 and the answers 0 , 1 , … , 1 0 . Now we mandate that student ( a , b ) answers question x with the answer a x + b ∈ F 1 1 . Note that if a 1 x + b 1 = a 2 x + b 2 , then x is determined uniquely as x = ( b 2 − b 1 ) / ( a 1 − a 2 ) , and this expression makes sense unless a 1 = a 2 , but in that case b 1 = b 2 and they are the same student. So the point is that each distinct pair of students have at most one answer in common.
Let's further stipulate that the 1 1 1 students are made up of the 1 1 0 ordered pairs ( a , b ) with a = 0 , plus the ordered pair ( 0 , 0 ) . And let's also stipulate that the correct answer to every question is 0 . Then the first 1 1 0 students all have exactly one correct answer: student ( a , b ) gets question − b / a right. And student ( 0 , 0 ) gets a perfect score. That adds up to 1 1 0 + 1 1 = 1 2 1 correct answers, which we already saw is the maximum.
So the answer is 1 2 1 + 1 1 1 = 2 3 2 .
The question does not state that the participants have to answer all the question. So they could just leave some blank. Then everyone could only answer the first question, get it right and leave the rest blanks. This would satisfy the condition. So that more than 11 people can get a question right. I believe that if they had to answer all the question then there could only be 11 people who get any one right. Is there something from the question that I am missing that suggests that all the question must be answered? Or am I just failing to figure it out properly?
Log in to reply
If participants could leave answers blank, and that doesn't count as "an answer in common", then something like 166 is achievable. I don't think that was intended to be an option, however.
Thanks for pointing out the ambiguity. I edited the question to rule this out.
When I first saw this question, there was no hint on how to begin. So I decided to just analyse what I was given and try to formulate some observations or lemmas as I will call them. A quick skim through the question show that:
no two students have more than one answer in common
seems quite fishy, in a sense that this suggests 2 approaches: pigeonhole and double counting. It turns out that the former works. Let us rephrase this into a lemma.
Lemma 1 : For each option of each question, there are at most 1 1 people choosing that option.
Proof: Consider to the contrary. Suppose for a certain question, ≥ 1 2 people chose the same option. By the pigeonhole principle, we have ≥ 2 people who will have chosen the same answer to another question since there are only 1 1 possible choices for the questions. This clearly violates the condition that no two students have more than one answer in common. □
Now having sorted out the phrasing of the problem, I turned to the numerical values provided. I asked myself, what is the correlation between 1 1 1 and 1 1 ? (This is a common tactic in problem solving, normally the values are chosen specially for a specific function, understanding why the variables are chosen helps us to select a proper approach.) Tinkering a bit, it dawned upon me that 1 1 1 > 1 1 0 = 1 0 × 1 1 whence the implication of pigeonhole. This provides some insight as to how many students chose a particular option. Another motivation for the following lemma is by drawing out an "adjacency matrix" (at least that's what my trainers refer to), as in a matrix where a i j = 0 , 1 . We would love to count the number of 1 s hence the inspiration.
Lemma 2 : For a particular question, the maximum number of students that choose a particular option is 1 1 .
Proof: We start off with the observation that 1 1 × 1 0 < 1 1 1 , so for each question. Once this is established, pigeonhole gives us that for any question, at least one option must have been chosen by ≥ 1 1 students. By Lemma 1 , however, each option of each question is chosen by at most 1 1 students. These two inequalities combined gives that there has to be an option that is chosen by exactly 1 1 students, whence the statement in the lemma. □
Notice the usefulness in proving n ≥ a and then n ≤ a to establish n = a . This is a common technique when no obvious bijection or correspondence is available.
Let us remark that the average mark of the students is in fact:
1 1 1 total marks of all of students ( ∗ )
Let us treat the numerator as a whole and see the big picture . Instead of treating individual students, let us instead look at individual questions (students' answers are sort of like variables - we try to keep things "constant"). Since the questions themselves are independent by definition, we can maximise the total score of the students by making as many students getting the question correct as possible. In other words, let the option that is chosen the most be the correct answer.
Well, by Lemma 2 and the above observation, exactly 1 1 students answered each of the 1 1 question correctly. Which means that the highest possible average mark (by subbing into ( ∗ ) ) is b a = 1 1 1 1 1 × 1 1 = 1 1 1 1 2 1 ⇔ a + b = 1 2 1 + 1 1 1 = 2 3 2 .
Nice. The one thing I would say is that your answer doesn't show that the situation stated in the problem is actually possible, i.e. that one can construct 1 1 1 tests that satisfy the conditions of the problem (the construction in my solution does this). But if you take that on faith as the problem asks you to do, then you can indeed see that the maximum is attained via these elementary Pigeonhole Principle arguments.
Log in to reply
You can't just 'take that on faith'. It is possible that the 'maximal' value cannot be constructed, and in which case the answer would be something else. Of course, having gotten the numerical answer correct, you could be more certain (but that doesn't constitute a proof) that such a construction exists.
This is why for such questions, it is important to show the maximum (minimum) bound, and also that it can be achieved.
Log in to reply
No, no, that's not what I mean. His answer does in fact show that the maximum is achievable, assuming the conditions of the problem.
What I mean is, the problem asks you to assume ("suppose") something that it's not immediately clear is possible--that 111 students took the exam, answered all the questions, and no two students agreed on more than one answer. Under that assumption, you don't need to work that hard to show that the maximum is achievable; as he notes, it falls out of the Pigeonhole Principle, because each question has to have an answer that at least 11 students gave. The heavy lifting has already been done for you by the assumption.
When I solved the problem, I didn't realize that, and so I came up with a construction that explicitly satisfied the conditions of the problem. But this was unnecessary if you assume the thing that the problem asks you to assume.
If no two students can have more than one answer in common, then the best case scenario would be when they all have 1 correct answer in common. This means the total they would score is: 1 1 1 × 1 = 1 1 1 But one of the students could have full marks making the maximum total of the marks 1 1 1 + 1 0 = 1 2 1 (as that student would score 10 more than everyone else). The average score for the students would then be 1 1 1 1 2 1 and so a + b = 1 1 1 + 1 2 1 = 2 3 2
Why would that be the best case scenario? Couldn't 55 students each have 2 correct answers, from each combination of 2 problems correct, and the other 56 students with any 1 question correct? That gives an average of 166/111
We prove that an optimal solution exists in which one of the students get a perfect score of 11, and all other students get a score of 1. The average mark in this case is ( 1 1 1 + 1 0 ) / 1 1 1 = 2 3 2 / 1 1 1 .
First we show that there is no solution with a higher average mark. Suppose we know the answers of all students but the last in an optimal solution, and we want to assign correct answers, as well as the last student's answers, such that the average mark is maximized. In other words, we want to maximize the total number of student-problem pairs s , p such that s correctly answers p . This quantity can be split into the sum of 11 separate numbers, one for each problem p . So the best we can do is, for each problem p , pick the most common answer among the first 110 students (if there are two answers appearing 55 times each, either will do). This fixes the number of correct answers for problem p involving the first 110 students. To maximize the number of correct answers including the last student, we can make him get a perfect score. Therefore there is an optimal solution in which a student gets a perfect score. Then the score of all other students is at most 1 since they only agree with him on at most one problem.
Next we show that this average mark can indeed be achieved by constructing an explicit solution. For this we use arithmetic over the finite field F 1 1 , i.e., arithmetic modulo the prime 11. The idea is that two different degree-1 polynomials over any field can only agree at one point, just like the students' answers to the problems. (In fact, the same idea is the basis behind the Reed-Solomon error-correcting code that is used, e.g., in CDs.)
Let A = { ( a , b ) ∈ F 1 1 × F 1 1 ∣ ( a = 0 ) ∨ ( a = 0 ∧ b = 0 ) } .
The size of A is ∣ A ∣ = 1 1 ⋅ 1 1 − 1 0 = 1 1 1 . We identify each element of A with one of the students. Also we number problems and answers from 0 to 10 inclusive (again, the elements of F 1 1 ). The answer that student ( a , b ) gives to problem x is given in our construction by ( a ⋅ x + b ) m o d 1 1 . The correct answer for every problem is answer number 0.
So far we have 111 students, 11 problems, 11 answers per problem and each student marks an answer for each problem. Note that student ( 0 , 0 ) gets a perfect score. Also, for any a 2 = 0 there is only one solution x = − b / a 2 ( m o d 1 1 ) to a 2 ⋅ x + b = 0 , so all other students get a score of 1.
It remains to be shown that no two distinct students ( a 1 , b 1 ) and ( a 2 , b 2 ) give the same answer for more than one problem. If a 1 or a 2 are zero, say a 1 = 0 (and therefore b 1 = 0 ), then this holds because as we saw, all other students only give the zero answer to one problem. When a 1 and a 2 are non-zero, the claim follows because $$a 1 \cdot x + b 1 \equiv a 2 \cdot x + b 2 \pmod {11}$$ implies $$(a 1 - a 2) x = b 2 - b 1.$$ If we also had ( a 1 − a 2 ) y = b 2 − b 1 for some y = x , then we would get ( a 1 − a 2 ) ( x − y ) = 0 . Dividing (modulo 11) by the non-zero element x − y , we conclude that a 1 = a 2 , so 0 = b 2 − b 1 and b 1 = b 2 as well. This means thay are the same student, as we wished to show.
The assignment of the last student to getting all the questions correct can make him violate the "no two students have the same answer for two questions" condition, so there is no guarantee that there exists an optimal solution that has one student getting all the answers correct. The rest of the proof is thus incorrect.
if 1 student achieves full marks (11) then the maximum the other students can achieve is 1 mark: 1x11 + 110x1=121 average: 121/111 and 121+111=232 Alternatively All the students can have one question correct ( the first one for example) (1x111) and ten students can have another one question correct (1x10).
Problem Loading...
Note Loading...
Set Loading...
For any question, if there are 12 or more people choosing the same option, then by the pigeonhole principle, at least 2 of these people must have the same answer to another question as well (since each question only has 11 options). These 2 people will thus have at least 2 answers in common, contradicting the condition in the question. Hence, there are at most 11 people choosing the same option for each question.
In addition, for any question, there are 11 options chosen by 111 students. By the pigeonhole principle, at least one option must be chosen by at least 11 students (since 1 1 × 1 0 = 1 1 0 < 1 1 1 ). We have established above that each option is chosen by a maximum of 11 students, so for every question, there must be at least one option that is chosen by exactly 11 students, with no other options being chosen by greater than 11 students. Thus, the option that is chosen the most number of times will have been chosen by exactly 11 students.
The total scores of the students for each of the 11 questions will sum up to the total score of the students for the entire paper. The answer to each of the 11 questions will only affect the total score of the students for that question and no other questions (i.e. they are independent), so this means that we can consider the questions separately and try to maximise the total score of the students for each of the questions. To do this, we just let the option that is chosen the most number of times for each question be the right answer for that question. This will mean that exactly 11 students answered each question correctly, so the maximum total score will be 1 1 × 1 1 = 1 2 1 .
Therefore, the highest possible average mark for the students is 1 1 1 1 2 1 , and the answer we want is 1 2 1 + 1 1 1 = 2 3 2 .