Ivan Shishkin, Rye (1878)

Problems/Probability and statisticsReviewed

The agony of group gifts

by Catalpa·translated by Ancient Tree·
20
Difficulty scaleÉchelle de difficulté

This score reflects both the level of the required concepts and the difficulty of the solution.Ce score tient compte à la fois du niveau des notions nécessaires et de la difficulté de la résolution.

  1. 110First steps / middle schoolPremiers pas / collège
  2. 1125Beginner / high schoolDébutant / lycée
  3. 2650Intermediate / undergraduateIntermédiaire / licence
  4. 5170Advanced / graduateAvancé / master
  5. 7190Expert / specializedExpert / spécialisé
  6. 91100Research levelNiveau recherche
These levels are approximate guides.Ces niveaux sont des repères approximatifs.
·
English
EnglishFrançais

You organized a group gift for a friend’s birthday, and you advanced the money on behalf of the nn other participants in the gift. It is now time to get reimbursed by each of them.

To do this, you must call the nn other contributors one by one on the phone - but your friends are a bit stingy, and try to avoid your call, so that you only have a probability p]0,1[p \in \,]0,1[ of reaching the person you’re calling...

We assume that the calls are independent trials (the participants don’t warn each other), and we let XX denote the random variable representing the number of gift participants you manage to reach.

  1. What is the distribution of XX?

The next day, you call back, under the same conditions, the nXn - X most stingy participants who did not answer the day before. Let YY denote the random variable representing the number of people you manage to reach during this second attempt.

  1. For i[ ⁣[0,n] ⁣]i \in [\![0, n]\!], determine P(Y=kX=i)P(Y = k \mid X = i) for kNk \in \mathbb{N}.

  2. Prove that Z=X+YZ = X + Y follows a binomial distribution, and determine its parameter.

  3. Determine the expectation and the variance of ZZ.

I solved itMark it doneAdd to my listKeep it in your list

Hints

1

Hint 1

Open this only if you want a small nudge before looking at the solutions.

Solutions

2
Reveal solutionsAre you sure? Give it a try first.

Solution by Ancient Tree

Discussions0 useful votes
  1. Because there are nn independant calls with probability pp of success, XX follows a binomial distribution B(n,p)\mathcal{B}(n,p):
    P(X=i)=(ni)pi(1p)niP(X=i)=\binom{n}{i}p^{i}(1-p)^{n-i}

  2. In the second round, the calls are still independant and each with probability pp to succeed, but this time, if X=iX=i, there are only nin-i calls. So the conditional probability is again a binomial distribution B(ni,p)\mathcal{B}(n-i,p):
    P(Y=kX=i)=(nik)pk(1p)nikP(Y=k\mid X=i)=\binom{n-i}{k}p^{k}(1-p)^{n-i-k}

  3. Disclaimer : there is probably a better way than the brute force method I did, but I can’t find the right interpretation: when X=iX=i, then we have to have Y=kiY=k-i. This means that :
    P(X+Y=k)=i=0kP(X=iandY=ki)P(X+Y=k)=\sum_{i=0}^{k}P(X=i\quad\text{and}\quad Y=k-i)But these joint probabilities can be expressed using the conditional probabilities which were computed in the previous question, and using the first question:
    P(X+Y=k)=i=0kP(X=i)P(Y=kiX=i)=i=0k(ni)pi(1p)ni(niki)pki(1p)nkP(X+Y=k)=\sum_{i=0}^{k}P(X=i)P(Y=k-i\mid X=i)=\sum_{i=0}^{k}\binom{n}{i}p^{i}(1-p)^{n-i}\binom{n-i}{k-i}p^{k-i}(1-p)^{n-k}and a few terms do not depend on ii, so they can be extracted from the sum:
    P(X+Y=k)=pk(1p)nki=0k(ni)(niki)(1p)niP(X+Y=k)=p^{k}(1-p)^{n-k}\sum_{i=0}^{k}\binom{n}{i}\binom{n-i}{k-i}(1-p)^{n-i}Here there is a nice combinatorial identity:
    (ni)(niki)=(ki)(nk)\binom{n}{i}\binom{n-i}{k-i}=\binom{k}{i}\binom{n}{k}which yields:
    P(X+Y=k)=pk(1p)nk(nk)i=0k(ki)(1p)niP(X+Y=k)=p^{k}(1-p)^{n-k}\binom{n}{k}\sum_{i=0}^{k}\binom{k}{i}(1-p)^{n-i}This sum is very close to a binomial type of expression. We make a kk appear:
    P(X+Y=k)=pk(1p)nk(nk)i=0k(ki)(1p)nk(1p)ki=pk((1p)nk)2(nk)i=0k(ki)(1p)kiP(X+Y=k)=p^{k}(1-p)^{n-k}\binom{n}{k}\sum_{i=0}^{k}\binom{k}{i}(1-p)^{n-k}(1-p)^{k-i}=p^{k}\left((1-p)^{n-k}\right)^{2}\binom{n}{k}\sum_{i=0}^{k}\binom{k}{i}(1-p)^{k-i}=pk((1p)nk)2(nk)(1p+1)k=pk((1p)nk)2(nk)(2p)k=p^{k}\left((1-p)^{n-k}\right)^{2}\binom{n}{k}(1-p+1)^{k}=p^{k}\left((1-p)^{n-k}\right)^{2}\binom{n}{k}(2-p)^{k}By letting q=p(2p)=2pp2q=p(2-p)=2p-p^{2}, we have that 1q=12p+p2=(1p)21-q=1-2p+p^{2}=(1-p)^{2}, we just obtained:
    P(X+Y=k)=(nk)qk(1q)nkP(X+Y=k)=\binom{n}{k}q^{k}(1-q)^{n-k}This is again a binomial distribution B(n,q)\mathcal{B}(n,q).

  4. The expected value of Z=X+YZ=X+Y is nqnq, while the variance is nq(1q)nq(1-q).

Solution by Ancient TreeFR

Discussions0 useful votes
  1. Comme il y a nn appels indépendants, chacun avec une probabilité pp de succès, XX suit une loi binomiale B(n,p)\mathcal{B}(n,p) :
    P(X=i)=(ni)pi(1p)niP(X=i)=\binom{n}{i}p^{i}(1-p)^{n-i}

  2. Lors du deuxième appel, les appels restent indépendants, chacun avec une probabilité pp de succès, mais cette fois, si X=iX=i, il ne reste que nin-i appels à passer. La probabilité conditionnelle suit donc à nouveau une loi binomiale B(ni,p)\mathcal{B}(n-i,p) :
    P(Y=kX=i)=(nik)pk(1p)nikP(Y=k\mid X=i)=\binom{n-i}{k}p^{k}(1-p)^{n-i-k}

  3. Attention : il y a probablement beaucoup mieux que cette méthode "force brute", mais je n’ai toujours pas trouvé la bonne interprétation : Lorsque X=iX=i, il faut nécessairement que Y=kiY=k-i. Cela signifie que :
    P(X+Y=k)=i=0kP(X=ietY=ki)P(X+Y=k)=\sum_{i=0}^{k}P(X=i\quad\text{et}\quad Y=k-i)Or ces probabilités jointes peuvent s’exprimer à l’aide des probabilités conditionnelles calculées à la question précédente, ainsi que du résultat de la première question :
    P(X+Y=k)=i=0kP(X=i)P(Y=kiX=i)=i=0k(ni)pi(1p)ni(niki)pki(1p)nkP(X+Y=k)=\sum_{i=0}^{k}P(X=i)P(Y=k-i\mid X=i)=\sum_{i=0}^{k}\binom{n}{i}p^{i}(1-p)^{n-i}\binom{n-i}{k-i}p^{k-i}(1-p)^{n-k}et certains termes ne dépendent pas de ii, donc on peut les sortir de la somme :
    P(X+Y=k)=pk(1p)nki=0k(ni)(niki)(1p)niP(X+Y=k)=p^{k}(1-p)^{n-k}\sum_{i=0}^{k}\binom{n}{i}\binom{n-i}{k-i}(1-p)^{n-i}On dispose ici d’une identité combinatoire remarquable :
    (ni)(niki)=(ki)(nk)\binom{n}{i}\binom{n-i}{k-i}=\binom{k}{i}\binom{n}{k}ce qui donne :
    P(X+Y=k)=pk(1p)nk(nk)i=0k(ki)(1p)niP(X+Y=k)=p^{k}(1-p)^{n-k}\binom{n}{k}\sum_{i=0}^{k}\binom{k}{i}(1-p)^{n-i}Cette somme fait beaucoup penser à un binôme de Newton, mais la puissance n’est pas la bonne. Faisons apparaître un kk :
    P(X+Y=k)=pk(1p)nk(nk)i=0k(ki)(1p)nk(1p)ki=pk((1p)nk)2(nk)i=0k(ki)(1p)kiP(X+Y=k)=p^{k}(1-p)^{n-k}\binom{n}{k}\sum_{i=0}^{k}\binom{k}{i}(1-p)^{n-k}(1-p)^{k-i}=p^{k}\left((1-p)^{n-k}\right)^{2}\binom{n}{k}\sum_{i=0}^{k}\binom{k}{i}(1-p)^{k-i}=pk((1p)nk)2(nk)(1p+1)k=pk((1p)nk)2(nk)(2p)k=p^{k}\left((1-p)^{n-k}\right)^{2}\binom{n}{k}(1-p+1)^{k}=p^{k}\left((1-p)^{n-k}\right)^{2}\binom{n}{k}(2-p)^{k}En posant q=p(2p)=2pp2q=p(2-p)=2p-p^{2}, on a 1q=12p+p2=(1p)21-q=1-2p+p^{2}=(1-p)^{2}, d’où l’on obtient :
    P(X+Y=k)=(nk)qk(1q)nkP(X+Y=k)=\binom{n}{k}q^{k}(1-q)^{n-k}Il s’agit à nouveau d’une loi binomiale B(n,q)\mathcal{B}(n,q).

  4. L’espérance de Z=X+YZ=X+Y vaut nqnq, tandis que la variance vaut nq(1q)nq(1-q).

Report

For an unclear, ambiguous, or possibly incorrect statement, please use the Discussion tab on the right. Report content that needs moderator intervention, such as dangerous, clearly non-mathematical, or plagiarized content.