Michael's expected value

Three distinct numbers are selected uniformly at random from the ten-term geometric sequence with first term 10 9 \dfrac{10}{9} and common ratio 2 2 . What is the expected value of their sum?

This problem is shared by Michael T .


The answer is 341.

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.

11 solutions

Oliver Welsh
Nov 25, 2013

Firstly, note that there are ( 10 3 ) = 120 {10 \choose 3} = 120 ways of choosing 3 3 distinct numbers from the geometric series. Now to find the expected value of the sum, we need to sum each possible combination and divide the total by 120 120 (because the numbers are selected uniformly at random). So the sum is: ( 10 9 + 20 9 + 40 9 ) + ( 10 9 + 20 9 + 80 9 ) + 120 \frac{\left (\frac{10}{9} + \frac{20}{9} + \frac{40}{9}\right ) + \left (\frac{10}{9} + \frac{20}{9} + \frac{80}{9}\right ) + \dots}{120} Now, notice that each term will appear ( 9 2 ) = 36 {9 \choose 2} = 36 times in the sum, so the sum can be simplified to: 36 ( 10 9 + 20 9 + 40 9 + + 5120 9 ) 120 \frac{36 \cdot \left (\frac{10}{9} + \frac{20}{9} + \frac{40}{9} + \dots + \frac{5120}{9} \right )}{120} Now taking out factors we further simplify the sum to: 10 36 ( 1 + 2 + 4 + + 512 ) 9 120 = 1 + 2 + 4 + + 512 3 \frac{10 \cdot 36 \cdot (1 + 2 + 4 + \dots + 512)}{9 \cdot 120} = \frac{1 + 2 + 4 + \dots + 512}{3} Now we will evaluate the geometric series 1 + 2 + 4 + + 512 1 + 2 + 4 + \dots + 512 . This is equal to: S 10 = 1 ( 2 10 1 ) 2 1 = 2 10 1 = 1023 S_{10} = \frac{1 \cdot (2^{10} - 1)}{2 - 1} = 2^{10} - 1 = 1023 Therefore, the expected value is: 1023 3 = 341 \frac{1023}{3} = \fbox{341}

Jus' like I did. Good job!

Guilherme Dela Corte - 7 years, 6 months ago

Good job Oliver!

Michael Tang - 7 years, 6 months ago
Anis Abboud
Nov 24, 2013
  • The sum of the series is a 1 q n 1 q 1 = 10 9 2 1 0 1 2 1 = 10 341 3 a_1 \dfrac{q^n - 1}{q - 1} = \dfrac{10}{9} \dfrac{2^10 - 1}{2 - 1} = 10 \dfrac{341}{3} .
  • The average of the 10 numbers is then 341 3 \dfrac{341}{3} .
  • So on average, the sum of three numbers would be three times this number 341 \Rightarrow \boxed{341} .

"So on average, the sum of three numbers would be three times this number" Why can we state that?

Guilherme Dela Corte - 7 years, 6 months ago

Log in to reply

I believe because of the linearity of expectation in probability... i.e., if X 1 , X 2 , X 3 X_1, X_2, X_3 are the three random variables representing the numbers, we know that for each, E [ X i ] = μ E[X_i] = \mu , so E [ X 1 + X 2 + X 3 ] = E [ X 1 ] + E [ X 2 ] + E [ X 3 ] = μ + μ + μ = 3 μ E[X_1 + X_2 + X_3] = E[X_1] + E[X_2] + E[X_3] = \mu + \mu + \mu = 3\mu .

http://en.wikipedia.org/wiki/Expected_value#Linearity

Anis Abboud - 7 years, 6 months ago
Sandip Nath
Jan 12, 2014

Let us calculate the result for: 1, 2, 4, ... 512

We will multiply this result by 10 / 9 at the end.

With any of these numbers selected, we can select any 2 other numbers from the rest in 9 * 8/2 = 36 ways Adding all the cases for all the numbers will give us: (1+2+4+8+...+512) * 36 = 1023 * 36

The total number of ways of selecting 3 nums is of course the number of ways of selecting 3 things from 10, which is 120

Hence, result = 1023 * 36 / 120

Multiplying this result by 10 / 9 we get , final result = 1023 * 36 / 120 * 10 / 9 = 341

Nahom Yemane
Jan 3, 2014

The number of ways of picking three numbers from ten is ( 10 3 ) 10\choose3 = 120 =120 If we fix a certain number of the geometric series- the number of ways we could choose two other is ( 9 2 ) 9\choose2 = 36 = 36 So if we pick a number from the series for our three number sum it could appear in 36 36 sums.

Now, our E x p e c t e d S u m Expected Sum = S u m o f A l l P o s s i b l e S u m s N u m b e r o f A l l P o s s i b l e S u m s =\frac{Sum of All Possible Sums}{Number of All Possible Sums}

If you think about the S u m o f A l l P o s s i b l e S u m s Sum of All Possible Sums its really just 36 × 36\times the sum of all the numbers in the geometric series. Since in adding all of the possible sums each member of the geometric series will appear 36 36 times. We can factor out the 36 36 to get a sum of a geometric series. So assuming the reader is aware of the sum of a geometric series we get:

E x p e c t e d S u m Expected Sum = S u m o f A l l P o s s i b l e S u m s N u m b e r o f A l l P o s s i b l e S u m s = 36 ( 10 9 ) ( 2 10 1 ) 120 = 341 =\frac{Sum of All Possible Sums}{Number of All Possible Sums} =\frac{36(\frac{10}{9})(2^{10}-1)}{120}=\boxed{341}

I guess that was an interesting solution but a bit long winded It is better to take the average of all the numbers and just multiply by 3

Nahom Yemane - 7 years, 5 months ago

all ideas are about symmetric.

let x 1 . x 2 , x 3 x_1.x_2,x_3 are the numbers that choose from the set of a i , i = 1 , 2 , 3 , . . . , 10 a_i,i=1,2,3,...,10 .

easily to find the number of ways to write ( x 1 , x 2 , x 3 ) (x_1,x_2,x_3) that is ( 10 3 ) = 120 \binom{10}{3}=120 . when you list ( x 1 , x 2 , x 3 ) (x_1,x_2,x_3) for each x 1 x_1 you will find ( 9 2 ) = 36 \binom{9}{2}=36 ways to choose x 2 , x 3 x_2,x_3 .

so the expected value that you want will be

i = 1 10 [ 1 ( 10 3 ) ( 9 2 ) a i ] = 341 \sum_{i=1}^{10}[\frac{1}{\binom{10}{3}}\cdot \binom{9}{2} \cdot a_i] = 341

Tanishq Aggarwal
Nov 24, 2013

The key to solving this problem is to write out the expected value as an expression, and then cleverly simplify the expression.

Note that the expected value of the sum is

( 1 ( 10 3 ) ) ( 10 9 + 20 9 + 40 9 ) + ( 1 ( 10 3 ) ) ( 10 9 + 20 9 + 80 9 ) + ( 1 ( 10 3 ) ) ( 10 9 + 20 9 + 160 9 ) . . . (\frac{1}{\binom{10}{3}})(\frac{10}{9}+\frac{20}{9}+\frac{40}{9}) + (\frac{1}{\binom{10}{3}})(\frac{10}{9}+\frac{20}{9}+\frac{80}{9}) + (\frac{1}{\binom{10}{3}})(\frac{10}{9}+\frac{20}{9}+\frac{160}{9})...

where the sum continues for all possible triplets of elements. If we expand and collect, we get

( 1 ( 10 3 ) ) ( 10 9 + 10 9 + 10 9 10 9 + 20 9 + 20 9 + 20 9 20 9 + 40 9 + 40 9 + 40 9 40 9 . . . 5120 9 + 5120 9 + 5120 9 5120 9 ) (\frac{1}{\binom{10}{3}})(\frac{10}{9}+\frac{10}{9}+\frac{10}{9} \ldots \frac{10}{9} + \frac{20}{9}+\frac{20}{9}+\frac{20}{9} \ldots \frac{20}{9}+\frac{40}{9}+\frac{40}{9}+\frac{40}{9} \ldots \frac{40}{9}...\frac{5120}{9}+\frac{5120}{9}+\frac{5120}{9} \ldots \frac{5120}{9})

How many times does each term above repeat? Each term is used in a triplet with two other distinct elements of the sequence. There are ( 9 2 ) \binom{9}{2} ways to choose the other two elements, so each term must appear ( 9 2 ) \binom{9}{2} times. The simplification of the above quantity is now trivial (and is left to the reader), leaving a final answer of 341 \boxed{341} .

Or you could use E[x+y+z]=E(x)+E(y)+E(z). Let the three numbers are x,y and z. The sum of the series is 10230 9 \frac{10230}{9} and probability of selecting x,y or z is 1 10 \frac{1}{10} each. E(x)=E(y)=E(z)= 10230 9 × 1 10 = 341 3 \frac{10230}{9} \times \frac{1}{10} = \frac{341}{3} E(x+y+z)= 341 \boxed{341}

Snehdeep Arora - 7 years, 6 months ago

Log in to reply

I did it the same way. But I am somewhat confused. If we select first number, there will be only 9 9 choices for second number. Similarly, there will be only 8 8 choices for the third number. Shouldn't the probabilities and then corresponding expectation values change for second and third number? How do we justify equating E ( x ) , E ( y ) E(x),E(y) and E ( z ) E(z) ?

Snehal Shekatkar - 7 years, 6 months ago

Log in to reply

x,y and z are random variables,the probability is calculated for the values these random variables can take.We have ten values and each value is equally likely for x,y and z.

Snehdeep Arora - 7 years, 6 months ago

Log in to reply

@Snehdeep Arora No.. that is different issue.. my question is.. shouldn't we use conditional probability to calculate E ( y ) E(y) and E ( z ) E(z) ?

Snehal Shekatkar - 7 years, 6 months ago

Log in to reply

@Snehal Shekatkar No, I don't think this has anything to do with conditional probability.We are using Linearity of expectation, meaning to calculate the expected sum of random variables we add their * individual expectations *.

Snehdeep Arora - 7 years, 6 months ago

@Snehal Shekatkar You can read the technique trainer post about Expected Value to learn about expectation and linearity of expectation. In the case of this problem, using linearity of expectation is a way for us to simplify calculations by not having to do conditional probability.

This explanation might help a bit, too. Suppose we are doing this conditionally. If we chose a small value for x x then E ( y ) E(y) will be larger because we cannot have that large value for y . y. . If we chose a large value for x x then E ( y ) E(y) will be smaller because we cannot have that large value for y . y. Since we must consider all values of x x , these changes in E ( y ) E(y) will balance out to be the same as E ( x ) . E(x).

Note that this only applies because we are taking the expected value of a sum.

Lino Demasi - 7 years, 6 months ago

Log in to reply

@Lino Demasi Exactly! That is the whole point! So suppose we don't ask for expected value of sum. Then can we say in this case that E ( x ) = E ( y ) = E ( z ) E(x)=E(y)=E(z) ?

Snehal Shekatkar - 7 years, 6 months ago
Aivin Solatorio
Nov 29, 2013

The approach I used to solve this problem is the following.


Definitions:

We first note that the n t h n^{th} term of a geometric sequence is given by a n = a 1 r n 1 a_n = a_1 r^{n-1} . We then define the set S S containing the ten-term geometric sequence where a 1 = 10 9 a_1 = \frac{10}{9} and r = 2 r = 2 .

S = { 10 9 × 2 n 1 } n = 1 10 \ \ \ \ \ \ \ \ \ S = \{\frac{10}{9} \times 2^{n-1}\} _{n=1}^{10}

Furthermore, we define the set T T containing all the triples formed from S S .

T = { s S s = 3 } \ \ \ \ \ \ \ \ \ T = \{ s \in S\ \ \ |\ \ \ \ |s| = 3\}

The number of these triples is equal to T = 10 C 3 = 120 |T| = \ _{10}C_{3} = 120 . The set of the sum of these triples can then be defined as σ = { l = 1 s s n , l } s n T \sigma = \{\sum\limits_{l = 1}^{|s|} s_{n,l} \}_{s_n \in T} .

Here, we should note that a triple s n s_n has the form 10 9 ( 2 i , 2 j , 2 k ) n \frac{10}{9}(2^i, 2^j, 2^k)_n where i j k i\neq j\neq k and i , j , k [ 0 , 9 ] i,j,k \in [0,9] .

We can then get the solution by solving the mean value of σ \sigma .


Calculation of mean value of σ \sigma :

The mean value of σ \sigma is defined as σ ˉ = i = 1 σ σ i σ \bar{\sigma} = \frac{\sum\limits_{i=1}^{|\sigma|} \sigma_i}{|\sigma|} , where σ = T = 120 |\sigma| = |T| = 120 . With this, we're left with the task of calculating the value of i = 1 σ σ i \sum\limits_{i=1}^{|\sigma|} \sigma_i .

We can simplify the summation by noting that in the entire collection of triples, a specific number is repeated 36 times. This is found by the fact that a certain number in the set S S will be paired with any other two numbers in 9 C 2 = 36 \ _{9}C_{2} = 36 ways.

Since we are solving for the sum of all the triples, we find that the value of the summation is equal to 36 × 10 9 × n = 1 10 2 n 1 36 \times \frac{10}{9} \times \sum\limits_{n=1}^{10} 2^{n-1} .

Combining the numerator and the denominator, we get

σ ˉ = 36 × 10 9 × n = 1 10 2 n 1 120 \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \bar{\sigma} = \frac{36 \times \frac{10}{9} \times \sum\limits_{n=1}^{10} 2^{n-1}}{120}

σ ˉ = 36 × 10 9 × 1023 120 \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \bar{\sigma} = \frac{36 \times \frac{10}{9} \times 1023}{120}

σ ˉ = 341 \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \bar{\sigma} = 341


Answer: σ ˉ = 341 \bar{\sigma} = 341

William Cui
Nov 26, 2013

Though this question may appear to be quite complicated, it is actually pretty simple.

The numbers in the sequence are 10 9 , 20 9 , 40 9 , 5120 9 \frac{10}{9}, \frac{20}{9}, \frac{40}{9}, \ldots \frac{5120}{9} We know this because the first term is 10 9 \frac{10}{9} and the common ration is 2.

The sum of these terms is equal to 10 + 20 + 40 + . . . + 2560 + 5120 9 = 10230 9 \frac{10+20+40+...+2560+5120}{9}=\frac{10230}{9} by using the fact that

2 0 + 2 1 + + 2 n = 2 n + 1 \qquad \qquad \qquad \quad \quad \quad 2^0+2^1+\ldots + 2^n = 2^{n+1}

However, we do not want to add all ten; we just want to choose three of the ten. So, we have

3 10 10230 9 = 3 10230 10 9 = 1023 3 = 341 \qquad\qquad \qquad\qquad \frac{3}{10}\cdot\frac{10230}{9}=\frac{3\cdot10230}{10\cdot 9}=\frac{1023}{3}=\boxed{341}

This is essentially the method I used but the sum of powers of 2 should equal: 2 n + 1 1 2^{n+1} - 1 (which I see you used correctly).

Andrew Sellek - 7 years, 6 months ago

Log in to reply

Yes, oops, that was a typo.

William Cui - 7 years, 6 months ago
Sanjay Banerji
Nov 25, 2013

We will choose 3 numbers from 10 numbers in 10C3 = 120 ways. Since in each choice we choose 3 numbers we are actually choosing (3*120) = 360 numbers. Thus each number is chosen (360 / 10) = 36 times

So we multiply 36 to each of the numbers 10/9,20/9,.....,5120/9 and get 40,80,...,20480.

Adding this (40 + 80 + .... + 20480) we get 40920 and since this is the sum of all the numbers taken in all possible groups of three dividing this by 120 gives us the answer as 341

Marco Rossi
Nov 29, 2013

First of all, I consider the sum of all possible groups of three numbers. To do that, I calculate the number of times that each member appears in this sum. We select a random term. We can select the other two terms in ( 9 2 ) = 36 \binom{9}{2}=36 . So this sum is: 36 10 9 1 + 2 + 4 + . . . + 512 = 40 1023 36\cdot \frac{10}{9} \cdot {1+2+4+...+512}=40 \cdot 1023 since the sum of the first n n powers of two is 2 n + 1 1 2^{n+1}-1 . To obtain the average value, we have to divide this sum by the number of possible terns, which is ( 10 3 ) = 120 \binom{10}{3}=120 . So 40 1023 120 = 341 \frac{40 \cdot 1023 }{120}=341 which is the answer.

Timothy Zhou
Nov 26, 2013

The sequence runs (10/9) * 1, (10/9) * 2, (10/9) * 4, ... (10/9) * 512. Now each term will be added into the sum in (9 choose 2) triples , since we choose the other two distinct numbers from the remaining nine. Then 10/9 (1+2+4+... +512) * (9 c 2) is the total sum. Also, 1/(10 c 3) is the probability that each triple will occur. So we multiply the two together to get the expected value. Using the identity 1+2+4...+ 2^n = 2^(n+1) -1, we can evaluate this expression quickly and get the answer 341.

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...