Generation phi

Let's define a series S S , with 0 t h 0^{th} element a 0 { a }_{ 0 } as:

S ( a 0 ) = { a 1 , a 2 , a 3 , a 4 , . . . } S({ a }_{ 0 })={ \quad \{ a }_{ 1 },{ a }_{ 2 },{ a }_{ 3 },{ a }_{ 4 },...\}

Where a n = φ a n 1 + 1 φ o r a n = a n 1 φ 1 φ 2 { a }_{ n }=\quad \varphi { a }_{ n-1 }+\frac { 1 }{ \varphi } \quad or\quad { a }_{ n }=\frac { { a }_{ n-1 } }{ \varphi } -\frac { 1 }{ { \varphi }^{ 2 } } with equal probability.

If a 0 = 31 { a }_{ 0 }=31 , the expected value of a 51 {a}_{51} can be expressed as:

a b c d e f \frac { { a }^{ b }\sqrt { c } }{ { d }^{ e } } -f

Find: a + b + c + d + e + f a+b+c+d+e+f

Details and Assumptions:

  1. φ = 5 + 1 2 \varphi=\frac { \sqrt { 5 } +1 }{ 2 }
  2. a , b , c , d , e , f a,b,c,d,e,f are integers. a a and d d are co-prime. c c is a square-free number.
This is part of my set Powers of the ordinary .


The answer is 84.

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.

2 solutions

Okay, consider this:

Define operation O 1 ( a n 1 ) = α a n 1 + β { O }_{ 1 }({ a }_{ n-1 })=\alpha { a }_{ n-1 }+\beta and

O 2 ( a n 1 ) = a n 1 α β α { O }_{ 2 }({ a }_{ n-1 })=\frac { { a }_{ n-1 } }{ \alpha } -\frac { \beta }{ \alpha }

Observe that O 1 { O }_{ 1 } and O 2 { O }_{ 2} are inverse of each other. The series in the question S ( a 0 ) S({ a }_{ 0 }) is formed by repeatedly performing these operations(with equal probability) on the terms.

Now, let's say we did this a few times. Some of the operations we made may be O 1 { O }_{ 1 } and the rest O 2 { O }_{2} . Note that performing O 1 { O }_{ 1 } negates the effect of one O 2 { O }_{ 2} , and vice-versa.

So, if O 1 { O }_{ 1 } is done p p times and O 2 { O }_{ 2} , q q times while making 2 m + 1 2m+1 terms of S ( a 0 ) S({a}_{0}) .

p + q = 2 m + 1 \Rightarrow p+q=2m+1

Now we need only bother about which one of p p or q q is more. let me illustrate: assume p = 5 , q = 4 p=5, q=4 . This means that we performed O 1 , 5 { O }_{ 1 }, 5 times and O 2 , 4 { O }_{ 2}, 4 times. Since doing O 2 { O }_{ 2 } 4 4 times negates the effect of 4 4 O 1 { O }_{ 1 } s we can see that the net effect after performing the 9 9 operations is equivalent to performing O 1 { O }_{ 1 } once.

Also note that since 2 m + 1 2m+1 is odd, p p and q q may differ only by odd numbers.

Hence, to find the expected value, we repeat the following algorithm from p q = ( 2 m + 1 ) p-q=-(2m+1) to p q = 2 m + 1 p-q=2m+1 :

  1. Find probability P ( p ) P(p) of p q p-q taking given value.

  2. Multiply P P with value of a 0 {a}_{0} when O 1 {O}_{1} is performed on it p p times and O 2 {O}_{2} is performed q q times.

  3. Add to result.

The probability of p p being a given value is easy to find and is:

P ( p ) = ( 2 m + 1 p ) 2 m + 1 P(p)=\frac { \left( \begin{matrix} 2m+1 \\ p \end{matrix} \right) }{ { 2 }^{ m+1 } }

for example, P ( 2 ) P(2) will give probability of p p being 2 2 or alternatively, the probability of q q being 2 m 1 2m-1 .

Now what's left to do is to find the general term for when O 1 {O}_{1} and O 2 {O}_{2} are performed n n times.

I am writing the formula here without proof, and leaving it as an exercise to the reader to arrive at this formula.

X n = O 1 ( O 1 ( . . n t i m e s ( a 0 ) ) ) = ( a 0 + β α 1 ) α n β α 1 Y n = O 2 ( O 2 ( . . n t i m e s ( a 0 ) ) ) = ( a 0 + β α 1 ) 1 α n β α 1 { X }_{ n }={ O }_{ 1 }({ O }_{ 1 }(..n\quad times({ a }_{ 0 })))=({ a }_{ 0 }+\frac { \beta }{ \alpha -1 } ){ { \alpha } }^{ n }-\frac { \beta }{ \alpha -1 } \\ { Y }_{ n }={ O }_{ 2 }({ O }_{ 2 }(..n\quad times({ a }_{ 0 })))=({ a }_{ 0 }+\frac { \beta }{ \alpha -1 } )\frac { 1 }{ { \alpha }^{ n } } -\frac { \beta }{ \alpha -1 }

After performing the algorithm for finding expected value, the required value( Q Q ) is:

Q = 1 2 2 m + 1 × [ i = 1 m P ( m + i ) X 2 i + 1 + j = 1 m P ( m + j ) Y 2 j + 1 ] Q=\frac { 1 }{ { 2 }^{ 2m+1 } } \times \left[ \sum _{ i=1 }^{ m }{ P(m+i){ X }_{ 2i+1 } } +\sum _{ j=1 }^{ m }{ P(m+j){ Y }_{ 2j+1 } } \right]

= 1 2 2 m + 1 × [ i = 1 m ( 2 m + 1 m + i ) X 2 i + 1 + j = 1 m ( 2 m + 1 m + j ) Y 2 j + 1 ] =\frac { 1 }{ { 2 }^{ 2m+1 } } \times \left[ \sum _{ i=1 }^{ m }{ \left( \begin{matrix} 2m+1 \\ m+i \end{matrix} \right) { X }_{ 2i+1 } } +\sum _{ j=1 }^{ m }{ \left( \begin{matrix} 2m+1 \\ m+j \end{matrix} \right) { Y }_{ 2j+1 } } \right]

After substituting values for X i {X}_{i} and Y j {Y}_{j} and using binomial theorem, we get:

Q = ( a 0 + β α 1 ) ( α + 1 α ) 2 m + 1 2 2 m + 1 β α 1 Q=\left( { a }_{ 0 }+\frac { \beta }{ \alpha -1 } \right) \frac { { (\alpha +\frac { 1 }{ \alpha } ) }^{ 2m+1 } }{ { 2 }^{ 2m+1 } } -\frac { \beta }{ \alpha -1 }

Now putting α = φ \alpha =\varphi , β = 1 φ \beta =\frac { 1 }{ \varphi } , a 0 = 31 {a}_{0}=31 , 2 m + 1 = 51 2m+1=51 , we get:

Q = 5 25 5 2 46 1 Q=\frac { { 5 }^{ 25 }\sqrt { 5 } }{ { 2 }^{ 46 } } -1

Thus, the required answer is 5 + 25 + 5 + 2 + 46 + 1 = 84 5+25+5+2+46+1=\boxed{84}

Shubham Garg
Jul 1, 2015

According to the question;

a n = φ a n 1 + 1 φ o r a n = a n 1 φ 1 φ 2 a_{n}=\varphi a_{n-1} + \frac{1}{\varphi} \quad or \quad a_{n}=\frac{a_{n-1}}{\varphi}-\frac{1}{{\varphi}^2}

with equal probabilities.

This means if A n A_{n} is the expected value then

A n = 1 2 ( φ A n 1 + 1 φ ) + 1 2 ( A n 1 φ 1 φ 2 ) A_{n}= \frac{1}{2}(\varphi A_{n-1} + \frac{1}{\varphi}) +\frac 1 2 (\frac{A_{n-1}}{\varphi}-\frac{1}{{\varphi}^2})

A n = 1 2 ( φ A n 1 + 1 φ + A n 1 φ 1 φ 2 ) \Rightarrow A_{n}= \frac{1}{2}(\varphi A_{n-1} + \frac{1}{\varphi} + \frac{A_{n-1}}{\varphi}-\frac{1}{{\varphi}^2})

A n = 1 2 ( A n 1 ( φ + 1 φ ) + ( 1 φ 1 φ 2 ) ) \Rightarrow A_{n}= \frac{1}{2}(A_{n-1}(\varphi +\frac{1}{\varphi}) + (\frac{1}{\varphi}-\frac{1}{{\varphi}^2}))

Putting the value of φ = 5 + 1 2 \varphi=\frac{\sqrt{5}+1}{2} and 1 φ = 5 1 2 \frac{1}{\varphi}=\frac{\sqrt{5}-1}{2} we get ,

A n = 1 2 ( A n 1 5 + 5 2 ) \Rightarrow A_{n}= \frac{1}{2}(A_{n-1}\sqrt{5} + \sqrt{5}-2)

A n = 1 2 ( 5 ( A n 1 + 1 ) 2 ) \Rightarrow A_{n}= \frac{1}{2}(\sqrt{5}(A_{n-1} + 1) -2)

A n = 1 2 ( 5 ( A n 1 + 1 ) ) 1 \Rightarrow A_{n}= \frac{1}{2}(\sqrt{5}(A_{n-1} + 1)) -1

A n + 1 = 5 2 ( A n 1 + 1 ) \Rightarrow A_{n} + 1= \frac{\sqrt{5}}{2}(A_{n-1} + 1)

Now let A n + 1 = T n A_{n}+1=T_{n} with T 0 = 32 T_0=32

T n = 5 2 T n 1 \Rightarrow T_{n} = \frac{\sqrt{5}}{2}T_{n-1}

Now a GP if formed with common ratio 5 2 \frac{\sqrt{5}}{2}

This gives us T 51 = T 0 ( 5 2 ) 51 T_{51}=T_{0} (\frac{\sqrt{5}}{2})^{51}

and A 51 = T 51 1 = ( 32 ) ( 5 2 ) 51 1 A_{51}=T_{51}-1=(32) (\frac{\sqrt{5}}{2})^{51}-1

A 51 = 5 25 5 2 46 1 \Rightarrow A_{51}=\frac{5^{25} \sqrt{5}}{2^{46}}-1

a + b + c + d + e + f = 5 + 25 + 5 + 2 + 46 + 1 = 84 \Rightarrow a+b+c+d+e+f=5+25+5+2+46+1=\boxed{84}

Nice solution using the value of ϕ \phi beforehand to make calculation easier! (+1)

Raghav Vaidyanathan - 5 years, 11 months ago

Log in to reply

Thank you :D

Shubham Garg - 5 years, 11 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...