I'm trying to solve this recurrence, but I don't know how to unfold it.
T(n)=2T((n+2)/3) + 1
Can I ignore that "+2" and solve it as it was 2T(n/3) + 1?
This comes from a from a problem that uses a V[a..b] array and makes this return:
return V(X) + f(V, a, Y) + f(V, Z, b)
Where Y is (2a+b)/3 and Z is (a+2b)/3
So: ((b-a+3)/3) = ((n+2)/3)
Sort of. The rigorous version of this trick is to set U(n) = T(n+1) and write
U(n) = T(n+1)
= 2T((n+1+2)/3) + 1
= 2T(n/3 + 1) + 1
= 2U(n/3) + 1.
Then solve for U (e.g., U(n) = O(n^log3(2))) and then you should be able to find an asymptotic expression for T of the same order.
Related
T(n) = 1/2(T(n − 1) + T(n − 2)) + cn, with c > 0
I am having trouble understanding how to solve recurrences with multiple T(n)s. I did a lot of practices by solving recurrence with just one T(n) and following the definition I can do it well. But this is not a recurrence directly solvable with the Master theorem. Anyway I can start a good approach to this question?
solve the homogeneous recurrence:
T_H(n) = 1/2(T_H(n − 1) + T_H(n − 2))
r^2 - r/2 - 1/2 = 0
r = 1 or r = -1/2
T_H(n) = alpha * 1^n + beta * (-1/2)^n (alpha and beta to be determined by initial conditions)
solve the special solution
(1) we want to find a s(n) such that s(n) = 1/2(s(n-1)+s(n-2)) + cn
we know cn is a polynome (in n) so special solution can be found as a polynome too.
Trying with s(n) = an leads to:
an = 1/2(an-1 + an-2) + cn and all terms in an simplify themselves so try the next degree: s(n)=an^2 + bn
an^2 + bn = 1/2 (a(n-1)^2 + b(n-1) + a(n-2)^2 + b(n-2) ) + cn
developping everybody then identifying we get
a = c/3
b = 5c/9
A quick check if we don't trust our ability to make valid calculus:
since s(n) must be valid for all n, let's put arbitrarily n=2, c=7 and check whether s(2) still verifies (1) idem
n = 2, c=7
s(n)-1/2(s(n-1)+s(n-2))-cn ?= 0
below octave shows that indeed s(2) = 0
octave:1> n=2
n = 2
octave:2> c=7
c = 7
octave:3> c/3*n^2 + 5*c/9*n - 1/2*(c/3*(n-1)^2 + 5*c/9*(n-1) +c/3*(n-2)^2 + 5*c/9*(n-2))-c*n
ans = 0
Complexity
T(n) = T_H(n) + sp(n) = alpha + beta (-1/2)^n + c/3n^2 + 5c/9n
so T(n) is in O(n^2)
This program calculate fibonacci numbers. I want to find out its time complexity using recurrence relation.
Fib(n)
if n<=1
return n
else
x= Fib(n-1)
y= Fib(n-2)
return x+y
The recurrence equation for this program is
T(n)=T(n-1)+T(n-2)+c
I tried to expend it , but couldn't find the solution.
=2T(n-1)+T(n-3)+c+c
=3T(n-3)+2T(n-4)+c+c+3c
=5T(n-4)+3T(n-3)+c+c+3c+5c
-------------------------
-------------------------
-------------------------
You need to think about how many calls you make to your function.
Each call makes 2, so it makes a binary tree:
n
(n-1)--------------(n-2)
(n-2)--(n-3)------(n-3)---(n-4)
and so on.
Consider the level of the tree when you first reach 1 and ignore everything below.
This happens on level n/2 (since the lowest number of each level is the rightmost and it always decreases by 2).
It's clear that the nodes on each level up to n/2 are always twice as many as on the previous level.
Thus the total number of nodes is 1 + 2 + 2^2 + ... + 2^(n/2) = 2^(n/2+1) - 1 = O(2^(n/2))
This means that the time complexity is at least exponential.
You can probably compute it even more accurately, but for all practical purposes this should be enough to avoid this implementation.
The given recurrence relation is,
T(n) = T(n-1) + T(n-2) + c ------ 1
T(n-1)= T(n-2) + T(n-3) + c ------ 2
1-2 -> T(n) = 2T(n-1) - T(n-3) ----- 3
T(n) - 2T(n-1) + T(n-3) = 0 ----- 4
The characteristic equation of 4 is x3 - 2x2 + 1 = 0 ---- 5
Solve equation 5,
The solutions are x = 1, x = (1 + √5)/2 and x = (1 −√5)/2
There for the general solution is,
Tn = a((1 + √5)/2)n + b((1 - √5)/2)n + c . 1n
There for Tn = a((1 + √5)/2)n + b((1 - √5)/2)n + c
Let us assume T(0) = 0, from equation 1 we get T(1) = c and T(2) = 2c
There for,
T(0) = a + b + c = 0 ---- 6
T(1) = a((1 + √5)/2) + b((1 - √5)/2) + c = c
There for a((1 + √5)/2) + b((1 - √5)/2) = 0 ----- 7
T(2) = a((1 + √5)/2)2 + b((1 - √5)/2)2 + c = 2c
There for a((1 + √5)/2)2 + b((1 - √5)/2)2 = c ---- 8
solve 6, 7 and 8, to get the values of a, b and c.
The general solution is,
Tn = a((1 + √5)/2)n + b((1 - √5)/2)n + c
since (1 + √5)/2 < 2,
T(n) = O(2n).
The thing to note about your recurrence relation is that it's the same as the Fibonacci recurrence itself. This means that you're doing c units of work times whatever Fibonacci number you're calculating. You can see it yourself from the first few steps that you computed. The c's start growing like the Fibonacci numbers.
Basically your recurrence comes down to O(Fib(n)). Fibonacci numbers are exponential in n, so you're going to do exponential work.
A better way to do this is to remember one of the numbers. Like this:
Fib(n):
if n <= 2:
return 1,0
else:
x,y = Fib(n-1)
return x+y,x
So when you call Fib(n), you're getting two values, Fib(n) and Fib(n-1). That extra x that you return "remembers" Fib(n-2) so you don't have to compute it twice. This recurrence comes down to T(n) = T(n-1) + c, which is O(n).
Once you have that, you can reduce this to a nice little for loop:
x = 1, y = 0
for i from 3 to n:
x,y = x+y,x
I am refreshing on Master Theorem a bit and I am trying to figure out the running time of an algorithm that solves a problem of size n by recursively solving 2 subproblems of size n-1 and combine solutions in constant time.
So the formula is:
T(N) = 2T(N - 1) + O(1)
But I am not sure how can I formulate the condition of master theorem.
I mean we don't have T(N/b) so is b of the Master Theorem formula in this case b=N/(N-1)?
If yes since obviously a > b^k since k=0 and is O(N^z) where z=log2 with base of (N/N-1) how can I make sense out of this? Assuming I am right so far?
ah, enough with the hints. the solution is actually quite simple. z-transform both sides, group the terms, and then inverse z transform to get the solution.
first, look at the problem as
x[n] = a x[n-1] + c
apply z transform to both sides (there are some technicalities with respect to the ROC, but let's ignore that for now)
X(z) = (a X(z) / z) + (c z / (z-1))
solve for X(z) to get
X(z) = c z^2 / [(z - 1) * (z-a)]
now observe that this formula can be re-written as:
X(z) = r z / (z-1) + s z / (z-a)
where r = c/(1-a) and s = - a c / (1-a)
Furthermore, observe that
X(z) = P(z) + Q(z)
where P(z) = r z / (z-1) = r / (1 - (1/z)), and Q(z) = s z / (z-a) = s / (1 - a (1/z))
apply inverse z-transform to get that:
p[n] = r u[n]
and
q[n] = s exp(log(a)n) u[n]
where log denotes the natural log and u[n] is the unit (Heaviside) step function (i.e. u[n]=1 for n>=0 and u[n]=0 for n<0).
Finally, by linearity of z-transform:
x[n] = (r + s exp(log(a) n))u[n]
where r and s are as defined above.
so relabeling back to your original problem,
T(n) = a T(n-1) + c
then
T(n) = (c/(a-1))(-1+a exp(log(a) n))u[n]
where exp(x) = e^x, log(x) is the natural log of x, and u[n] is the unit step function.
What does this tell you?
Unless I made a mistake, T grows exponentially with n. This is effectively an exponentially increasing function under the reasonable assumption that a > 1. The exponent is govern by a (more specifically, the natural log of a).
One more simplification, note that exp(log(a) n) = exp(log(a))^n = a^n:
T(n) = (c/(a-1))(-1+a^(n+1))u[n]
so O(a^n) in big O notation.
And now here is the easy way:
put T(0) = 1
T(n) = a T(n-1) + c
T(1) = a * T(0) + c = a + c
T(2) = a * T(1) + c = a*a + a * c + c
T(3) = a * T(2) + c = a*a*a + a * a * c + a * c + c
....
note that this creates a pattern. specifically:
T(n) = sum(a^j c^(n-j), j=0,...,n)
put c = 1 gives
T(n) = sum(a^j, j=0,...,n)
this is geometric series, which evaluates to:
T(n) = (1-a^(n+1))/(1-a)
= (1/(1-a)) - (1/(1-a)) a^n
= (1/(a-1))(-1 + a^(n+1))
for n>=0.
Note that this formula is the same as given above for c=1 using the z-transform method. Again, O(a^n).
Don't even think about Master's Theorem. You can only use Masther's Theorem when you're given master's theorem when b > 1 from the general form T(n) = aT(n/b) + f(n).
Instead, think of it this way. You have a recursive call that decrements the size of input, n, by 1 at each recursive call. And at each recursive call, the cost is constant O(1). The input size will decrement until it reaches 1. Then you add up all the costs that you used to make the recursive calls.
How many are they? n. So this would take O(2^n).
Looks like you can't formulate this problem in terms of the Master Theorem.
A good start is to draw the recursion tree to understand the pattern, then prove it with the substitution method. You can also expand the formula a couple of times and see where it leads.
See also this question which solves 2 subproblems instead of a:
Time bound for recursive algorithm with constant combination time
May be you could think of it this way
when
n = 1, T(1) = 1
n = 2, T(2) = 2
n = 3, T(3) = 4
n = 4, T(4) = 8
n = 5, T(5) = 16
It is easy to see that this is a geometric series 1 + 2+ 4+ 8 + 16..., the sum of which is
first term (ratio^n - 1)/(ratio - 1). For this series it is
1 * (2^n - 1)/(2 - 1) = 2^n - 1.
The dominating term here is 2^n, therefore the function belongs to Theta(2^n). You could verify it by doing a lim(n->inf) [2^n / (2^n - 1)] = +ve constant.
Therefore the function belongs to Big Theta (2^n)
I can find the sum of each row (n/log n-i) and also I can draw its recursive tree but I can't calculate sum of its rows.
T(n)=2T(n/2)+n/logn
T(1) = 1
Suppose n = 2^k;
We know for harmonic series (euler formula):
Sum[i = 1 to n](1/i) ~= log(n) [n -> infinity]
t(n) = 2t(n/2) + n/log(n)
= 2(2t(n/4) + n/2/log(n/2)) + n/log(n)
= 4t(n/4) + n/log(n/2) + n/log(n)
= 4(2t(n/8) + n/4/log(n/4)) + n/log(n/2) + n/log(n)
= 8t(n/8) + n/log(n/4) + n/log(n/2) + n/log(n)
= 16t(n/16) + n/log(n/8) + n/log(n/4) + n/log(n/2) + n/log(n)
= n * t(1) + n/log(2) + n/log(4) + ... + n/log(n/2) + n/log(n)
= n(1 + Sum[i = 1 to log(n)](1/log(2^i)))
= n(1 + Sum[i = 1 to log(n)](1/i))
~= n(1 + log(log(n)))
= n + n*log(log(n)))
~= n*log(log(n)) [n -> infinity]
When you start unrolling the recursion, you will get:
Your base case is T(1) = 1, so this means that n = 2^k. Substituting you will get:
The second sum behaves the same as harmonic series and therefore can be approximated as log(k). Now that k = log(n) the resulting answer is:
Follow Extended Masters Theorem Below.
Using Extended Masters Theorem T(n)=2T(n/2)+n/logn can be solved easily as follows.
Here n/log n part can be rewritten as n * (logn)^-1,
Effictively maaking value of p=-1.
Now Extended Masters Theorem can be applied easily, it will relate to case 2b of Extended Masters Theorem .
T(n)= O(nloglogn)
Follow this for more detailed explanation
https://www.youtube.com/watch?v=Aude2ZqQjUI
I'm not sure if this is the right place to post this, but the problem actually belongs to a programming assignment. This recursion is something I probably should know how to solve but Im having a bit of trouble with it.
Solve the recursion:
T(0) = 2;
T(n) = T(n-1) + 2;
Solution:
T(n) = 2(n+1)
Could someone please show me how they got to that solution?
Please not that its not the main part of the assignment to solve this particular problem.
You have to figure out what is solution and then you can use induction, to prove it.
To figure solution is simple.
Value is previous value + 2.
2, 2+2, 2+2+2, 2+2+2+2, 2+2+2+2+2, ...
Use induction to prove:
T(0) = 2
T(n) = T(n-1) + 2;
Solution
T(n) = 2(n+1)
Proof:
T(n) = T(n-1) + 2 => 2((n-1)+1) + 2 = 2(n+1)
Check for n=0
2(0+1)=2
End of proof
Try writing out the first few values - it should then be obvious.
Take T(5):
T(5)
|
+-> T(4) + 2
|
+-> T(3) + 2
|
+-> T(2) + 2
|
+-> T(1) + 2
|
+-> T(0) + 2
|
+-> 2
Now count the number of 2's that are added together for T(5).
Then try to figure out how many 2's would be added for T(n).
It's an arithmetic progression with ratio common difference 2.
The first term is T[0] = 2 and the ratio common difference is r = 2 so the n + 1th term (n + 1th because there are n + 1 numbers in 0, 1, 2, ..., n) is T[0] + r*(n + 1 - 1) = 2 + 2*n = 2*(n + 1).
No guessing required, just recognize it as an arithmetic progression.
Each time n decreases by one, 2 is added. This gives a variable term of 2n. Since T(0) is fixed at 2, this gives a constant term of 2. Adding them together gives 2n + 2, or 2(n + 1).
I'd solve it as follows:
Assume that T(n) = a*n + b for some a and b.
T(0) = 2. So a * 0 + b = 2, thus b = 2.
T(n) = T(n-1) + 2, so
a * n + b = (a * (n-1) + b) + 2 consequently
a * n + b = a * n - a + b + 2 and
0 = - a + 2, thus a = 2.
So we have T(n) = 2 * n + 2 = 2 (n+1).
This one is pretty straightforward to solve by hand as the other answers point out, but in case it's ever useful, Mathematica is pretty good solving recurrence relations like this.
Evaluating
RSolve[{T[0] == 2, T[n] == T[n-1] + 2}, T[n], n]
returns
{{T[n] -> 2 (1 + n)}}
It can, for example, find the closed form of the nth Fibonacci number as well:
RSolve[{F[1] == 1, F[2] == 1, F[n] == F[n-1] + F[n-2]}, F[n], n] //FunctionExpand
returns
{{F[n] -> (((1 + Sqrt[5])/2)^n - (2/(1 + Sqrt[5]))^n*Cos[n*Pi])/Sqrt[5]}}