#Asymptomatically tight solution

26 messages · Page 1 of 1 (latest)

karmic groveBOT
torpid notch
#

Is this well-defined? What's the domain?

coarse stratus
torpid notch
#

The domain is what values of n you're allowed to plug in. Can you ever find T(5)?

coarse stratus
#

T(5) would just be T(sqrt(5)) + log_2 5 which would continually drop to sqrt(n) until n is between 1 and 4 which would then give a result of 1

#

i think

#

thats what im interpreting the question as

#

and i am asked to Give asymptotically tight solutions to the following recurrences.

#

I thought about maybe trying to draw a recursion tree to solve it, but tbh im pretty fuzzy on them so idrk how I would go abt that if thats the way to solve something like this

torpid notch
#

Is n a natural number or a real number?

#

I suppose it shouldn't really matter, but you probably have different tools to work with in different classes

coarse stratus
#

pretty sure n is just any number between 1-4 so it can be like 1.2, 3, 2.7, etc

torpid notch
#

I would just look at the sequence 2,4,16,256,256^2,...

#

Does it say somewhere that log is supposed to be log base 2?

coarse stratus
#

we automatically assume log base 2 in my class

torpid notch
#

What class is this?

coarse stratus
#

analysis of algorithms

dense quartz
#

here's what I found

#

by iterating the process of plugging sqrt(n) in the definition, you can essentially end up with T(n) = T(sqrt(n))+lg(n) = T(n^(1/4)) + 3/2lg(n) = T(n^(1/8)) + 7/8lg(n) etc...

#

so you see we're descending towards 1 inside T (which I assume to represent a complexity), and approaching 2lg(n) on the other term

#

so that motivates proving the following statement : for all integers n, T(n) <= 1+2lg(n)

#

and you can prove that by induction very easily

#

but what this gives you for this algorithm is that the complexity is O(log(n)) which is pretty good!

#

though I'd be pretty curious as to what this algorithm does with such a strange recursion expression (if it even exists outside of this exercise lmao)

#

@coarse stratus ^

coarse stratus