[ACE] Recurrence Relation

Recurrence Relation

Acknowledgement

The following content is summarized by G52ACE class and MIT open source course(Introduction to Algorithms).

Main Body

We don’t care the official definition of recurrence relation, but we need to know:
Suppose that the runtime of a program is T(n), then a recurrence relation will express T(n) in terms of its values at other (smaller) values of n.
Example:
Suppose the runtime of merge-sort of an array of n integers is T(n). Then T(n) = 2 T(n/2) + b + a n:
enter image description here
• “2 T(n/2)” is due to having to sort the two sub-arrays each of size n/2
• “b” is the cost of doing the split
• “a n” is the cost of doing the merge (and any copying to/from the workspace, etc.)
Also, what is our base case? ===> T(1) = 1
Thus, how can we solve this recurrence relation?

Solving Recurrence using induction(introduced by G52ACE course)

General pattern
1. Starting from the base case, use the recurrence to work out many cases, by directly substituting and working upwards in values of n
2. Inspect the results, look for a pattern and make a hypothesis for the general results
3. Attempt to prove the hypothesis – typically using some form of induction
enter image description here
If we regard 2^k to n
T(n) = n + n * log n * a +(n-1) * b = O(n log n)
Often then extract the large n behavior using big-Oh family

Solving Recurrence using Recurrence Tree(introduced by Book)

First we should cast the formula to this

Put the θ(n) to cn then we can draw a recursion tree
What we need to do is to figure out how many leaves are in the lowest level, what is the height of the tree and what is the total of each level

Because the height is logn, the total of each level is cn for internal leaves and θ(n) for external leaves.
Thus, the total is (cn) * lgn + θ(n), we ignore θ(n), it is θ(n log n)

There is not a usual way to solving recurrence, so in terms of different question, we should use different method, here are two other important method– Substitution method and Master Theorem

Substitution method

First, guess the form of solution
Second, verify by induction
Third, solve the constant
It is somehow similar to the method we learned in the class and i need to further check with my tutor.
But i’d like to introduce this method here:
Here is a example:

In my opinion, it is similar to the method we learned before but it is not so precise. It just can work out O(n^3) in this example but it is not Θ(n^3) because it is O(n^2) in this example too. So it just can narrow your assumption down to some range but not exact complexity.

The master theorem(Both introduced by lecture and book)

MT applies to
T(n) = a T(n/b) + f(n) where a>=1, b>1, f(n) is asymptotically positive
  • what does it mean by “asymptotically positive”?
  • It is when n is bigger enough, f(n) should be positive===> f(n) > 0 for n>= 0;
Results are split into 3 cases, according to comparing the growth rate of f(n) to n^ (log b a)
There are a lot of similar format of MT but i pick the one on Wikipedia
- Case 1:
enter image description here
- Case 2:
enter image description here
- Case 3:
enter image description here
Keep in mind that there are a lot of situation that MT can not applies to.
Here are two examples solving actually question in MT(Binary Search and Merge Sort)
enter image description here

So how to prove MT? why f(n) should be compared to n ^ (log b a)?

The prove in Introduction to Algorithm is using tree, but i tried to prove it using mathematical expressions:
enter image description here
Here we can see we work out n ^ (log b a), and each case, what we do is to compare it to see which is the dominance of the expression
Okay, it is just my opinion, let’s see how it should be proved officially
enter image description here
We are using a recursive tree.
If we add the total work of the tree, what we will get?!
Cool! Its what i proved above in the last equation!
n ^ (log b a) is the sum of the leaves at the bottom
The above sum of each level will be showed as the summation formula!
It is the same as induction by mathematic
In Case 1: the weight or dominance is on the external leaves than the roots(and internal leaves), so we should use the sum of leaves which is n ^ (log b a)
In Case 2: the weight is similar between them so we should add them up
In Case 3: the weight is mostly on the internal leaves so we should ignore external leaves and it should be O( f(n) )

评论

发表评论

此博客中的热门博文

[MLE] Linear Classification

图片
Linear Classification Discriminant functions Least squares for classification Fisher’s linear discriminant K-Nearest Neighbour(KNN) Classification What’s decision boundaries/linearly separable? The goal in classification is to take an input vector x and to assign it to one of K discrete classes \(C_k\) where k = 1,…,K. In the most common scenario, the classes are taken to be disjoint, so that each input is assigned to one and only one class. The input space is thereby divided into decision regions whose boundaries are called decision boundaries or decision surfaces. In this blog, we consider linear models for classification , by which we mean that the decision surfaces are linear functions of the input vector x and hence are defined by (D-1) -dimensional hyperplanes within the D-dimensional input space. For example: in 3D input space, the decision boundary will be a 2D surface What is Linear Separability Linearly separable data: Datasets whose classes can be separated by
阅读全文

[AIM] MetaHeuristics

图片
Day2 Outline Metaheuristics -definition Single point based search method(simple stochastic local search) Iterated local search Scheduling Tabu Search Metaheuristic what is a metaheuristic? A metaheuristic is a high-level problem independent algorithmic framework that provides a set of guidelines or strategies to develop heuristic optimization algorithms. what is the difference between heuristic and metaheuristic? Heuristic : Heuristics are problem-dependent techniques. As such, they usually are adapted to the problem at hand and they try to take full advantage of the particularities of this problem. However, because they are often too greedy, they usually get trapped in a local optimum and thus fail, in general, to obtain the global optimum solution. Metaheuristic : Meta-heuristics, on the other hand, are problem-independent techniques . As such, they do not take advantage of any specificity of the problem and, therefore, can be used as black boxes . In general, t
阅读全文

[CS231] Neural Networks

图片
Neural Network Table of Contents: Quick intro without brain analogies Modeling one neuron Biological motivation and connections Single neuron as a linear classifier Commonly used activation functions Neural Network architectures Layer-wise organization Summary Additional references Quick intro It is possible to introduce neural networks without appealing to brain analogies. In the section on linear classification we computed scores for different visual categories given the image using the formula \( s = W x \), where \(W\) was a matrix and \(x\) was an input column vector containing all pixel data of the image. In the case of CIFAR-10, \(x\) is a [3072x1] column vector, and \(W\) is a [10x3072] matrix, so that the output scores is a vector of 10 class scores. An example neural network would instead compute \( s = W_2 \max(0, W_1 x) \). Here, \(W_1\) could be, for example, a [100x3072] matrix transforming the image into a 100-dimensional intermediate vector. The funct
阅读全文