Asymptotic Analysis : Big-Oh Notation
Been learning about asymptotic analysis.
Definition: Asymptotic analysis is the study of the growth of an algorithm’s running time or space requirement as a function of the input size n, for large values of n, while ignoring constant factors and lower-order terms.
In other words, we analyze how the resource consumption of an algorithm scales as the problem size becomes large.
Time Complexity != Time taken by the algorithm to run
Time complexity is a function f(n) that represents the number of elementary operations performed by an algorithm as a function of input size n ∈ ℕ.
T(n) = number of primitive steps executed for input size n
Asymptotic analysis is independent of
- Hardware specifications
- Programming language
- Operating system and runtime environment
- Input distribution (for a fixed n)
Big-Oh Notation (O-notation)
Definition: Let f(n) and g(n) be functions mapping positive integers to positive real numbers. We say that g(n) is O(f(n)) (read “g of n is big-oh of f of n”) if there exist positive constants c ∈ ℝ⁺ and n₀ ∈ ℕ such that:
g(n) ≤ c·f(n) ∀ n ≥ n₀
The function f(n) is called an asymptotic upper bound for g(n). Intuitively, this means that g(n) grows no faster than f(n), up to a constant multiplicative factor, for sufficiently large values of n.
Visually , if we plot both g(n) and f(n) and can find some point n₀ in the input axis beyond which c·f(n) always stays above g(n). Then g(n) is O(f(n)). i.e We use Big-Oh if we are concerned with the worst case performance of the algorithm
eg:
Say f(n) = 3n² + 5n + 10 , g(n) = n². g(n) = O(f(n)) & f(n) = O(g(n))
If we analyse this
We need to find constants c ∈ ℝ⁺ and n₀ ∈ ℕ such that:
f(n) ≤ c·g(n) ∀ n ≥ n₀
Substituting: 3n² + 5n + 10 ≤ c·n²
For n ≥ 1: – 5n ≤ 5n² – 10 ≤ 10n²
Therefore: 3n² + 5n + 10 ≤ 3n² + 5n² + 10n² = 18n²
Conclusion: Choose c = 18 and n₀ = 1. Then f(n) ≤ 18·g(n) ∀ n ≥ 1.
Hence, f(n) = O(g(n)) or more specifically, f(n) = O(n²).
Again (n) ≤ c·f(n) ∀ n ≥ n₀
Substituting: n² ≤ c(3n² + 5n + 10)
Since f(n) = 3n² + 5n + 10 > 3n² for all n ≥ 1, we have:
n² ≤ 3n² + 5n + 10
This is clearly true for all n ≥ 1.
Conclusion: Choose c = 1 and n₀ = 1. Then g(n) ≤ 1·f(n) ∀ n ≥ 1.
Hence, g(n) = O(f(n)).
Big-Oh Represents a Set of Functions
Say , some T(n) = n³ then
3n² + 5n + 10 = O(T(n))..................i
5n+7 = O(T(n)).........................ii
log(n) = O(T(n))_____iii
if we analytically approach eq(i),eq(ii),eq(iii). All 3 are True But eq(i) != eq(ii) != eq(iii)
When we write f(n) = O(g(n)), the ”=” is not a true equality.
More Precisely: O(g(n)) represents an infinite set of all functions that grow no faster than g(n). So f(n) = O(g(n)) really means f(n) ∈ O(g(n))
Example: O(n²) is the set containing functions like: – n² – 3n² + 5n + 10 – n² + 100n – 50n + 1000 – 5 (constant) – log n
All these functions belong to the set O(n²) because they all grow no faster than n².
When we write O(n²), we're saying that as the input size n becomes very large, the running time grows at most proportionally to n². The constant factors and lower-order terms become negligible in comparison to the dominant term, which is why we drop them in Big-Oh notation. For instance, an algorithm that takes 3n² + 5n + 10 operations is simply O(n²) because the n² term dominates as n approaches infinity.
i.e: We simply ignore the lower order and constant terms when were trying to find the time complexity in Big-Oh
Some Commonly Seen Cases
O(1) – Constant Time: These operations take the same amount of time regardless of input size. Accessing an array element by index, performing arithmetic operations, and returning a value are all O(1) operations. No matter if our array has 10 elements or 10 million elements, accessing the element at index 5 takes the same time.
O(log n) – Logarithmic Time: Algorithms that repeatedly divide the problem space in half exhibit logarithmic complexity. Binary search is the classic example, where we eliminate half of the remaining elements with each comparison. As n doubles, the running time increases by only a constant amount.
O(n) – Linear Time: When we must examine every element exactly once, we have linear complexity. A simple loop that processes each element in an array demonstrates O(n) behavior. If we double the input size, the running time doubles as well.
O(n log n) – Linearithmic Time: Efficient sorting algorithms like merge sort and heap sort operate in O(n log n) time. This complexity arises when we perform a logarithmic number of linear operations or divide-and-conquer with linear merging.
O(n²) – Quadratic Time: Nested loops where each loop runs n times typically result in quadratic complexity. Simple sorting algorithms like bubble sort and selection sort fall into this category. Doubling the input size quadruples the running time.
O(n³) – Cubic Time: Triple-nested loops often produce cubic complexity. Matrix multiplication using the naive algorithm is a prime example.
O(2ⁿ) – Exponential Time: The recursive Fibonacci function is a classic example, where each call branches into two more calls. The running time doubles with each increment of n, making it impractical for even moderately sized inputs.
O(n!) – Factorial Time: Generating all permutations of n elements produces factorial complexity. For n = 10, that's 3,628,800 operations. Yes calculated that in my head,absolutely did not google.
O(1) < O(log n) < O(n) < O(n log n) < O(n²) < O(n³) < O(2ⁿ) < O(n!)
Analyzing Some Algorithms
Example 1: Array Sum
def array_sum(arr):
total = 0
for element in arr:
total += element
return total
The loop runs once per element. Each iteration does constant work. Time complexity: O(n).
Example 2: Finding Maximum Element
def find_max(arr):
if len(arr) == 0:
return None
max_val = arr[0]
for i in range(1, len(arr)):
if arr[i] > max_val:
max_val = arr[i]
return max_val
Single pass through the array with constant-time operations per iteration. Time complexity: O(n).
Example 3: Bubble Sort
def bubble_sort(arr):
n = len(arr)
for i in range(n):
for j in range(n - i - 1):
if arr[j] > arr[j + 1]:
arr[j], arr[j + 1] = arr[j + 1], arr[j]
Nested loops result in roughly n² comparisons. Time complexity: O(n²).
Analyzing a Slightly More Complex Program
Example 4: Binary Search
def binary_search(arr, target):
left = 0
right = len(arr) - 1
while left <= right:
mid = (left + right) // 2
if arr[mid] == target:
return mid
elif arr[mid] < target:
left = mid + 1
else:
right = mid - 1
return -1
Initially, the search space has size n. After each iteration: – 1st iteration: n elements – 2nd iteration: n/2 elements – 3rd iteration: n/4 elements – kth iteration: n/2^(k-1) elements
The loop continues until the search space is reduced to 1 element or becomes empty. We need to find k such that:
n/2^(k-1) ≤ 1
Solving for k: – n ≤ 2^(k-1)
log₂(n) ≤ k – 1
k ≥ log₂(n) + 1
Therefore, the loop executes at most ⌈log₂(n)⌉ + 1 times (ceiling of log₂(n) plus 1).
Since each iteration performs O(1) operations (comparisons, assignments, arithmetic), the total time complexity is : T(n) = O(1) + O(log n)·O(1) + O(1) = O(log n) .
i.e T(n) = O(log(n)) for Binary Search
Conclusion:
- Big-Oh gives an upper bound on how an algorithm scales as input size increases
- We analyse for n → ∞
- Constant terms and lower order terms are ignored
print("Titas,Signing Out!")