DSA Patterns — One‑Page Cheat Sheet
Compact list of common problem‑solving patterns, when to use them, complexity notes, and
1‑line problem prompts to practice.
1. Two Pointers
What: Use two indices moving from ends or same direction on sorted/linear structures. When: Sorted
arrays, pair-sum, removing duplicates, merging lists. Complexity: O(n) / O(1) extra. Try: "Find pair with
given sum (sorted)."
2. Fast & Slow Pointers
What: Two pointers at different speeds to find cycles, midpoints, or k‑th from end. When: Linked lists, cycle
detection, palindrome checks. Try: "Detect cycle in linked list (Floyd)."
3. Sliding Window (Fixed & Variable)
What: Maintain window [i..j] that satisfies a condition; expand/contract. When: Subarray/substring sums,
longest/shortest subarray with property. Complexity: O(n) typical. Try: "Longest substring without
repeating chars."
4. Prefix Sum / Cumulative Sum
What: Precompute running sums for O(1) range queries. When: Range sums, subarray equals k (with
hashmap). Try: "Subarray sum equals K."
5. Binary Search (Index & Answer)
What: Classic on sorted arrays or binary search on answer space. When: Search, minimizing/maximizing
parameter, monotonic predicate. Complexity: O(log n) × check. Try: "Aggressive cows / split arrays to
minimize largest sum."
6. Sorting + Two-Pointer / Sweep
What: Sort and then use linear pass or pointers. When: Interval merging, pair problems, scheduling. Try:
"Merge intervals."
1
7. Interval Patterns (Merge, Cover, Scheduling)
What: Sort by start/end; greedy choices (earliest finish). When: Meeting rooms, max non-overlapping
intervals. Try: "Minimum number of conference rooms."
8. Hashing (Maps/Sets)
What: Use hash tables for O(1) lookups and counting. When: Frequency, complements, deduplication,
sliding window with counts. Try: "Two-sum (unsorted)."
9. Monotonic Stack / Queue
What: Maintain monotone structure to answer next/previous greater/smaller queries. When: Sliding-
window max (deque), histogram largest rectangle, stock span. Try: "Daily temperatures (next warmer day)."
10. Heaps (Priority Queues)
What: Keep top‑k or extract min/max dynamically. When: Merge k sorted lists, streaming median (two
heaps), Dijkstra. Try: "Find k smallest elements."
11. Graph Traversals: BFS / DFS
What: BFS for shortest unweighted path / level order; DFS for components, backtracking, topological sort
helper. When: Connectivity, shortest path in unweighted graphs, tree problems. Try: "Number of connected
components."
12. Topological Sort / DAG DP
What: Order DAG nodes; perform DP along topological order. When: Scheduling with dependencies,
longest path in DAG. Try: "Course schedule / finish order."
13. Shortest Paths (Dijkstra / Bellman‑Ford)
What: Dijkstra for nonnegative weights; Bellman‑Ford for negative edges. When: Weighted shortest path
problems. Try: "Single-source shortest path with nonnegative weights."
14. Union‑Find (Disjoint Set)
What: Merge sets, detect cycles, count components (use union by size/rank + path compression). When:
Connectivity, Kruskal MST, friend circles. Try: "Number of connected components in graph."
2
15. Backtracking / Recursion
What: Explore choices with recursion + pruning. When: Permutations, subsets, combinatorics, constraint
satisfaction. Try: "Generate all subsets."
16. Dynamic Programming (DP) Patterns
Categories & Hints: - 1D DP (linear states): fib, climb stairs, dp[i] depends on previous few. - 2D DP (grid):
unique paths, edit distance. - Knapsack / DP on weights: capacity-based. - DP on sequences (LIS / LCS):
patience sorting for LIS O(n log n). - Tree DP: postorder aggregation. Try: "Longest increasing
subsequence."
17. Greedy Algorithms
What: Make local optimal choices aiming for global optimum; require proof. When: Interval scheduling,
Huffman coding, coin change (canonical systems). Try: "Activity selection (max non-overlapping intervals)."
18. Divide & Conquer (Merge Sort style)
What: Split, solve halves, merge answers. When: Sorting, inversion count, closest pair of points. Try: "Count
inversions using merge sort."
19. Bit Manipulation
What: Use bitwise ops for parity, masks, subset iteration. When: XOR problems, single-number, bit DP,
subset generation with masks. Try: "Single number (XOR)."
20. Trie (Prefix Tree)
What: Tree of characters for fast prefix queries. When: Autocomplete, longest common prefix, word search.
Try: "Implement prefix search / word dictionary."
21. Rolling Hash / Rabin‑Karp
What: Hash substrings for pattern matching with O(1) rolling update. When: String search, duplicate
substring detection. Try: "Find repeated DNA sequences."
22. Segment Tree / Fenwick (BIT)
What: Range queries + point/range updates; Fenwick simpler for sums. When: Dynamic range queries,
order statistics, offline queries. Try: "Range sum with updates."
3
23. Square‑Root Decomposition / Mo’s Algorithm
What: Block decomposition or offline reordering for query optimization. When: Many range queries with
offline processing. Try: "Answer offline distinct‑count queries."
24. Matrix Manipulation (DFS on grid)
What: 2D DFS/BFS, rotations, spirals, dynamic programming on grid. When: Islands count, path counts,
rotate image. Try: "Number of islands."
25. Misc / Math Patterns
• GCD / LCM for number theory problems.
• Sieve / Primes for primality ranges.
• Combinatorics (nCr / DP) for counting.
• Probability / Expectation when stochastic.
How to Use This Sheet
1. Identify structure: array? string? graph? tree? matrix? numbers?
2. Match pattern: sorted → two pointers/binary search; dynamic updates → segment tree/heap.
3. Check constraints: n ≤ 10⁵ → O(n log n) or better; n ≤ 20 → consider bitmask DP/backtracking.
4. Combine patterns: e.g., sliding window + hashing for anagram search.
Quick Practice Plan (30 problems)
• 6 Array/string basics: two pointers, sliding window, prefix sum.
• 6 Trees/graphs: DFS/BFS, union‑find, Dijkstra.
• 6 DP/backtracking: LIS, knapsack, subsets/permutations.
• 6 Heaps/monotonic structures: k‑largest, sliding max, stack problems.
• 6 Advanced: segment tree, trie, rolling hash, Mo’s algorithm.
Need a printable PDF, short example code templates, or curated problem list per pattern? Click the canvas title to
view and ask me to export.
