Module 1 · SBI Clerk Reasoning Ability

Puzzles & Seating

Updated 17 Nov 2026 · 5,716 words · illustrated
Updated 21 Nov 2026
Linear, circular, floor-based puzzles with logical constraints.
Linear Arrangement · Circular · Floor Puzzles · Comparison
🎯 Take Mock Test
1234ABCDLinear ArrangementDirection: Left ← → RightCircular ArrangementAdjacent: A-B, B-C, C-D, D-AOpposite: A-C, B-DExam Hook: 'Facing Center → Left = Clockwise'vlymbooq
Seating Arrangement Types: Linear vs Circular — Compares linear and circular seating layouts with key directional rules for puzzle solving. · Open the full diagram →

Learning Objectives

  • Understand the fundamental concepts of Puzzles & Seating
  • Apply key formulas and techniques to solve problems
  • Practice with exam-level questions to build speed and accuracy

What are Puzzles & Seating Arrangements?

Puzzles and seating arrangement questions are a cornerstone of banking exams, and the exam. They test your ability to organize information, identify relationships, and deduce logical conclusions from given clues. These questions typically involve arranging people or objects in a specific order — linear rows, circular tables, or rectangular arrangements — and answering questions about their positions relative to each other.

Types of Seating Arrangements

1. Linear Arrangement: People sit in a straight row facing north or south. Clues describe who sits next to whom, who is at the ends, who is between two persons, etc. Example: A, B, C, D, E sit in a row. C is at one end. B sits second to the right of A, etc.

2. Circular Arrangement: People sit around a circular table, typically facing the centre. Clues involve immediate neighbours, opposite persons, and positions relative to each other (to the left/right). In circular arrangements, 'to the left' means anti-clockwise and 'to the right' means clockwise when facing the centre.

3. Rectangular/Square Arrangement: People sit around a rectangular table, usually with persons on each side. Some face the centre, others face outward. This adds a direction component to the puzzle.

4. Floor-Based Puzzles: People live on different floors of a building (typically 5-8 floors). Clues describe who lives above/below whom, which floor has a specific person, etc. Sometimes combined with additional parameters like professions or favourite colours.

5. Complex Puzzles with Multiple Parameters: Combine two or more variables — e.g., 8 persons in a circular arrangement with different professions and cities. These require grid or table method to solve.

Key Strategies

Always start by identifying fixed positions. Use a diagram or grid. Note down negative clues (who cannot be placed where). Update your diagram step by step as you process each clue. When multiple possibilities arise, use the branching method — consider both possibilities and eliminate when contradictions appear.

Solved Examples

Example 1 (Linear Arrangement)
Five persons — P, Q, R, S, T — sit in a row facing north. R sits exactly in the middle. P sits second to the left of R. S sits to the immediate left of Q. T is not at either end. Who sits at the extreme left end?

a) P b) Q c) S d) T
Solution: a) P. Positions: 1 _ 2 _ 3 _ 4 _ 5. R is at position 3 (middle). P is second to left of R → P at position 1. Therefore P is at extreme left. S sits immediate left of Q → positions 2 and 4 or 4 and 5. T is not at ends → T at position 4. So S at 2, Q at 5, T at 4. Final: P, S, R, T, Q.
Example 2 (Circular Arrangement)
Six friends — A, B, C, D, E, F — sit around a circular table facing the centre. A sits third to the left of C. D sits to the immediate right of A. B sits second to the right of E. Who sits opposite F?

a) A b) B c) C d) D
Solution: a) A. Place C at top. A is third left of C → A is 3 positions anti-clockwise from C. D is immediate right of A. B is second right of E → place E, then B two positions clockwise. The remaining position is F. The arrangement (clockwise): C, E, B, A, D, F. A sits opposite C. F sits opposite A.
Example 3 (Floor-Based Puzzle)
Seven persons live in a 7-storey building (G=ground, 1 to 6). M lives on an even-numbered floor. Two persons live between M and N. O lives immediately above P. Q lives on floor 3. Three persons live between Q and R. Who lives on the topmost floor?

a) M b) N c) O d) R
Solution: b) N. Q on floor 3. Three persons between Q and R → R on floor 7 (top) or floor -1 (invalid), so R = floor 7.

Shortcut Techniques

  • Draw a rough diagram — it saves time over mental juggling of positions.
  • For circular arrangements, fix one person first and arrange everyone relative to them.
  • Use tick (✓) and cross (✗) marks in a grid table for multi-parameter puzzles.
  • Read all clues first — sometimes a clue at the end gives a definite position you missed.
  • Negative clues (e.g., "A does not sit next to C") are as important as positive ones.
Pro Tip
In banking exams, puzzles-seating carry 5-10 marks in the reasoning section. Focus on speed — practice with a timer. Most puzzles can be solved in 3-5 minutes if you avoid drawing and redrawing the diagram multiple times. Use a pencil for easy corrections.

Advanced Problem Types

Competitive exams feature advanced variations combining multiple reasoning concepts. Master these to score high.

Multi-Step Problems

Some questions require applying multiple reasoning techniques sequentially. Break problems into clear steps.

Common Traps

  • Assuming a pattern continues without checking all elements
  • Misinterpreting negative statements in syllogisms
  • Forgetting wrapping in alphabet coding (Z to A transition)
  • Confusing directions (East vs West)
  • Overlooking either-or cases in syllogisms
  • Not considering all possible arrangements in puzzles
Pro Tip
For puzzles, start with the most concrete clue and build around it. Draw diagrams for seating arrangements and family trees for blood relations.

Multi-Parameter Grid Puzzles: Integrating Seating with Attributes

When a puzzle supplies more than one attribute for each person — such as profession, favourite colour, vehicle, and floor — the most reliable solving framework is a two‑dimensional grid. Rows represent the entities (persons, floors, seats) and columns represent the attributes. Begin by listing every distinct value that appears in the clues for each attribute. For a seven‑person circular arrangement with three attributes, the grid will have seven rows and four columns (person + three attributes).

Clues that give a direct match (e.g., “The engineer sits opposite the person who likes red”) are entered as a tick in the intersecting cell and crosses in all other cells of that row and column for the same attribute. Negative clues (“A does not drive a sedan”) are recorded as crosses. The power of the grid lies in the propagation of constraints: once a row has a single empty cell for an attribute, that value is forced. Similarly, if a column has only one possible row for a value, the placement is fixed.

Circular adjacency clues (“B sits second to the left of the doctor”) are translated into relative seat indices. Assign provisional seat numbers 1‑N clockwise. The clue becomes an equation: seat(B) = seat(doctor) – 2 (mod N). Keep a separate seat‑index column in the grid; when a person’s seat becomes certain, propagate the index to all attributes of that person. This dual‑track (attribute grid + seat index) prevents the common error of solving attributes in isolation and then trying to fit them into seats later.

Iterate through the clue list repeatedly. After each full pass, scan the grid for rows or columns with a single candidate — these are “singletons” and must be committed. Commitment may create new singletons, so the process is recursive. If a contradiction appears (a row requires two different values for the same attribute), backtrack to the last ambiguous decision point and test the alternative branch. In practice, most competitive‑exam puzzles are designed so that a single linear pass with careful bookkeeping yields a unique solution without deep backtracking.

Time‑saving tip: use a pencil‑and‑paper grid of 1 cm squares. Write attribute abbreviations (Eng, Doc, Red, Blu, Sed, SUV) in the header row. Tick marks are small ✓; crosses are tiny ✗. This visual density lets the eye spot singletons instantly, which is faster than mental juggling.

Conditional Either‑Or Clues in Circular Arrangements

Either‑or clues appear in two flavours: exclusive (exactly one of the two statements is true) and inclusive (at least one is true, possibly both). The phrasing “Either A sits opposite B or C sits opposite D, but not both” signals exclusivity. “Either A sits opposite B or C sits opposite D” without “but not both” is inclusive. Misclassifying the type is the single biggest source of wrong arrangements.

To handle an exclusive either‑or, create two parallel sub‑diagrams (branch A and branch B). In branch A, enforce the first statement as a hard constraint and add the negation of the second statement as an additional clue. In branch B, do the opposite. Solve each branch independently using the standard circular‑arrangement method (fix one person, place others relative). If a branch leads to a contradiction (e.g., two persons assigned to the same seat), discard it. The surviving branch yields the unique arrangement.

For inclusive either‑or, the logic is softer: at least one statement holds. The practical approach is to first assume both statements are true and test for consistency. If consistent, the puzzle may have multiple valid arrangements; the question will then ask for a “must be true” or “could be true” answer, which is evaluated across all consistent arrangements. If the double‑true assumption creates a contradiction, then exactly one statement is true, reducing the inclusive clue to an exclusive one.

Conditional clues (“If A sits next to B, then C sits opposite D”) are treated as logical implications. Convert to contrapositive: “If C does not sit opposite D, then A does not sit next to B.” Both the original and contrapositive are added to the clue pool. During solving, whenever the antecedent becomes certain, the consequent is forced. Whenever the consequent becomes impossible, the antecedent is forced false. This bidirectional propagation dramatically reduces the search space.

A worked illustration: eight persons around a circle, facing centre. Clue set includes “Either P sits third to the left of Q or R sits second to the right of S, but not both” and “If T sits opposite U, then V sits immediate left of W.” Fix P at seat 1. Branch 1: place Q at seat 4 (third left). Add “R is not second right of S”. Continue placing remaining persons using other clues. If a seat conflict arises, abandon branch 1. Branch 2: place R second right of S, add “Q is not third left of P”. The branch that survives gives the final circle. The conditional clue is monitored throughout; once T‑U opposition is decided, V‑W adjacency is locked.

Worked Example: Exclusive Either‑Or with Conditional
Six persons — A, B, C, D, E, F — sit around a circular table facing the centre. Clues:
1. Either A sits opposite C or B sits opposite D, but not both.
2. If E sits opposite F, then A sits immediate left of B.
3. C sits second to the left of E.
4. F does not sit next to A.
Question: Who sits opposite B?
Step 1: Fix E at seat 1 (arbitrary). Clue 3 → C is second left of E → seat 5 (clockwise numbering).
Step 2: Branch X (A opposite C). Opposite of seat 5 is seat 2 → A = seat 2. Add “B not opposite D”.
Step 3: Remaining seats 3,4,6 for B,D,F. Clue 4: F not next to A (seats 1 and 3). Seat 1 is E, so F ≠ seat 3 → F ∈ {4,6}.
Step 4: Test conditional clue 2. In Branch X, E (seat 1) opposite F would require F = seat 4 (opposite of 1). If F = seat 4, then antecedent true → consequent forces A immediate left of B. Immediate left of B (clockwise) means seat of A = seat(B) – 1. A is seat 2, so B must be seat 3. Then D gets seat 6. Check “B not opposite D”: opposite of seat 3 is seat 6 → B opposite D, violates branch condition. Hence F ≠ seat 4. Therefore F = seat 6.
Step 5: With F = seat 6, antecedent of clue 2 false (E not opposite F), so no further restriction. Remaining seats 3,4 for B,D. No direct clue, but “B not opposite D” eliminates B=3,D=4 (opposite pair) and B=4,D=3 (also opposite). Thus Branch X yields no valid assignment → discard.
Step 6: Branch Y (B opposite D). Opposite pairs in 6‑seat circle: (1,4), (2,5), (3,6). Seats 1 and 5 are taken (E, C). So B‑D must be (2,5) or (3,6). Seat 5 is C, so B‑D = (3,6). Two sub‑cases: B=3,D=6 or B=6,D=3.
Sub‑case Y1: B=3, D=6. Remaining seats 2,4 for A,F. Clue 4: F not next to A. Adjacent pairs: (1,2), (2,3), (3,4), (4,5), (5,6), (6,1). Seat 1=E, seat 3=B, seat 5=C, seat 6=D. If A=2,F=4 → A adjacent to E (allowed), F adjacent to B (allowed), A and F not adjacent (2 and 4 separated by 3) → condition satisfied. Check conditional clue 2: E opposite F? Opposite of 1 is 4 → F=4 makes antecedent true → then A immediate left of B. Immediate left of B (seat 3) is seat 2 → A=2, which holds. All clues satisfied. Arrangement (clockwise from seat 1): E(1), A(2), B(3), F(4), C(5), D(6). Opposite of B (seat 3) is seat 6 → D.
Sub‑case Y2: B=6, D=3. Remaining seats 2,4 for A,F. Clue 4: F not next to A. Test A=2,F=4 → same adjacency check passes. Conditional clue 2: E opposite F? F=4 → antecedent true → A immediate left of B. Immediate left of B (seat 6) is seat 5 → A would need to be seat 5, but seat 5 is C. Contradiction. So Y2 invalid.
Unique solution: D sits opposite B. Answer: D.
Exam‑Day Branching Discipline
When an exclusive either‑or appears, draw two tiny circles labelled “Branch 1” and “Branch 2” in the margin. Solve the first branch completely before touching the second. If the first branch yields a full valid arrangement, you can often answer “must be true” questions without solving the second branch — but verify the question type first. For “could be true” you must check both branches. Keep branch diagrams compact: reuse the same seat numbers, only change the forced placements.

Floor Puzzles with Multiple Variables: Profession, Colour, and Vehicle

Floor‑based puzzles frequently attach three or more independent variables to each resident — profession, favourite colour, vehicle type, and sometimes a pet. The solving skeleton is a vertical table: floors as rows (ground = 0 or 1 depending on convention), columns for each variable. Begin by noting the total number of floors and the number of residents; they are usually equal, but occasionally a floor may be vacant — treat vacancy as an extra “resident” with a “None” value for every variable.

Direct clues (“The doctor lives on floor 3”) are entered immediately. Relative clues (“Two floors separate the engineer and the artist”) translate into a difference constraint: |floor(engineer) – floor(artist)| = 3 (since two floors between means a gap of three). Adjacency clues (“The red‑car owner lives immediately above the blue‑car owner”) become floor(red) = floor(blue) + 1. Conditional clues (“If the teacher drives a sedan, then the lawyer lives on the top floor”) are handled exactly as in circular puzzles: add both the implication and its contrapositive to the clue pool.

Because each variable set (professions, colours, vehicles) is a permutation of the residents, the grid obeys Latin‑square constraints: each profession appears exactly once, each colour exactly once, etc. This permits powerful cross‑elimination. For instance, if the grid shows that only floor 2 and floor 5 can host the engineer, and the artist must be three floors apart, then the artist can only occupy the complementary floors. Mark these as candidate pairs and propagate.

A systematic solving loop: (1) Apply all direct and negative clues. (2) For each relative clue, list all floor pairs satisfying the numeric relation given current candidates. (3) Intersect candidate sets across clues — a floor that cannot satisfy any clue for a variable is eliminated. (4) When a variable on a floor reduces to a single candidate, commit it and purge that candidate from all other floors for the same variable. (5) Repeat until stable. If multiple solutions persist, the question will be of “could be true” type; enumerate the remaining possibilities.

Speed tip: use a compact notation in the margin — e.g., “Eng: {2,5}” — and update it in place. Avoid rewriting the whole table after each deduction; instead, strike through eliminated candidates with a thin line. This keeps the visual field clean and the mind focused on the remaining degrees of freedom.

Linear Arrangements with Facing Directions and Gap Constraints

Linear puzzles often add a facing direction (north/south or left/right) for each person, and may include empty seats (gaps). The core representation is a one‑dimensional array of positions 1…N, where N = persons + gaps. Each position holds either a person or a blank. Facing direction is an attribute attached to the person, not the seat.

Clues such as “A faces north and sits third to the left of B who faces south” are parsed as: position(A) = position(B) – 3, and direction(A) = north, direction(B) = south. “Third to the left” assumes a common reference orientation — usually the observer looks at the line from the left end, so “left” means decreasing index. Confirm the convention from the first clue that uses “left/right”; if ambiguous, adopt the standard that positions increase left‑to‑right.

Gap clues (“There are exactly two empty seats between C and D”) translate to |pos(C) – pos(D)| = 3 (two empties plus the two persons). If the total number of seats is not given, deduce it from the maximum span implied by the clues plus the number of persons. For example, five persons with a maximum required gap of three empties between the farthest pair implies at least 5 + 3 = 8 seats.

Conditional facing clues (“If E faces north then F sits at an extreme end”) are handled by adding both the implication and its contrapositive. Because facing direction is binary, a single forced direction often collapses a branch. When a person’s direction becomes known, any clue referencing that direction becomes a hard positional constraint.

Solving algorithm: (1) List all persons and create a direction column (N/S/unknown). (2) Determine the minimum line length from gap clues. (3) Place the most anchored person first — usually someone with a fixed position (“G sits at the leftmost seat”) or a fixed direction plus a relative clue. (4) Propagate relative positions using the index equations. (5) After each placement, update the set of free indices and re‑evaluate gap constraints. (6) When a contradiction appears (two persons assigned to same index, or a gap requirement impossible), backtrack to the last decision point.

Because linear puzzles with gaps are essentially constraint‑satisfaction problems on a line, the “most constrained variable” heuristic works well: always place the person with the fewest legal indices next. This mirrors the singleton‑search in grid puzzles and keeps the search tree shallow.

Hybrid Puzzles: Seating Combined with Ranking and Blood Relations

The most demanding puzzles merge a seating arrangement (circular or linear) with an independent ranking (e.g., test scores, heights) and a blood‑relation tree. The key is to treat each domain as a separate sub‑puzzle linked only by shared entities (the same persons appear in all domains). Solve each sub‑puzzle to the maximum extent possible before attempting cross‑domain deductions.

Start with the domain that offers the most direct clues. Ranking puzzles often give pairwise comparisons (“P scored higher than Q but lower than R”). Convert these into a directed acyclic graph and compute a partial order. If the ranking is total (no ties), the graph will eventually become a Hamiltonian path — the exact order. Blood‑relation clues (“M is the sister of N’s father”) build a family tree; use standard notation (M = female, sibling of N’s father → M is N’s aunt). Represent the tree with nodes and edges; gender clues are critical for later “who is the brother of…” questions.

Cross‑domain clues appear as bridges: “The person who sits opposite the tallest person is the uncle of the person who ranked third.” Translate each reference into a variable: seat(opposite(tallest)) = uncle(rank3). Because each domain is already partially resolved, such a clue often pins down a unique individual in one domain, which then resolves the linked individual in the other domain.

Iterative cross‑propagation loop: (1) Solve each domain to a set of candidate assignments (e.g., possible seats for each person, possible ranks, possible family roles). (2) Apply every cross‑domain clue to intersect candidate sets. (3) If any domain yields a singleton, commit it and propagate within that domain (e.g., fixing a rank may fix a seat via a bridge clue). (4) Repeat until stable. Because each domain is a permutation problem, the total search space is the product of the domain sizes; cross‑clues dramatically reduce this product.

Exam strategy: allocate the first 60 seconds to reading all clues and classifying them by domain. Sketch three mini‑diagrams side by side — circle for seating, vertical list for ranking, tree for relations. Use a consistent colour or symbol for each person across diagrams (e.g., A = red dot). This visual linking makes bridge clues instantly recognizable. Practice with at least ten hybrid sets before the test; the pattern of “bridge appears late in the clue list” is common, so do not finalize any domain until the last clue is processed.

  • Treating an inclusive either‑or as exclusive and discarding valid arrangements.
  • Ignoring the contrapositive of a conditional clue, leaving a propagation path unused.
  • Assuming the line orientation (left/right) without verifying the first directional clue.
  • Failing to update all three domains after a cross‑domain singleton is found.
  • Overlooking the possibility of a vacant floor or empty seat when the total count is not explicitly given.
  • Committing a branch in an either‑or before checking all negative clues that could invalidate it.
  • Using mental arithmetic for gap calculations instead of writing the index equation, leading to off‑by‑one errors.

Circular Arrangements with Variable Seat Counts and Directional Constraints

When the number of seats around a circle is not fixed, the solver must treat the seat count as an additional variable bounded by the clues. Start by listing every person and every explicit directional clue (e.g., “A sits third to the left of B”). Convert each relative clue into a modular equation: if the total seats are n, then the index of A equals (index of B − 3) mod n. Collect all such equations; they form a system of congruences that restricts the possible values of n. The smallest n satisfying every congruence and the “no two persons share a seat” condition is the minimal viable circle. If clues also specify facing directions (inward/outward), add a binary variable for each person. A clue such as “the person facing outward sits opposite C” translates to index(opposite) = (index(C) + n/2) mod n, which forces n to be even. Use the parity requirement to prune impossible n values early. After the minimal n is identified, place the most anchored person (often the one with a fixed position or a fixed direction plus a relative clue) and propagate using the modular equations. Whenever a placement creates a conflict — two persons assigned to the same index or a direction requirement violated — increase n to the next candidate that satisfies all congruences and repeat. This incremental search guarantees the first consistent arrangement is the unique solution because any larger circle would introduce empty seats not referenced by any clue, which contradicts the “all seats occupied” principle unless explicitly allowed.

Practical tip: write the congruences in a table with columns for person, known offset, and direction. Solve the system by hand for small n (≤ 12) using the Chinese Remainder Theorem logic; for larger n a quick spreadsheet modulo column works. Always verify that the final arrangement respects every “left/right” clue in both clockwise and anticlockwise senses, because the wording “to the left” assumes a fixed orientation (usually facing the centre). If the orientation is ambiguous, treat both possibilities as separate branches and discard the one that fails any clue.

Linear Arrangements with Overlapping Group Constraints

Overlapping group constraints appear when clues define multiple subsets that share members, such as “the three engineers sit together” and “the two engineers who are also managers sit at the ends of the engineer block”. Represent each group as an interval on the line. Let the line have positions 1…L (L unknown). For each group, introduce a start variable s and a length k (known from the clue). The interval occupies positions s, s+1, …, s+k‑1. Overlap between groups translates into equalities or inequalities between their start variables. For the example above, let the engineer block start at e (length 3). The manager‑engineer sub‑block of length 2 must start at either e or e+1 (the two possible end‑pairs of a length‑3 block). This yields a disjunction: (subStart = e) ∨ (subStart = e+1). Encode each disjunction as a binary branch. Simultaneously, any “gap” clues (e.g., “at least two seats between the doctor block and the teacher block”) become linear inequalities: |start₁ − start₂| ≥ length₁ + gap + 1. Collect all equalities, inequalities, and disjunctions into a constraint‑satisfaction problem. Solve by first fixing the most constrained group — usually the one with the smallest length and the most overlap conditions. Propagate its start value through equalities; each propagation reduces the domain of other start variables. When a disjunction is encountered, explore both branches depth‑first, but prune immediately if any inequality becomes impossible (e.g., required gap exceeds remaining free seats). Because the line length L is the maximum of (start + length − 1) across all groups, L emerges automatically; if clues give an absolute position (“X sits at position 5”), treat it as an equality fixing that group’s start. The algorithm terminates when all groups have a single feasible start and no inequality is violated. The final arrangement is read by expanding each interval in order of start position; any empty seats not covered by groups remain vacant unless a clue forbids vacancies.

Technique: Branch‑and‑Bound on Overlapping Intervals
Eight persons A–H must sit in a row. Clues: (1) A, B, C sit consecutively in that order. (2) D and E sit together. (3) The block (A,B,C) and the block (D,E) have exactly one empty seat between them. (4) F sits at the leftmost seat. (5) G sits immediately to the right of H. Determine the seating order.
Create intervals: I₁ = (A,B,C) length 3, order fixed; I₂ = (D,E) length 2, order unknown; I₃ = (G,H) length 2 with order H then G (since G right of H). Clue 4 fixes position 1 = F. Let start₁ = s₁, start₂ = s₂, start₃ = s₃. From clue 3: |s₁ − s₂| = 3 + 1 + 1 = 5 (because one empty seat between blocks means distance between starts = length₁ + 1 + 1). Since I₁ is left of I₂ or right; try s₂ = s₁ + 5 first. F occupies position 1, so s₁ ≥ 2. Minimum s₁ = 2 → I₁ occupies 2‑4. Then s₂ = 7 → I₂ occupies 7‑8. Remaining positions: 5,6,9,10… but total persons = 8, so line length at least 8. Positions 5‑6 are free for I₃ (length 2) → s₃ = 5 works (H at 5, G at 6). All persons placed: 1 F, 2 A, 3 B, 4 C, 5 H, 6 G, 7 D, 8 E. Check clues: all satisfied. The alternative branch s₁ = s₂ + 5 would push I₁ right of I₂, making F at position 1 impossible because I₂ would need to start ≤ ‑4. Hence unique solution: F A B C H G D E.
Exam‑Day Strategy
When you see overlapping group clues, sketch a timeline with labelled blocks before writing equations. Assign a distinct colour or symbol to each block; this visual map makes the “exactly one empty seat” type gaps obvious and prevents off‑by‑one errors. Solve the block with the most fixed internal order first — it reduces branching dramatically.

Floor Puzzles with Dynamic Vacancy and Multi‑Level Conditions

Dynamic vacancy means the number of floors or the number of occupants per floor is not given directly; it must be inferred from relational clues. Begin by listing all entities (persons, pets, vehicles, etc.) and the attributes that vary per floor (profession, colour, vehicle). Each floor can host multiple entities, but clues often state “exactly one person per floor” or “at most two”. Treat each floor as a bucket with a capacity variable cₖ. Clues such as “the engineer lives two floors above the doctor” translate to floor(engineer) = floor(doctor) + 2. If a clue says “the red car is on a floor with a vacancy”, introduce a Boolean vacancy variable for each floor. The system becomes a mixed integer‑linear problem: integer floor assignments for each entity, binary vacancy flags, and capacity constraints Σ entities on floor k ≤ cₖ. Because the total number of floors is usually ≤ 10, a manual branch‑and‑bound works: guess the total floor count F from the maximum floor difference implied by clues (e.g., “topmost floor” and “bottommost floor” give a span). For each candidate F, create a table of floors 1…F. Place the most anchored entity first — often the one with an absolute floor reference (“the manager lives on floor 3”). Propagate relative clues, updating possible floor sets for each remaining entity. When a floor’s occupancy reaches its capacity, mark remaining slots as unavailable. If a clue requires a vacancy on a specific floor, ensure at least one capacity slot stays empty. Continue until all entities are placed or a contradiction forces a new F. Multi‑level conditions (e.g., “the person who owns the blue bike lives on a floor immediately above a floor where a doctor lives”) are handled by chaining the floor equations: floor(bikeOwner) = floor(doctor) + 1. Because each entity appears in at most a few clues, the propagation depth stays shallow. The final check verifies that every attribute combination (profession‑colour‑vehicle) is unique if the puzzle demands it.

Sequential Logic Chains Across Multiple Domains

Sequential logic chains arise when a clue links a position in one domain to a rank in another, which in turn links to a relation in a third domain. Model each domain as a separate permutation: seating (circular or linear), ranking (total order), and relations (family tree). Assign each person a unique identifier. For each domain, build a constraint graph: seating graph edges are adjacency or opposition; ranking graph edges are “higher than” directed arcs; relation graph edges are parent/child/sibling with gender labels. Cross‑domain clues become hyper‑edges connecting nodes across graphs, e.g., seat(opposite(tallest)) = uncle(rank3). Translate each hyper‑edge into a logical equivalence: the person occupying seat X must be the same individual who holds rank Y and relation Z. The solving procedure is an iterative fixed‑point algorithm: (1) Initialise each domain with all permutations consistent with intra‑domain clues. (2) For every cross‑domain clue, intersect the candidate sets of the linked variables across domains. (3) If any variable’s candidate set reduces to a singleton, commit that assignment and propagate within its domain (e.g., fixing a rank may fix a seat via a bridge clue). (4) Repeat until no further reduction occurs. Because each domain is a permutation problem, the candidate sets are subsets of the symmetric group; intersection is set‑intersection of possible persons. The algorithm converges quickly when cross‑clues are dense. If convergence stalls with multiple candidates, pick the most constrained variable (smallest candidate set) and branch, recursing the whole fixed‑point loop. This mirrors the DPLL algorithm for SAT and guarantees completeness. In practice, limit branching to at most two levels; exam puzzles are designed so that a single cross‑clue cascade resolves the rest.

Optimization Techniques for Large‑Scale Permutation Puzzles

When the number of persons exceeds ten, exhaustive search becomes impractical. Apply constraint‑programming heuristics: variable ordering (most constrained first), value ordering (least constraining value), and forward checking. Represent each clue as a constraint object with a propagate() method that prunes domains. For seating, domains are possible indices; for ranking, possible rank numbers; for relations, possible roles. Implement a simple recursive solver in a notebook or on paper using a stack of decisions. At each decision point, choose the person with the smallest domain size > 1. Try each value, snapshot the current domains, propagate all constraints, and if any domain empties, backtrack. Use symmetry breaking: if the puzzle is invariant under rotation (circular) or reflection (linear), fix one person’s position arbitrarily (e.g., person A at index 1) to cut the search space by a factor of n. For linear puzzles with facing directions, also fix the global orientation by assuming the leftmost seat faces north unless a clue contradicts. Memoise partial assignments: store a hash of the current domain state; if the same state reappears, prune. These techniques reduce a 12‑person circular puzzle with 30 clues from millions of permutations to a few hundred nodes, solvable within the two‑minute exam window. Practice by implementing a tiny solver for 8‑person puzzles; the pattern recognition gained transfers directly to manual solving.

  • Assuming a fixed number of seats or floors when the puzzle only gives relative clues, leading to missed vacancy possibilities.
  • Treating overlapping group constraints as independent blocks instead of linking their start variables, causing double‑counting of shared members.
  • Ignoring the parity requirement (even/odd seat count) when a clue forces an opposite pair in a circular arrangement.
  • Failing to propagate a cross‑domain singleton immediately, leaving stale candidate sets that generate false branches.
  • Overlooking symmetry breaking in circular or linear puzzles, inflating the search space and wasting time.
  • Using mental arithmetic for gap calculations instead of writing the index equation, resulting in off‑by‑one errors.
  • Neglecting to verify that every attribute combination remains unique after the final arrangement is read.

Practice Questions

Syllogism →