Direction Sense – Advanced Study Notes
Subject: Reasoning Ability
Test‑Type: Direction Sense (Relative Orientation Problems)
Level: MIT‑Level / First‑Principles Treatment
In a direction‑sense problem we are given a set of relative statements of the form “(X) is N, NE, S, etc. of (Y)” and must deduce a specific position or orientation. These problems are ubiquitous in reasoning‑skill tests, aptitude exams, and even in AI constraint‑solving.
1. Foundations
| Concept | Symbol | Explanation | Example |
|---|---|---|---|
| Cardinal Directions | (N, E, S, W) | 90° apart on a compass. | “A is north of B” |
| Intermediate Directions | (NE, SE, SW, NW) | 45° between cardinal points. | “C is southeast of D” |
| Relative Position | (X \xrightarrow{D} Y) | X is in direction (D) relative to Y. | (A \xrightarrow{NE} B) |
| Orientation Graph | (G=(V,E)) | Vertex set (V) = persons; directed edge (X\to Y) labelled by direction. | (A\to B) labelled (NE) |
| Transitivity | (X \xrightarrow{D_1} Y, Y \xrightarrow{D_2} Z \Rightarrow X \xrightarrow{D_3} Z) | (D_3) is the vector sum of (D_1) and (D_2). | (A \xrightarrow{NE} B, B \xrightarrow{E} C \Rightarrow A \xrightarrow{NE+E=E+NE}) |
| Partial Order | (\leq) on directions | If all directions lie in a plane, the set of directions forms a lattice. | (N \leq NE \leq E) |
1.1 Coordinate Representation
Assign each person a 2‑D vector (\mathbf{p}_i=(x_i,y_i)).
Direction (D) corresponds to a unit vector (\mathbf{d}_D):
| Direction | (\mathbf{d}_D) |
|---|---|
| (N) | ((0,1)) |
| (NE) | ((\frac{\sqrt2}{2},\frac{\sqrt2}{2})) |
| (E) | ((1,0)) |
| ... | ... |
Relative statement “(X) is in direction (D) of (Y)” translates to:
[ \mathbf{p}_X - \mathbf{p}_Y = \lambda \mathbf{d}_D,\qquad \lambda>0 ]
(Scale (\lambda) is irrelevant for pure direction reasoning.)
2. Modeling as a Constraint Satisfaction Problem (CSP)
| Variable | Domain | Constraint |
|---|---|---|
| (\mathbf{p}_i) | (\mathbb{R}^2) | For each directed edge (i \xrightarrow{D} j): (\mathbf{p}_i - \mathbf{p}_j \parallel \mathbf{d}_D) and ((\mathbf{p}_i-\mathbf{p}_j)\cdot \mathbf{d}_D>0) |
- Parallelism ensures the relative direction is correct.
- Positive dot product guarantees correct orientation (not opposite direction).
- All constraints are linear equations and inequalities in 2‑D.
Transitive Closure
For any path (i \to k_1 \to k_2 \dots \to j), we can derive a composite direction (D) by adding unit vectors. If the resulting direction does not match a known cardinal direction, we can deduce that the problem is inconsistent or the given statements are incomplete.
3. Theoretical Properties
3.1 Consistency Checking
-
Acyclicity: If the orientation graph contains a directed cycle, the product of the direction vectors around the cycle must be the identity (i.e., zero displacement).
- Example: (A \xrightarrow{NE} B, B \xrightarrow{SW} A) → product = (NE \times SW = 0) → consistent.
-
Transitivity Violations: If (A \xrightarrow{N} B) and (B \xrightarrow{E} C) but the problem states (A \xrightarrow{SW} C), that is a contradiction because (N+E \neq SW).
-
Ambiguity: If multiple orientations satisfy all constraints, the problem admits multiple solutions. Standard tests usually avoid such cases.
3.2 Complexity
- For planar problems with only cardinal directions, consistency checking is linear in the number of statements (graph traversal + vector addition).
- When intermediate directions and distance constraints (e.g., “two units apart”) are included, the CSP becomes NP‑complete