Direction Sense – Advanced Study Notes
(Reasoning Ability – MIT‑Level Treatment)
Goal – Equip you with a rigorous, first‑principles framework for reasoning about spatial direction.
Audience – Graduate students in mathematics, physics, or computer science who wish to master the logical analysis of directional statements (e.g., “A is left of B”, “C is north‑east of D”).
1. Prelude: Why Formalise Direction?
-
Ambiguity in natural language
- “Left” depends on the observer’s orientation.
- “North‑east” is a compound direction that can be interpreted as a unit vector or a cone in 2‑D space.
-
Need for precision in reasoning tests
- Tests often ask: “If A is left of B and B is above C, where is A relative to C?”
- A formal system guarantees uniqueness or consistency of conclusions.
-
Connections to other fields
- Robotics & path planning
- Computational geometry
- Graph theory & constraint satisfaction
2. Foundations – From Geometry to Vectors
2.1 Euclidean Plane & 3‑D Space
| Notation | Meaning |
|---|---|
| ( \mathbb{R}^2) | 2‑D Euclidean plane |
| ( \mathbb{R}^3) | 3‑D Euclidean space |
| ( \mathbf{e}_x, \mathbf{e}_y, \mathbf{e}_z) | Standard basis vectors |
| ( \mathbf{v} = (v_x, v_y, v_z)) | Position vector |
First Principle – Every point can be represented uniquely by a vector from a chosen origin.
2.2 Direction as a Unit Vector
A direction (D) in (\mathbb{R}^n) is a unit vector ( \mathbf{d}) with (\lVert \mathbf{d}\rVert = 1).
- “North” → (\mathbf{d} = (0,1)) in 2‑D.
- “Left” relative to an observer facing east → (\mathbf{d} = (0,1)).
Advantages
- Directions are vectors, so they can be added, rotated, and compared using dot products.
2.3 Relative Position Constraints
Given two points (P, Q) with position vectors (\mathbf{p}, \mathbf{q}):
Definition – (P) is “to the left of (Q)” iff
[ \mathbf{p} = \mathbf{q} + \lambda , \mathbf{e}_y + \mathbf{v}, \quad \lambda>0,\ \mathbf{v}\cdot \mathbf{e}_y = 0 ]
i.e., (P) lies somewhere on the line through (Q) in the direction (\mathbf{e}_y).
In practice, we only know the inequality ((p_y>q_y)) and no constraint on (p_x).
3. Systematic Representation of Directional Statements
3.1 Encoding Directions as Inequalities
| Symbol | Direction | Inequality |
|---|---|---|
| (>) | “above” (north) | (p_y > q_y) |
| (<) | “below” (south) | (p_y < q_y) |
| “(\rightarrow)” | “right” (east) | (p_x > q_x) |
| “(\leftarrow)” | “left” (west) | (p_x < q_x) |
Compound direction (NE) → (p_x>q_x \ \text{and}\ p_y>q_y).
3.2 Graph‑Theoretic View
- Vertices – Points (A, B, C, \dots).
- Edges – Directed edges labeled with inequalities.
Example:
(A \xrightarrow{<} B \xrightarrow{>} C) means
(p_A < p_B) (in x) and (p_B > p_C) (in y).
3.3 Constraint Satisfaction
You now have a system of linear inequalities.
- Feasibility – Does there exist an assignment of coordinates satisfying all constraints?
- Uniqueness – Are there multiple solutions?
This reduces to a 2‑D linear programming problem, which can be solved by the simplex algorithm or graph‑based topological sorting.
4. Advanced Theory – Group Theory & Rotations
4.1 The Rotation Group (SO(2))
- Definition – The set of all orientation‑preserving linear transformations of (\mathbb{R