How Mathematical Modeling Can Supercharge Your Civil Service Exam Prep

Core Modeling Paradigm

Any quantitative problem can be expressed as: real situation → abstraction → mathematical model → solution → result → back to original question → answer. Identify core variables and constraints; exam questions are designed so that the number of equations equals unknowns, guaranteeing a unique solution.

Specific Features of Civil‑service Exam Modeling

Only a limited, enumerable set of model types, data designed for rapid calculation, and a single correct choice among four options. Candidates rely on pattern recognition to invoke pre‑stored models and substitution verification to confirm answers.

Mathematical Models for Common Question Types

Engineering (Work) Problems

Basic model: W = E × T where W is total work, E efficiency, T time.

Modeling trick: Let L be the least common multiple of individual completion times; then efficiencies become integers.

Co‑operation example: A needs 12 days, B 18 days, C 9 days. Set L = lcm(12,18,9) = 36. The combined time is computed as 1 / (1/12 + 1/18 + 1/9) = 4 days.

Travel (Rate‑Distance) Problems

Encounter model: Two objects move toward each other, initial distance D, speeds v₁ and v₂. Meeting time t = D / (v₁ + v₂).

Chase model: Same direction, initial gap D, relative speed v_rel = v_fast – v_slow. Catch‑up time t = D / v_rel.

River‑boat model: Boat speed in still water v_b, current speed v_c. Downstream speed v_b + v_c, upstream speed v_b – v_c.

Permutations, Combinations and Classical Probability

Permutation model (ordered selection): Number of ways P(n,k) = n! / (n‑k)!.

Combination model (unordered selection): Number of ways C(n,k) = n! / (k! (n‑k)!).

Classical probability: P = N_favorable / N_total. When direct counting is difficult, use the complement principle: P = 1 – P(complement).

Profit‑Related Problems

Basic relationships: Profit = SellingPrice – Cost. If cost = C, list price = P, discount = d, actual selling price = P·(1‑d), then Profit = P·(1‑d) – C.

Break‑even model: Fixed cost F, variable cost per unit v, unit price p, sales volume q. Break‑even when F + v·q = p·q → q = F / (p – v).

Concentration (Mixture) Problems

Mixing two solutions: C_final = (m₁·C₁ + m₂·C₂) / (m₁ + m₂), where m denotes mass and C concentration.

Time‑Optimization Model

Weighted Time Allocation

Total exam time T_total = 120 minutes. Let N_i be the number of questions in module i and k_i a weighting coefficient. Allocated time for module i is t_i = k_i × N_i (normalized to the total time).

Typical coefficients (provincial‑level exam):

General knowledge: k = 0.5 → 8‑10 min

Verbal comprehension: k = 1.0 → 35 min

Quantitative relations: k = 1.5 → 15‑20 min

Logical reasoning: k = 1.0 → 35 min

Data analysis: k = 1.2 → 20‑25 min

Decision Model for Question Selection

For a given question, estimate time t, success probability p, and score s. Expected benefit per minute is EB = (p × s) / t. Prioritize questions with higher EB. Classification:

Must‑do: High benefit (e.g., engineering, simple travel, basic probability).

Optional: Medium benefit (complex permutations, multiple encounters).

Skip: Low benefit (number‑theory, complex geometry).

Learning‑Curve and Review Models

Power‑Law Learning Curve

Time for the n ‑th practice: T_n = a × n^{‑b}, with learning‑rate exponent b ≈ 0.2–0.5. Early practice yields rapid improvement; marginal gains diminish later.

Ebbinghaus Forgetting Curve

Retention decays exponentially: R(t) = R₀ e^{‑λt}, where λ depends on material difficulty and review frequency. Effective mitigation uses spaced repetition—review just before the forgetting threshold.

Recommended review intervals after learning a new model: 1 day, 3 days, 7 days, 14 days, then progressively longer.

Marginal Utility of Practice

If U = f(Q) denotes ability gain from solving Q questions, marginal utility ΔU = f(Q+1) – f(Q) approaches zero as Q grows. When ΔU is negligible, shift to new problem types or higher difficulty to reactivate learning.

Practical Recommendations

Master the five core models; practice 30‑50 representative questions per model to develop instant pattern recognition.

Apply the weighted time‑allocation table in timed mock exams; record module‑wise accuracy and adjust focus.

During the exam, scan quickly, answer high‑benefit items first, and mark difficult items for later review.