📚 A-Level Edexcel Further Maths Decision 2 High-Scoring Tips | Edexcel 进阶数学决策数学 2 高分技巧
Decision Mathematics 2 (D2) takes your understanding of algorithms, optimisation and networks to a deeper, more abstract level. It rewards precision, logical flow and a disciplined approach to layout. This guide breaks down the core topics with actionable high-scoring techniques, common pitfalls and exam-focused strategy. Mastering these will boost your confidence and your final grade.
决策数学 2（D2）将你对算法、优化和网络的理解提升到更深入、更抽象的层次。它奖励精确性、逻辑流程和严谨的排版。本指南通过可操作的高分技巧、常见错误和应试策略，逐一剖析核心主题。掌握这些内容将提升你的信心和最终成绩。
1. Transportation Problems: North-West Corner, Stepping-Stone and VAM | 运输问题：西北角法、石阶法与 Vogel 近似法
In a transportation problem, your first task is to find an initial feasible solution. The North-West Corner method is simple but rarely optimal; Vogel's Approximation Method (VAM) usually gives a much better starting solution. For high marks, always show VAM penalty row and column calculations step by step.
在运输问题中，首要任务是找出初始可行解。西北角法简单却很少最优；Vogel 近似法（VAM）通常给出好得多的起始解。为获高分，务必逐步展示 VAM 惩罚行和列的计算过程。
ng the solution with the stepping-stone method, use a neat grid with shadow costs uᵢ and vⱼ nd then calculate all other u and v values from occupied cells. The improvement indices Δᵢⱼ = cᵢⱼ − (uᵢ + vⱼ index to bring into the solution, and carefully form the closed loop to adjust quantities.
使用石阶法改进解时，使用整洁的网格，清晰标注影子成本 uᵢ 和 vⱼ。记住设定 u₁ = 0，然后通过占用的格子计算所有其他 u 和 v 值。必须对所有未使用的路线计算改进指数 Δᵢⱼ = cᵢⱼ − (uᵢ + vⱼ)。选择最小的负指数引入解中，并仔细形成闭环来调整数量。
High-scoring tip: always double-check the total supply equals total demand. In unbalanced problems, introduce a dummy row or column with zero costs, and remember that dummy cells carry no real product.
高分技巧：务必再次确认总供应量等于总需求量。在不平衡问题中，引入一行或一列零成本的虚拟行/列，并记住虚拟格子不携带实际产品。
2. Allocation Problems and the Hungarian Algorithm | 分配问题与匈牙利算法
The Hungarian algorithm solves minimisation assignment problems efficiently, but many students lose marks on maximisation or unbalanced cases. For maximisation, subtract all entries from the largest entry in the matrix (or multiply by −1 and add a constant). Always write this transformed matrix explicitly.
匈牙利算法高效解决最小化分配问题，但许多学生会在最大化或不平衡情形中丢分。遇到最大化，用矩阵中的最大值减去所有项（或乘以 −1 再加一个常数）。务必明确写出变换后的矩阵。
For an unbalanced matrix where jobs differ from workers, add enough dummy rows or columns filled with zeros so that the cost matrix becomes square. Then proceed with row reduction, column reduction, and drawing the minimum number of lines to cover all zeros. A high-scoring answer shows the covering lines clearly on the final reduced matrix.
对于工作数与工人数不同的不平衡矩阵，添加足够多行或列为零的虚拟行/列，使成本矩阵变为方阵。然后依次进行行归约、列归约，以及用最少的直线覆盖所有零。高分答案会在最终归约矩阵上清晰画出覆盖线。
b 3, and state the minimum cost or maximum profit. If not optimal, reduce the uncovered elements and repeat the covering step.
如果直线数量等于矩阵的阶数，则可直接做出最优分配。清晰列出分配关系，如工人 A → 作业 3，并说明最小成本或最大利润。若未达最优，则缩减未被覆盖的元素，重复覆盖步骤。
3. Linear Programming: Two-Stage Simplex and Big-M Method | 线性规划：两阶段单纯形法与 Big-M 法
bleau correctly, clearly marking the objective row. For the two-stage method, stage 1 minimises W = sum of artificial variables; for Big-M, the objective is maximised with a large penalty M.
当需要引入人工变量时，可使用两阶段法或 Big-M 法。考官期望你正确建立初始单纯形表，清晰标记目标行。两阶段法中，第一阶段最小化 W = 人工变量之和；Big-M 法则在目标中用大 M 惩罚，求最大化。
During pivoting, choose the pivot column from the most negative coefficient in the objective row (for maximisation). Then compute θ-ratios using the RHS divided by the positive entries in the pivot column. The row with the smallest non-negative θ gives the pivot row. A common mistake is including rows with negative or zero denominator in θ-ratio calculations — never do this.
旋转时，对最大化问题从目标行最负系数选主元列。然后计算 θ 比值：RHS 除以主元列中的正项。最小非负 θ 对应的行即为主元行。一个常见错误是把分母为负或零的行纳入 θ 计算——绝对不要这样做。
High marks come from displaying the full sequence of tableaux neatly, with the entering and leaving variables labelled at each step. The final tableau must clearly state the optimal values of all variables and the optimal objective value. If the problem requests integer solutions, apply Gomory cuts or simply interpret the result if integer conditions are satisfied naturally.
高分在于整洁地展示完整的表序列，每一步标明进基和出基变量。最终表必须清晰陈述所有变量的最优值及最优目标值。若问题要求整数解，应用 Gomory 割平面，或若整数条件自然满足则可直接解释结果。
4. Game Theory: Saddle Points, Mixed Strategies and Graphic Solutions | 博弈论：鞍点、混合策略与图解方法
For a two-person zero-sum game, first check for a saddle point by finding maximin and minimax. If the maximin (row minimum's maximum) equals the minimax (column maximum's minimum), you have a stable solution and can state the value of the game immediately. Write the reasoning clearly.
对于两人零和博弈，首先通过寻找最大最小值和最小最大值检查鞍点。若最大最小值（行最小中的最大）等于最小最大值（列最大中的最小），则存在稳定解，可立即给出博弈值。要清晰写出推理过程。
×2 game is to use the graphical approach: plot the expected payoffs for each of row player' envelope (for row) or lower envelope (for column) will give the maximin or minimax point.
若无鞍点，则需混合策略。为任一局中人建立线性规划问题。对于 2×n 或 m×2 博弈，快速方法是使用图解法：绘制行局中人每种策略关于混合策略概率 p 的期望支付。上包络（行方）或下包络（列方）将给出最大最小点或最小最大点。
High-scoring answers label the axes clearly, indicate the scale, and mark the intersection points exactly by solving simultaneous equations. The optimal strategy probabilities must be given as exact fractions, not rounded decimals. Always verify that the sum of probabilities is 1.
高分答案要清晰标注坐标轴，指明尺度，并通过解联立方程精确求出交点。最优策略概率必须以精确分数给出，不可四舍五入为小数。始终验证概率之和为 1。
5. Network Flows: Maximum Flow and Minimum Cut | 网络流：最大流与最小割
The max-flow min-cut theorem guarantees that the maximum possible flow from source to sink equals the capacity of the minimum cut. To build a flow-augmenting path, start with initial feasible flows (often all zero) and label nodes systematically. Use forward arcs with spare capacity and backward arcs with existing flow.
最大流最小割定理保证了从源点到汇点的最大可能流等于最小割的容量。为构造增流路径，从初始可行流（通常全为零）开始，系统标记节点。使用尚有容量的前向弧和已有流量的后向弧。
On each path, the bottleneck capacity δ is the minimum of the spare capacities on forward arcs and of the existing flows on backward arcs. Update the flow: add δ on forward arcs and subtract δ on backward arcs. Repeat until no augmenting path can be found. This final flow is maximum.
在每条路径上，瓶颈容量 δ 是前向弧剩余容量和后向弧现有流量的最小值。更新流量：前向弧加 δ，后向弧减 δ。重复直至无法找到任何增广路径。最终所得即为最大流。
Identify a minimum cut by examining the final labelled nodes. The cut is the set of arcs from labelled to unlabelled nodes. Sum the capacities of these forward arcs to confirm the max-flow value. Clearly separate the labelled set S and unlabelled set T in your diagram. This is a direct way to earn full marks on the cut argument.
通过检查最终已标记的节点来识别最小割。割是从标记节点到未标记节点的所有弧的集合。对这些前向弧的容量求和，验证等于最大流值。在图中清晰分隔已标记集 S 和未标记集 T。这是割论证上拿满分的直接方法。
6. Critical Path Analysis: Activity on Node, Floats and Resource Levelling | 关键路径分析：节点活动图、时差与资源均衡
D2 uses activity-on-node (AON) networks. Learn to construct the diagram quickly: each activity is a box with earliest start (EST), earliest finish (EFT), latest start (LST) and latest finish (LFT) slots. The critical path consists of activities with zero total float. Always calculate total float as LFT − EFT or LST − EST; they must match.
D2 采用节点活动图（AON）。学会快速绘制：每一项活动为一个方框，内含最早开始时间 (EST)、最早完成时间 (EFT)、最晚开始时间 (LST) 和最晚完成时间 (LFT)。关键路径由总时差为零的活动组成。始终将总时差计算为 LFT − EFT 或 LST − EST；两者必须吻合。
Resource histograms can catch you out. Draw the histogram using the earliest start times first, and then smooth resources by shifting non-critical activities within their floats. A high-scoring technique is to present both 'before' and 'after' histograms with the maximum resource level clearly reduced. Annotate which activities you moved and by how many days.
资源直方图容易失分。先用最早开始时间绘出直方图，然后在其时差范围内移动非关键活动以平滑资源。高分技巧是同时展示“调整前”和“调整后”直方图，明显降低最大资源水平。注明移动了哪些活动以及移动了多少天。
When scheduling with limited resources, use a priority rule (e.g., smallest float first) and build a Gantt chart. Show cumulative resource usage step by step, and explain why delays occur. This logical explanation distinguishes top-band answers.
在资源受限的调度中，使用优先规则（如最小总时差优先）并构建甘特图。逐步展示累积资源用量，并解释延迟的原因。这种逻辑性解释是高分答案的标志。
7. Dynamic Programming: Bellman's Principle and Table Construction | 动态规划：贝尔曼原理与建表技巧
Dynamic programming breaks a problem into stages, with state variables defining the possible states at each stage. The optimal value function fₙ(s) denotes the best return from stage n onward, starting in state s. A clean table is crucial: columns for state, decision, immediate return, future value and total return.
动态规划将问题划分为若干阶段，状态变量定义每个阶段可能的状态。最优值函数 fₙ(s) 表示从阶段 n 开始、状态 s 出发的最佳收益。整洁的表格至关重要：列出状态、决策、即时收益、未来值和总收益等列。
diate return + fₙ₊₁( next state ) }. For maximisation problems, always highlight the optimal decision with an asterisk or bold text. For minimisation, choose the smallest total.
在最后阶段（边界），fₙ(s) 通常等于即时收益。然后利用递推关系逆向运算：fₙ(s) = max 或 min { 即时收益 + fₙ₊₁(下一状态) }。最大化问题中，始终用星号或粗体标出最优决策。最小化问题则选择最小的总计值。
At the end, read the optimal policy forward from the initial state. Trace through the starred decisions to get the sequence of actions. High-scoring answers include a brief written summary of the optimal policy, not just numbers.
最终，从初始状态正向读出最优策略。追踪带星号的决策以得到动作序列。高分答案不仅包含数字，还包含对最优策略的简短文字总结。
8. Integer Programming: Gomory Cuts and Branch-and-Bound | 整数规划：Gomory 割平面与分支定界法
he Simplex method yields fractional solutions but integer answers are required, use Gomory' ed by taking the fractional parts of the coefficients, with a ≤ inequality and a non-negative slack variable.
当单纯形法得出分数解而需要整数解时，使用 Gomory 割平面法。找出 RHS 为分数的一行，写成分数形式，分离整数部分和小数部分。利用系数的小数部分构成割，使用 ≤ 不等式并添加一个非负松弛变量。
Insert the new constraint into the optimal tableau and use the dual simplex method to restore feasibility. Watch out for negative entries in the cut row during dual pivoting — the pivot row is chosen by the most negative RHS, and the pivot column by the smallest non-negative ratio of objective row to negative constraint coefficients.
将新约束插入最优表，使用对偶单纯形法恢复可行性。对偶旋转时要注意割行中的负项——主元行由最负的 RHS 选出，主元列由目标行与负约束系数的最小非负比值选出。
constraints x ≤ k or x ≥ k+1. For high marks, draw a clear tree diagram showing each node' athoming. Prune a node when its bound is no better than the current best integer solution.
另一种方法是分支定界，通过添加约束 x ≤ k 或 x ≥ k+1 分割可行域。为得高分，画出清晰树形图，展示每个节点的解、界定和探明。当一个节点的界不优于当前最佳整数解时，将其剪枝。
9. Common Pitfalls and How to Avoid Them | 常见失分点与回避方法
Many students lose marks by not showing sufficient working. In D2, the steps of an algorithm are as important as the final answer. Write every pivot calculation, every reduced matrix and every θ-ratio explicitly. Examiners award method marks even if a numerical slip occurs, so visibility matters.
许多学生因展示步骤不足而失分。在 D2 中，算法步骤同最终答案一样重要。明确写出每一次旋转计算、每一个归约矩阵和每一个 θ 比值。即便出现数值错误，考官仍会授予方法分，因此可见性至关重要。
tract a surplus variable and add an artificial variable if needed. Label all variables clearly to avoid confusion.
线性规划约束中不等式方向的理解错误是另一处陷阱。对于 ≤ 约束，添加松弛变量；对于 ≥，减去剩余变量并在必要时添加人工变量。清晰标注所有变量以避免混淆。
In network flow, forgetting to update backward arcs correctly can destroy the flow conservation. Always check that flow into a node equals flow out for every intermediate node. A quick box-check of node balances can catch mistakes early.
在网络流中，忘记正确更新后向弧会破坏流量守恒。始终检查每个中间节点的流入量是否等于流出量。对节点平衡的快速核查能及早发现错误。
Finally, manage your time. D2 questions are long; allocate roughly 1.5 minutes per mark. If stuck on a tableau, move on and return later. A partially correct sequence of tableaux often secures over half the marks for that part.
最后，管理好时间。D2 题目篇幅较长；大致按每题 1.5 分钟分配时间。若卡在某个单纯形表，可先跳过稍后返回。部分正确的表序列通常能确保该部分一半以上的分数。
10. Exam Preparation Strategy and Practice Regimen | 备考策略与练习方案
Start your revision by reworking every worked example from the textbook without looking at the solution. Decision maths is procedural; you build fluency through repeated execution. Use previous Edexcel D2 papers to simulate exam conditions, strictly timing each session.
从重新演练教材上每个例题开始复习，不看答案。决策数学是程序性的；通过反复执行建立熟练度。使用历年 Edexcel D2 试卷模拟考试环境，严格计时。
Build a personal "mistake log" ing the wrong θ. Review this log the night before the exam. This targeted reflection prevents repeat errors.
建立个人“错误日志”，记录每一个失误：例如，把最大化误作最小化、忘记设定 u₁=0、或选错 θ 值。考前当晚复习该日志。这种针对性的反思能防止重复出错。
Active recall techniques work well for algorithms: recreate the Hungarian algorithm steps on a blank sheet, or draw a flow-augmenting example from memory. Pair up with a study buddy and swap explanations of Gomory cuts or dynamic programming recurrences.
主动回忆技巧对算法很有效：在空白纸上重现匈牙利算法步骤，或凭记忆画出一个增流示例。与学习伙伴结对，交换讲解 Gomory 割或动态规划递推关系。
On exam day, read every question twice. Underline command words such as "determine", "explain" and "hence". When a question says "hence", you must use previous results — otherwise you may waste time. Keep your work legible and well-spaced; this encourages examiners to award maximum marks.
考试当天，每题读两遍。在“确定”、“解释”和“因此”等指令词下划线。当题目出现“因此”时，必须使用前面的结果——否则可能浪费时间。保持书写清晰、间距合理，这有助于考官给出最高分。
Published by TutorHao | Further Pure & Decision Mathematics Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
或截图后用「扫一扫 → 相册」选择本图