📚 The Poisson Distribution | 泊松分布
The Poisson distribution is a discrete probability distribution that expresses the probability of a given number of events occurring in a fixed interval of time or space, provided these events happen with a known constant mean rate and independently of the time since the last event. Named after French mathematician Siméon Denis Poisson, it is one of the most widely used distributions for modelling count data, such as the number of calls received by a call centre in an hour, the number of decay events from a radioactive source, or the number of typing errors on a page. In the IB Mathematics curriculum, particularly in the 'Statistics and Probability' topics of Analysis & Approaches HL and Applications & Interpretation HL, the Poisson distribution offers a powerful tool for tackling problems that go beyond the binomial framework, helping students understand random processes that involve rare or random events.
泊松分布是一种离散概率分布，它描述了在固定时间或空间区间内某事件发生特定次数的概率，前提是这些事件以已知的恒定平均速率发生，且每次发生的时间相互独立。该分布以法国数学家西莫恩·德尼·泊松的名字命名，是建模计数数据时应用最广泛的分布之一，例如客服中心每小时接到的电话数、放射源衰变事件数或一页书上的打字错误数等。在IB数学课程中，特别是在分析与方法(AA)高水平以及应用与解释(AI)高水平的“统计与概率”专题中，泊松分布为学生提供了处理超越二项分布框架问题的强有力工具，帮助他们理解涉及稀有或随机事件的随机过程。
1. Introduction to the Poisson Distribution | 泊松分布简介
A Poisson experiment examines how many times a specific event occurs during a given interval. The interval could be time, distance, area, volume, or any other continuous measure. The key feature is that events occur randomly and independently, and the average rate of occurrence λ (lambda) is constant.
泊松试验考察在给定区间内某一特定事件发生的次数。这个区间可以是时间、距离、面积、体积或任何其他连续的度量。其关键特征是事件随机且独立地发生，且事件发生的平均速率 λ（拉姆达）是恒定的。
The random variable X follows a Poisson distribution with parameter λ, written as X ~ Po(λ), where λ > 0. It represents the expected number of occurrences in that interval. Typical textbook problems might ask: 'If a switchboard receives on average 4.5 calls per minute, what is the probability that exactly 2 calls arrive in a given minute?'
随机变量 X 服从参数为 λ 的泊松分布，记作 X ~ Po(λ)，其中 λ > 0。它代表在相应区间内事件发生次数的期望值。典型的课本问题可能会问：“如果一台总机平均每分钟接听 4.5 通电话，那么在某一分钟内恰好接到 2 通电话的概率是多少？”
2. Definition and Probability Mass Function | 定义与概率质量函数
The probability mass function (PMF) of a Poisson-distributed random variable X is given by:
泊松分布随机变量 X 的概率质量函数 (PMF) 由下式给出：
P(X = k) = (e−λ × λᵏ) / k! , for k = 0, 1, 2, …
Here, e is Euler's number (approximately 2.71828), λ is the average rate, and k! denotes k factorial. The expression includes e−λ to ensure that the sum of all probabilities equals 1.
这里 e 是欧拉数（约等于 2.71828），λ 是平均发生率，k! 表示 k 的阶乘。表达式中的 e−λ 保证了所有概率之和等于 1。
For example, if λ = 2, the probability of getting exactly 3 events is P(X = 3) = (e−2 × 2³) / 3! F is only defined for non-negative integer values of k, making it a discrete distribution.
例如，若 λ = 2，恰好发生 3 次事件的概率为 P(X = 3) = (e−2 × 2³) / 3! ≈ (0.1353 × 8) / 6 ≈ 0.1804。请注意，概率质量函数仅定义在 k 的非负整数值上，因此这是一个离散分布。
3. Conditions for a Poisson Distribution | 泊松分布的条件
To model a situation with a Poisson distribution, four conditions must be satisfied. First, events must occur randomly and independently – the occurrence of one event does not affect the likelihood of another. Second, the average rate λ must remain constant throughout the interval. Third, two events cannot occur at exactly the same instant (events are distinct). Fourth, the probability of an event occurring in a very small sub-interval is proportional to the length of that sub-interval.
要用泊松分布对实际情况建模，必须满足四个条件。第一，事件必须随机且独立地发生——一个事件的发生不影响另一个事件发生的可能性。第二，平均速率 λ 在整个区间内必须保持恒定。第三，两个事件不能在完全相同的瞬间发生（事件是离散的）。第四，在一个极小的子区间内事件发生的概率与该子区间的长度成正比。
In practice, these conditions translate into questions like counting the number of cars passing a point on a quiet country road in an hour (rare, independent events) or the number of mutations in a strand of DNA. When the conditions are not met – for instance, if events cluster or the rate changes – the Poisson model becomes inappropriate.
在实践中，这些条件可以转化为诸如计算一小时内在一条安静乡村公路上经过某一点的汽车数量（稀有、独立的事件）或一条DNA链上突变的数量等问题。当条件不满足时——例如事件成群出现或发生率发生变化——泊松模型就不再适用。
4. Mean and Variance | 均值与方差
= λ and Var(X) = λ. This property makes parameter estimation particularly straightforward – if a sample yields a mean number of occurrences , we often set λ = mean count.
泊松分布的一个显著性质是期望值 E(X) 和方差 Var(X) 都等于 λ。即 E(X) = λ，Var(X) = λ。这一性质使参数估计变得特别直接——如果样本得到平均发生次数，我们通常直接令 λ = 平均计数。
Because the mean equals the variance, a dataset where the sample variance greatly exceeds the sample mean (overdispersion) suggests that a Poisson model may not be suitable. On the other hand, if variance is much smaller than the mean (underdispersion), the Poisson is also questionable. In IB exam contexts, you might be asked to verify whether a set of counts is consistent with a Poisson distribution by comparing the sample mean and variance.
由于均值等于方差，若一个数据集的样本方差远大于样本均值（过度离散），就暗示泊松模型可能不合适。反之，如果方差远小于均值（欠离散），泊松模型同样值得怀疑。在IB考试中，你可能会被要求通过比较样本均值和方差来验证一组计数数据是否与泊松分布一致。
5. Derivation from the Binomial Distribution | 从二项分布推导
probability of success p approaches zero, and their product np remains constant, equal to λ. Let X ~ B(n, p) with large n and small p such that np = λ. Then, for a fixed k,
泊松分布可以作为二项分布 B(n, p) 的极限情况推导出来，此时试验次数 n 趋近无穷大，成功概率 p 趋近于零，且两者的乘积 np 保持恒定并等于 λ。设 X ~ B(n, p)，其中 n 很大、p 很小且满足 np = λ。那么对于固定的 k，有
P(X = k) = C(n, k) pᵏ (1 – p)n – k → (e−λ × λᵏ) / k! as n → ∞.
5. It links two of the most fundamental discrete distributions and deepens understanding of how the Poisson models rare events.
这一推导为在 n 较大、p 较小时使用泊松分布近似二项分布提供了依据，通常当 n ≥ 50 且 np ≤ 5 时该近似是合理的。它连接了两个最基本的离散分布，加深了学生对泊松分布如何模拟稀有事件的理解。
6. Shape and Skewness | 分布形状与偏度
tribution can be approximated by a normal distribution with mean λ and variance λ, applying a continuity correction.
当 λ 值较小（如 λ < 5）时，泊松分布呈现明显的右偏态，在大值方向有一条长尾。随着 λ 增大，分布变得越来越对称，开始呈现钟形曲线。当 λ ≥ 10 时，泊松分布可以用均值为 λ、方差为 λ 的正态分布来近似，并需要进行连续性校正。
t-skewed distribution. When λ = 8, probabilities are more evenly spread around the centre.
理解分布的形状有助于可视化概率计算。例如，当 λ = 1 时，P(X ≤ 2) 会远大于 P(X ≥ 5)，因为在一个严重右偏的分布中，概率质量集中在零点附近。而当 λ = 8 时，概率更均匀地围绕中心分布。
7. Sum of Independent Poisson Variables | 独立泊松变量之和
If X ~ Po(λ₁) and Y ~ Po(λ₂) are independent Poisson random variables, then their sum X + Y also follows a Poisson distribution with parameter λ₁ + λ₂. Formally, X + Y ~ Po(λ₁ + λ₂). This additive property extends to any finite number of independent Poisson variables.
如果 X ~ Po(λ₁) 和 Y ~ Po(λ₂) 是相互独立的泊松随机变量，那么它们的和 X + Y 也服从泊松分布，参数为 λ₁ + λ₂。形式上，X + Y ~ Po(λ₁ + λ₂)。这一可加性质可以推广到任意有限个独立的泊松变量。
This property is extremely useful for merging independent Poisson processes. For example, if two separate call centres receive calls at independent Poisson rates of 3 per minute and 5 per minute, the combined incoming calls follow a Po(8) distribution. This summation rule is often examined in IB problems dealing with combined sources of events.
这一性质在合并独立泊松过程中极为有用。例如，如果两个分别的客服中心以每分钟3通和每分钟5通的独立泊松速率接听电话，那么合并后的来电数服从 Po(8) 分布。在处理事件来源合并的IB问题中，这一求和规则经常被考查。
8. Using the Poisson Cumulative Distribution Tables | 泊松累积分布表的使用
: P(X = 4) = P(X ≤ 4) − P(X ≤ 3). For a range, use P(a ≤ X ≤ b) = P(X ≤ b) − P(X ≤ a − 1).
IB考试为选定的 λ 值提供了泊松累积分布表。这些表格列出了整数 x 对应的 P(X ≤ x)。要求出某个确切取值的概率，例如 P(X = 4)，可以相减表格值：P(X = 4) = P(X ≤ 4) − P(X ≤ 3)。对于区间概率，使用 P(a ≤ X ≤ b) = P(X ≤ b) − P(X ≤ a − 1)。
Here is a snippet of a cumulative Poisson table for λ = 2.5:
下面是一个 λ = 2.5 的累积泊松分布表片段：
| x | P(X ≤ x) |
|---|---|
| 0 | 0.0821 |
| 1 | 0.2873 |
| 2 | 0.5438 |
| 3 | 0.7576 |
| 4 | 0.8912 |
| 5 | 0.9580 |
5, which matches direct calculation. Being adept with these tables is essential for timed IB papers.
因此，P(X = 2) = 0.5438 − 0.2873 = 0.2565，这与直接计算相符。熟练使用这些表格对于限时的IB考试至关重要。
9. Applications: Modelling Rare Events | 应用：罕见事件建模
The Poisson distribution fundamentally models rare events. Classic examples include the number of earthquakes above a certain magnitude in a year, the number of defective items in a large batch produced under quality control, or the number of customers arriving at a bank in a ten-minute interval. In each case, the event is relatively uncommon in the chosen time frame, making the Poisson assumptions reasonable.
泊松分布从根本上讲就是用来对稀有事件建模。经典例子包括一年中发生超过某一震级的地震次数、在质量管控下大量生产批次中的残次品数量，或者十分钟内到达银行的顾客数量。在每种情形下，所选时间范围内的事件相对稀少，这使得泊松假设合理。
In biology, the distribution of yeast cells in a suspension under a microscope can be described by a Poisson distribution. In genetics, the number of mutations occurring in a stretch of DNA can be modelled similarly. Such applications highlight the versatility of the Poisson in pure and applied settings.
在生物学中，显微镜下悬液中酵母细胞的分布可以用泊松分布描述。在遗传学中，一段DNA上发生的突变数量也可类似建模。这些应用突出了泊松分布在纯数学和应用场景中的多功能性。
10. Approximating the Binomial with Poisson | 用泊松分布近似二项分布
ults. The approximation simplifies probability computations and is widely used in statistical practice.
当 n 很大（n ≥ 50）且 p 很小（通常 p ≤ 0.1，且 np ≤ 5）时，二项分布 B(n, p) 的计算变得繁琐。以 λ = np 为参数的泊松近似能给出相当准确的结果。这种近似简化了概率计算，并在统计实践中广泛应用。
For example, suppose 1% 0.99¹⁹⁷, which is tedious. Using Poisson with λ = 200 × 0.01 = 2, we compute P(X = 3) ≈ (e−2 o the exact value (0.1814). This illustrates the practical power of the Poisson approximation.
例如，假设生产的灯泡有 1% 是次品，我们抽取 200 个灯泡。恰有 3 个是次品的精确概率为 C(200,3)×0.01³×0.99¹⁹⁷，计算十分繁琐。使用 λ = 200 × 0.01 = 2 的泊松近似，我们算得 P(X = 3) ≈ (e−2 × 2³)/6 ≈ 0.1804，与精确值 0.1814 非常接近。这体现了泊松近似的实际威力。
11. Assumptions and Limitations | 假设与局限性
Despite its elegance, the Poisson distribution rests on strict assumptions. The rate λ must be constant; if the event rate varies over time (e.g., rush hour traffic vs. off-peak), a homogeneous Poisson process is inappropriate. Events must be independent; contagious occurrences (like the spread of a rumour) violate this. Moreover, the mean-variance equality may not hold in real data, requiring alternative models such as the negative binomial distribution.
尽管泊松分布十分优美，但它依赖于严格的假设。发生率 λ 必须恒定；如果事件发生率随时间变化（例如高峰时段交通与平峰时段的差异），则齐次泊松过程不再适用。事件必须相互独立；具有传染性的事件（如谣言的传播）违反了这一假设。此外，均值与方差相等的性质在真实数据中可能不成立，此时需要诸如负二项分布等替代模型。
In IB examinations, you may be asked to comment on the suitability of a Poisson model by examining whether the given context meets these assumptions. Recognising when a Poisson is not appropriate is as important as being able to use it correctly. Always check for clustering, trend, or dependence before applying the distribution.
在IB考试中，你可能会被要求通过检查给定情境是否满足这些假设来评述泊松模型的适宜性。识别出泊松分布不适用的情形与能够正确使用它同等重要。在应用该分布之前，务必检查是否存在聚集、趋势或相依性。
12. Worked Example | 例题讲解
Problem: A radioactive source emits particles at an average rate of 3.2 particles per second. Assume the number of particles emitted follows a Poisson distribution. Find the probability that in a randomly chosen one-second interval (i) exactly 2 particles are emitted, (ii) at most 1 particle is emitted, and (iii) between 3 and 5 particles inclusive are emitted.
题目：一个放射源平均每秒发射 3.2 个粒子。假设发射的粒子数服从泊松分布。求在随机选取的一秒间隔内 (i) 恰好发射 2 个粒子的概率，(ii) 最多发射 1 个粒子的概率，以及 (iii) 发射 3 到 5 个（含）粒子的概率。
Let X ~ Po(3.2). Using the formula or tables:
设 X ~ Po(3.2)。使用公式或查表：
(i) P(X = 2) = (e−3.2 × 3.2²) / 2! = (0.04076 × 10.24) / 2 ≈ 0.2087. | (i) P(X = 2) = (e−3.2 × 3.2²) / 2! = (0.04076 × 10.24) / 2 ≈ 0.2087。
(ii) P(X ≤ 1) = P(X = 0) + P(X = 1) = e−3.2 + (e−3.2 × 3.2) = 0.04076 + 0.1304 = 0.1712. | (ii) P(X ≤ 1) = P(X = 0) + P(X = 1) = e−3.2 + (e−3.2 × 3.2) = 0.04076 + 0.1304 = 0.1712。
λ = 3.2, P(X ≤ 5) ≈ 0.8946, P(X ≤ 2) ≈ 0.3799, so probability ≈ 0.8946 − 0.3799 = 0.5147. | (iii) P(3 ≤ X ≤ 5) = P(X ≤ 5) − P(X ≤ 2)。查 λ = 3.2 的累积表得 P(X ≤ 5) ≈ 0.8946，P(X ≤ 2) ≈ 0.3799，因此概率 ≈ 0.8946 − 0.3799 = 0.5147。
Thus, there is roughly a 20.9% chance of exactly two emissions, a 17.1% chance of at most one, and a 51.5% chance of observing three to five emissions. These calculations demonstrate direct use of the PMF and cumulative table, essential skills for the IB statistics component.
因此，恰好发射两个粒子的概率约为 20.9%，至多一个粒子的概率为 17.1%，观察到三到五个粒子的概率为 51.5%。这些计算展示了概率质量函数与累积表的直接应用，是IB统计模块的核心技能。
Published by TutorHao | 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