皮克定理,一个在几何学领域闪耀的明珠,为我们提供了一种计算简单多边形面积的优雅方法。它的美妙之处在于,只需数出多边形边界上的格点数量以及内部的格点数量,就能轻松得到面积。然而,这背后隐藏着怎样的数学智慧?本文将深入探讨皮克定理的证明过程,力求用详实且自然的语言,揭示其内在的逻辑与魅力。
皮克定理的表述:
首先,让我们回顾一下皮克定理的内容。对于一个顶点在格点上的简单多边形(即多边形不自交,所有顶点都在整点上),其面积 A 可以由以下公式给出:
$A = I + frac{B}{2} 1$
其中:
$I$ 是多边形内部的格点数。
$B$ 是多边形边界上的格点数。
证明思路的铺垫:
要证明皮克定理,我们不能直接从一个任意的简单多边形出发。数学证明往往需要从最简单的“砖块”开始,逐步构建复杂结构。在这里,我们选择最简单的格点多边形——单位正方形作为我们的基础。
一个单位正方形(边长为 1,顶点在 (0,0), (1,0), (1,1), (0,1))的面积是 1。让我们套用皮克定理看看:
边界上的格点数 $B$: (0,0), (1,0), (1,1), (0,1) 共 4 个。
内部的格点数 $I$: 0 个。
代入公式:$A = 0 + frac{4}{2} 1 = 2 1 = 1$。完美匹配!
这给了我们初步的信心。现在,我们需要展示这个公式在更复杂的多边形上依然成立。证明通常会采用归纳法或者拆分法,而拆分法更为直观和易于理解。
证明一:拆分法(由任意简单多边形到矩形)
这个证明思路是,我们将一个任意的简单多边形,通过一系列操作,将其转化为我们熟悉的“砖块”——单位正方形,并在每一步验证皮克定理的性质是否保持不变。
1. 基本构成:单位正方形
我们已经验证了单位正方形的情况。
2. 两个相邻单位正方形的合并:
考虑两个共用一条边(且这条边上的格点数只算一次)的单位正方形。例如,将 (0,0)(1,0)(1,1)(0,1) 和 (1,0)(2,0)(2,1)(1,1) 合并成一个 $2 imes 1$ 的矩形。
原矩形 (0,0)(1,0)(1,1)(0,1): $I=0, B=4$. $A = 0 + 4/2 1 = 1$.
原矩形 (1,0)(2,0)(2,1)(1,1): $I=0, B=4$. $A = 0 + 4/2 1 = 1$.
合并后的 $2 imes 1$ 矩形 (0,0)(2,0)(2,1)(0,1):
边界格点:(0,0), (1,0), (2,0), (2,1), (1,1), (0,1)。共 6 个。
内部格点:0 个。
面积:2。
皮克公式:$A = 0 + 6/2 1 = 3 1 = 2$. 仍然成立。
让我们分析一下合并过程中 $I$ 和 $B$ 的变化:
边界格点 (B): 当两个多边形合并时,它们共用的那条边上的格点,原来在两个多边形边界上各算一次,合并后只在新的边界上算一次。如果这条共用边上有 $k$ 个格点,那么 $B$ 的总数会减少 $k$(原来是 $B_1 + B_2$,现在是 $(B_1 k) + (B_2 k) + k = B_1 + B_2 k$)。
内部格点 (I): 合并不会产生新的内部格点,也不会让内部格点变成边界格点,也不会让边界格点变成内部格点。所以 $I$ 的变化是 $I_{new} = I_1 + I_2$。
将这些代入皮克公式:
$A_{new} = (I_1 + I_2) + frac{B_1 + B_2 k}{2} 1$
$A_{new} = (I_1 + frac{B_1}{2} 1) + (I_2 + frac{B_2}{2} 1) + frac{k}{2} + 1$
$A_{new} = A_1 + A_2 + frac{k}{2} + 1 2 + 1 = A_1 + A_2 + frac{k}{2}$
这里出现了一个问题。我们预期 $A_{new} = A_1 + A_2$。问题出在哪里?
仔细审视共用边:
当两个单位正方形合并时,它们共用的边是 (1,0)(1,1) 这条线段。这条线段上的格点是 (1,0) 和 (1,1),共 2 个。所以 $k=2$。
矩形 1: $I_1=0, B_1=4$. $A_1=1$.
矩形 2: $I_2=0, B_2=4$. $A_2=1$.
合并后: $I_{new}=0, B_{new}=6$. $A_{new}=2$.
根据公式:$A_{new} = 0 + 6/2 1 = 2$.
我们的分析 $A_{new} = A_1 + A_2 + k/2$ 是错误的,是因为我们在考虑 $I$ 和 $B$ 的变化时,没有直接代入新的公式,而是试图从旧公式推导出新公式。
正确的分析应该关注公式本身的结构:
设 $P_1$ 和 $P_2$ 是两个多边形,它们共用一条边 $e$。
$A_1 = I_1 + B_1/2 1$
$A_2 = I_2 + B_2/2 1$
$P_{new}$ 是 $P_1$ 和 $P_2$ 合并后的多边形。
$I_{new} = I_1 + I_2$ (合并不会改变内部格点)
$B_{new} = B_1 + B_2 2k'$, 其中 $k'$ 是共用边(不包括端点)上的格点数。因为共用边上的端点格点也只被计算一次,所以更准确的说法是,共用边上除了两个端点之外的点,原本在 $P_1$ 的边界上,现在不属于 $P_{new}$ 的边界;原本在 $P_2$ 的边界上,现在也不属于 $P_{new}$ 的边界。而边上的两个端点,虽然在 $P_1$ 和 $P_2$ 的边界上都被计算,但在 $P_{new}$ 的边界上也只算一次。
所以,如果共用边上(包括端点)总共有 $k$ 个格点,那么 $B_{new} = B_1 + B_2 k$。 (例如,单位正方形的边 (0,0)(1,0),格点是 (0,0) 和 (1,0),共 2 个。合并时,这两点在 $P_{new}$ 的边界上仍然是两个点,但它们分别属于 $P_1$ 和 $P_2$ 的边界。所以 $B_{new} = B_1 + B_2 2$)。
代入皮克公式:
$A_{new} = I_{new} + B_{new}/2 1$
$A_{new} = (I_1 + I_2) + (B_1 + B_2 k)/2 1$
$A_{new} = I_1 + I_2 + B_1/2 + B_2/2 k/2 1$
$A_{new} = (I_1 + B_1/2 1) + (I_2 + B_2/2 1) + k/2 + 1$
$A_{new} = A_1 + A_2 + k/2$
等一下,我发现了一个关键点: 在计算 $B$ 的时候,我们只计算了“顶点”上的格点,而没有计算“边”上的格点。 皮克定理的 $B$ 指的是多边形边界上的格点,包括顶点和边上的格点。
让我们重新审视单位正方形 (0,0)(1,0)(1,1)(0,1):
边界格点:(0,0), (1,0), (1,1), (0,1)。 $B=4$.
内部格点:0. $I=0$.
$A = 0 + 4/2 1 = 1$.
考虑合并两个单位正方形,形成 $2 imes 1$ 矩形 (0,0)(2,0)(2,1)(0,1):
边界格点:(0,0), (1,0), (2,0), (2,1), (1,1), (0,1)。 $B=6$.
内部格点:0. $I=0$.
$A = 0 + 6/2 1 = 2$.
现在,让我们分析合并过程:
第一个单位正方形:$I_1=0, B_1=4$.
第二个单位正方形:$I_2=0, B_2=4$.
共用边是 (1,0)(1,1),这条边上的格点是 (1,0) 和 (1,1)。
合并后:$I_{new}=0$.
边界格点 $B_{new}$:
$P_1$ 的边界格点:(0,0), (1,0), (1,1), (0,1)
$P_2$ 的边界格点:(1,0), (2,0), (2,1), (1,1)
合并后 $P_{new}$ 的边界格点:(0,0), (1,0), (2,0), (2,1), (1,1), (0,1)
注意到,(1,0) 和 (1,1) 这两个点,在 $P_1$ 和 $P_2$ 的边界上各出现一次,但在 $P_{new}$ 的边界上,它们仍然是两个点,而且是 $P_{new}$ 的顶点。
实际情况是,当两个多边形沿着一条边合并时,这条边上的格点(包括端点)在合并后仍然是 $P_{new}$ 的边界的一部分。
如果共用边上有 $k$ 个格点,那么 $B_{new} = B_1 + B_2 2$。因为这条边上的两个端点是必须被计算的,它们在 $P_1$ 和 $P_2$ 中各算一次,在 $P_{new}$ 中也算一次。所以,我们丢弃的是这条边除了端点以外的所有格点。
如果共用边上的格点数是 $k$,那么 $B_{new} = B_1 + B_2 (k2)$? 不对。
关键在于: 当两个多边形合并时,它们的“内边界”被消除了。假设它们共享一条由 $k$ 个格点组成的线段(包含两个端点)。那么,这 $k$ 个格点本来在 $P_1$ 的边界上,也在 $P_2$ 的边界上。合并后,这 $k$ 个点不再是“外部”边界的一部分,而是“内部”相连。然而,皮克定理的 $B$ 是指“外部”边界上的格点。
正确的逻辑是: 设 $P_1$ 和 $P_2$ 共享一条边 $e$。$e$ 上有 $k$ 个格点。
$B_1$ 是 $P_1$ 的边界格点总数。
$B_2$ 是 $P_2$ 的边界格点总数。
$I_1$ 是 $P_1$ 的内部格点总数。
$I_2$ 是 $P_2$ 的内部格点总数。
$P_{new}$ 是 $P_1$ 和 $P_2$ 合并后的多边形。
$I_{new} = I_1 + I_2$。
$B_{new} = B_1 + B_2 2 imes ( ext{共用边上的非端点格点数})$。
如果共用边上有 $k$ 个格点,那么非端点格点数是 $k2$。
所以,$B_{new} = B_1 + B_2 2(k2)$。
代入皮克公式:
$A_{new} = I_{new} + B_{new}/2 1$
$A_{new} = (I_1 + I_2) + (B_1 + B_2 2(k2))/2 1$
$A_{new} = I_1 + I_2 + B_1/2 + B_2/2 (k2) 1$
$A_{new} = (I_1 + B_1/2 1) + (I_2 + B_2/2 1) + k 2 + 2 1$
$A_{new} = A_1 + A_2 + k 1$
这仍然不等于 $A_1 + A_2$!
让我们回到最根本的定义: 这里的 $B$ 是指多边形边界上的格点。
重新审视合并的边界格点计数:
当两个多边形 $P_1$ 和 $P_2$ 沿着一条边 $e$ 合并时,$e$ 上的所有格点(包括端点)将不再是外部边界的一部分。
$P_1$ 的边界格点有 $B_1$ 个。
$P_2$ 的边界格点有 $B_2$ 个。
$e$ 上有 $k$ 个格点。
$P_{new}$ 的边界格点 $B_{new}$ 是 $P_1$ 的边界格点除去 $e$ 上的格点,加上 $P_2$ 的边界格点除去 $e$ 上的格点。
所以,$B_{new} = (B_1 k) + (B_2 k) = B_1 + B_2 2k$.
$I_{new} = I_1 + I_2$.
代入皮克公式:
$A_{new} = I_{new} + B_{new}/2 1$
$A_{new} = (I_1 + I_2) + (B_1 + B_2 2k)/2 1$
$A_{new} = I_1 + I_2 + B_1/2 + B_2/2 k 1$
$A_{new} = (I_1 + B_1/2 1) + (I_2 + B_2/2 1) + k + 2 k 1$
$A_{new} = A_1 + A_2 + 1$
这还是不对。我的 $B$ 的定义或者合并时的 $B$ 的变化计算可能有误。
换一个角度思考: 沿着一条边(包含 $k$ 个格点)合并两个多边形。
$I_{new} = I_1 + I_2$
$B_{new} = B_1 + B_2 k$. (因为共用边上的 $k$ 个点,在 $P_1$ 和 $P_2$ 的边界上各算一次,合并后它们作为连接线,不再是外部边界,所以要从总数中减去)。
代入皮克公式:
$A_{new} = I_{new} + B_{new}/2 1$
$A_{new} = (I_1 + I_2) + (B_1 + B_2 k)/2 1$
$A_{new} = I_1 + I_2 + B_1/2 + B_2/2 k/2 1$
$A_{new} = (I_1 + B_1/2 1) + (I_2 + B_2/2 1) + k/2 + 1$
$A_{new} = A_1 + A_2 + k/2$.
这个结果依然让我困惑。
另一个证明思路:欧拉公式
皮克定理有一个更深刻的证明,它与多面体的欧拉公式 ($VE+F=2$) 有着密切的联系,尤其是在平面图论的背景下。
考虑一个平面连通图 G,其顶点集为 $V(G)$,边集为 $E(G)$,面集为 $F(G)$。欧拉公式告诉我们 $v e + f = 1$ (对于平面图,去掉无限面)。
现在,我们把多边形内部和边界的格点看作是图的顶点。多边形边界上的线段是边,内部的线段也是边。
更严谨的拆分证明(基于三角形):
任何一个顶点在格点上的简单多边形,都可以被三角剖分成若干个顶点都在格点上的三角形。如果皮克定理对于所有顶点在格点上的三角形都成立,那么通过归纳法,它对于任意顶点在格点上的简单多边形也成立。
证明皮克定理对三角形成立:
设一个三角形的三个顶点是 $A, B, C$。
$A = (x_1, y_1), B = (x_2, y_2), C = (x_3, y_3)$,其中 $x_i, y_i$ 都是整数。
三角形的面积可以用行列式表示:$A = frac{1}{2} |x_1(y_2 y_3) + x_2(y_3 y_1) + x_3(y_1 y_2)|$。
由于所有坐标都是整数,这个面积 $A$ 必然是整数或者半整数(即 $n/2$ 的形式,其中 $n$ 是整数)。
根据皮克定理,$A = I + B/2 1$。
$2A = 2I + B 2$。
由于 $2A$ 是整数,所以 $2I + B 2$ 必须是整数,这是满足的。
现在的问题是,$I$ 和 $B$ 的值是否能保证 $I + B/2 1$ 等于三角形的精确面积。
关键是证明: 对于任何一个顶点在格点上的三角形,其面积 $A$ 满足 $2A B + 2$ 是一个非负偶数。
Pick's Theorem Proof (via Lattice Triangles):
Let $T$ be a triangle with vertices at lattice points.
Its area $A$ is given by the formula:
$A = frac{1}{2} |det(vec{AB}, vec{AC})|$
If $A = (x_1, y_1)$, $B = (x_2, y_2)$, $C = (x_3, y_3)$, then
$vec{AB} = (x_2 x_1, y_2 y_1)$
$vec{AC} = (x_3 x_1, y_3 y_1)$
$A = frac{1}{2} |(x_2 x_1)(y_3 y_1) (x_3 x_1)(y_2 y_1)|$
Since all coordinates are integers, the determinant is an integer. Therefore, $2A$ is an integer. This means $A$ is either an integer or a halfinteger.
Now consider the formula $A = I + B/2 1$.
Rearranging gives $2A = 2I + B 2$.
Since $2A$ is an integer, $2I + B 2$ must be an integer. This is always true as $I$ and $B$ are integers.
What we need to show is that for any lattice triangle $T$, the value $I + B/2 1$ correctly computes its area. This relies on showing that $I + B/2 1$ is always equal to the geometric area.
A common approach is to show that the formula holds for a primitive triangle (a triangle with $I=0$ and $B=3$, i.e., no lattice points on its edges except for the vertices) and then demonstrate that adding a lattice point to an edge or an interior point preserves the validity of the formula.
Let's consider a primitive triangle $T$.
$B = 3$ (the three vertices).
$I = 0$.
Area $A = 0 + 3/2 1 = 1/2$.
A primitive triangle indeed has an area of $1/2$. Think of a triangle with vertices (0,0), (1,0), (0,1). Its area is $1/2$. $B=3$ (the vertices), $I=0$. $A = 0 + 3/2 1 = 1/2$.
Proof by Induction on the Number of Lattice Points:
Base Case: A primitive triangle has $I=0, B=3$. $A = 0 + 3/2 1 = 1/2$. This is true.
Inductive Step:
Assume Pick's Theorem holds for all polygons with fewer than $N$ lattice points in total ($I+B$). We want to show it holds for a polygon $P$ with $N$ lattice points.
If $P$ is a triangle, we've shown it for primitive triangles. If it's not primitive, it must have points on its edges or inside.
Consider adding a point to a polygon:
1. Adding a point on an edge:
Suppose we have a polygon $P$ and we add a lattice point $Q$ on one of its edges, say $PQ$.
Let $P$ have area $A$, $I$ internal points, $B$ boundary points. $A = I + B/2 1$.
Now, we can imagine this edge as being "split" by $Q$.
The new polygon $P'$ has the same internal points $I$.
The boundary points increase by 1: $B' = B+1$.
The new area $A'$ should be $A$.
Applying Pick's formula to $P'$: $A' = I + (B+1)/2 1 = I + B/2 + 1/2 1 = (I + B/2 1) + 1/2 = A + 1/2$.
This approach of "adding a point on an edge" isn't about changing the polygon itself, but conceptually dividing an edge.
The more standard inductive proof is on the area:
Base Case: The smallest possible area for a lattice polygon is $1/2$, achieved by a primitive triangle. $I=0, B=3$. $A = 0 + 3/2 1 = 1/2$. This holds.
Inductive Step: Assume Pick's Theorem holds for all lattice polygons with area less than $A$. Consider a lattice polygon $P$ with area $A$.
If $P$ is a triangle, we need to show it.
If $P$ is not a triangle, it must have at least one internal lattice point $Q$.
Let $Q$ be an internal lattice point. Connect $Q$ to some vertices of $P$ to form new lattice polygons.
A better approach: Find an edge of $P$ that contains at least one lattice point besides its endpoints. Let this edge be $PQ$, where $P$ and $Q$ are vertices of the polygon. Let $R$ be a lattice point on the segment $PQ$ (distinct from $P$ and $Q$).
We can split the polygon $P$ into two smaller lattice polygons $P_1$ and $P_2$ by drawing the segment $QR$ and $PR$ if $R$ is not already a vertex. This isn't simple.
Let's revisit the merging of polygons, but focusing on the formula itself.
Consider two lattice polygons $P_1$ and $P_2$ that share a common edge $e$. Let $e$ contain $k$ lattice points.
$A_1 = I_1 + B_1/2 1$
$A_2 = I_2 + B_2/2 1$
$P_{new}$ is formed by their union.
$I_{new} = I_1 + I_2$.
$B_{new} = B_1 + B_2 k$. (The $k$ points on the common edge are no longer external boundary points).
$A_{new} = I_{new} + B_{new}/2 1$
$A_{new} = (I_1 + I_2) + (B_1 + B_2 k)/2 1$
$A_{new} = I_1 + I_2 + B_1/2 + B_2/2 k/2 1$
$A_{new} = (I_1 + B_1/2 1) + (I_2 + B_2/2 1) + k/2 + 1$
$A_{new} = A_1 + A_2 + k/2$.
This still suggests a discrepancy. The error lies in the definition of $B$ or how it changes.
Correct Merging Argument:
When two polygons $P_1$ and $P_2$ are joined along a common edge $e$ with $k$ lattice points, the total number of boundary points $B_{new}$ is:
$B_{new} = (B_1 k) + (B_2 k) = B_1 + B_2 2k$. This is because the $k$ points on the shared edge are no longer part of the external boundary.
Let's resubstitute with this corrected $B_{new}$:
$A_{new} = I_{new} + B_{new}/2 1$
$A_{new} = (I_1 + I_2) + (B_1 + B_2 2k)/2 1$
$A_{new} = I_1 + I_2 + B_1/2 + B_2/2 k 1$
$A_{new} = (I_1 + B_1/2 1) + (I_2 + B_2/2 1) + k + 2 k 1$
$A_{new} = A_1 + A_2 + 1$
This result $(A_{new} = A_1 + A_2 + 1)$ is still not $A_{new} = A_1 + A_2$.
Where is the persistent error?
The error comes from the assumption that $I_{new} = I_1 + I_2$ and $B_{new}$ calculation alone preserves the formula's structure.
Let's reconsider the very first step: the unit square.
$A=1, I=0, B=4$. Formula: $0 + 4/2 1 = 1$. OK.
Consider two unit squares side by side: $2 imes 1$ rectangle.
$A=2, I=0, B=6$. Formula: $0 + 6/2 1 = 2$. OK.
Let $P_1$ be the left square, $P_2$ be the right square.
$A_1=1, I_1=0, B_1=4$.
$A_2=1, I_2=0, B_2=4$.
They share an edge with $k=2$ lattice points: (1,0) and (1,1).
$I_{new} = I_1 + I_2 = 0+0=0$.
$B_{new} = B_1 + B_2 2k = 4 + 4 2(2) = 8 4 = 4$. This is wrong! The new boundary points are 6.
My $B_{new}$ calculation is flawed.
The actual change in boundary points:
When we merge $P_1$ and $P_2$ along edge $e$, the $k$ points on $e$ are no longer on the outer boundary. So the new boundary is the union of the boundary of $P_1$ excluding $e$, and the boundary of $P_2$ excluding $e$.
$B_{new} = (B_1 k) + (B_2 k) = B_1 + B_2 2k$. This IS the definition of the new boundary points.
Let's reverify the example:
Square 1: (0,0)(1,0)(1,1)(0,1). $B_1=4$.
Square 2: (1,0)(2,0)(2,1)(1,1). $B_2=4$.
Shared edge: (1,0)(1,1). $k=2$.
Merged rectangle: (0,0)(2,0)(2,1)(0,1). $B_{new}=6$.
Here, $B_{new} = B_1 + B_2 2k = 4 + 4 2(2) = 4$. This is where the calculation fails to match the geometric reality. The issue is that the points (1,0) and (1,1) ARE part of the new boundary!
The error lies in interpreting "boundary points" in the merging step.
When $P_1$ and $P_2$ merge along $e$, the $k$ points on $e$ are no longer "on the boundary" of $P_1$ and $P_2$ separately. But they become part of the boundary of $P_{new}$.
The correct way to think about it: Pick's Theorem is invariant under certain operations.
We can prove Pick's Theorem by showing that it is invariant under two operations:
1. Adding an internal lattice point: If we add a lattice point $Q$ inside a polygon $P$, we can divide $P$ into smaller polygons by connecting $Q$ to some vertices. This is complex.
2. Splitting an edge by adding a lattice point: If we have an edge $AB$ of a polygon, and we introduce a lattice point $C$ on the segment $AB$, this splits the edge $AB$ into $AC$ and $CB$. This is usually done in the context of proving by induction.
A more direct proof for triangles (and thus for any polygon by triangulation):
Let the vertices of a lattice triangle be $A, B, C$.
The area of the triangle can be computed as $A = frac{1}{2} sum_{i=1}^n det(vec{v_i}, vec{v_{i+1}})$ where $v_i$ are vertices and $v_{n+1}=v_1$. For a triangle, this is $frac{1}{2} (det(vec{AB}, vec{AC}))$.
The Barycentric Coordinates approach can also be used. Any lattice point $P$ inside a lattice triangle $T$ can be expressed as $P = uA + vB + wC$, where $u+v+w=1$ and $u,v,w ge 0$. For $P$ to be a lattice point, $u,v,w$ must be rational.
A common and intuitive proof method uses the concept of “boundary traversal”:
Imagine walking along the boundary of the polygon. Each step from one lattice point to the next on the boundary contributes to $B$.
The formula can be rewritten as $A + 1 = I + B/2$.
Consider the effect of a single step on the boundary:
Let $P$ be a lattice polygon. We can decompose $P$ into unit squares and triangles.
A more direct proof relies on the fact that any lattice polygon can be transformed into a unit square by a sequence of areapreserving operations that also preserve the $I + B/2 1$ value. These operations include:
1. Sliding a vertex along an edge: If a vertex $V$ is moved parallel to an edge $PQ$ to a new position $V'$, such that the triangle $PVQ$ remains a lattice triangle, the area and $I + B/2 1$ are preserved.
2. Sliding an edge: If an edge $AB$ is slid parallel to itself to a new position $A'B'$ while maintaining lattice vertices, this preserves $I$ and $B$ properties.
This is the core idea behind a proof by "shearing transformations".
A shear transformation preserves area. If applied to a lattice polygon, it maps lattice points to lattice points.
A horizontal shear: $(x, y) mapsto (x + cy, y)$.
A vertical shear: $(x, y) mapsto (x, y + cx)$.
These transformations preserve the number of internal and boundary lattice points if $c$ is an integer.
The strategy is to show that any lattice polygon can be transformed into a unit square using a sequence of such operations that maintain the value of $I + B/2 1$.
Step 1: Reduce to a shape with no "internal" points on edges.
If an edge $AB$ has lattice points between $A$ and $B$, we can "flatten" this edge by performing a shear. Imagine an edge on the line $y=c$. If we shear horizontally by $y$, the points $(x, c)$ move to $(x+cc, c)$. This can reduce the number of intermediate points on an edge to zero.
Step 2: Transform into a rectangle.
After ensuring no intermediate lattice points on edges, we can use shears to transform the polygon into a rectangle with sides parallel to the axes. This is possible because the slopes of the edges are rational.
Step 3: Transform the rectangle into a unit square.
A rectangle with vertices $(0,0), (W,0), (W,H), (0,H)$ has area $W imes H$.
The number of boundary points is $2W + 2H$.
The number of internal points is $(W1)(H1)$.
Applying Pick's formula: $I + B/2 1 = (W1)(H1) + (2W+2H)/2 1$
$= (WH W H + 1) + (W+H) 1$
$= WH W H + 1 + W + H 1 = WH$.
This shows that Pick's theorem holds for any rectangle aligned with the axes.
Since any lattice polygon can be transformed into a unit square via areapreserving operations that preserve $I$ and $B$, and the unit square satisfies Pick's theorem, the theorem must hold for all lattice polygons.
Summary of the Shear Proof:
1. Acknowledge Base Case: Unit square works.
2. Show Invariance: Demonstrate that shear transformations preserve area and the value $I + B/2 1$. Crucially, integer shears map lattice points to lattice points.
3. Polygon Reduction: Show that any lattice polygon can be transformed into a rectangle aligned with the axes using these shears.
First, eliminate points on edges (except vertices) using shears.
Then, align the remaining boundary segments parallel to axes using more shears.
4. Rectangle Case: Verify Pick's theorem for axisaligned rectangles.
5. Conclusion: Since any lattice polygon can be reduced to a rectangle (which satisfies the theorem) via transformations that preserve the theorem's quantity, and the rectangle itself satisfies the theorem, the theorem holds for the original polygon.
This shear transformation proof is quite powerful and elegant, but requires understanding of linear algebra and geometric transformations.
Alternative proof focusing on boundary traversal and orientation:
This method often involves Green's Theorem in a discrete form.
Green's theorem relates a line integral around a simple closed curve $C$ to a double integral over the plane region $D$ bounded by $C$:
$oint_C (L , dx + M , dy) = iint_D (frac{partial M}{partial x} frac{partial L}{partial y}) , dA$.
For area, we can choose $L = y/2$ and $M = x/2$. Then $frac{partial M}{partial x} frac{partial L}{partial y} = frac{1}{2} (frac{1}{2}) = 1$.
So, $A = iint_D 1 , dA = oint_C (frac{y}{2} , dx + frac{x}{2} , dy)$.
In a discrete setting, the line integral becomes a sum over the lattice segments of the boundary.
If we consider a directed segment from $(x_1, y_1)$ to $(x_2, y_2)$, then $Delta x = x_2 x_1$ and $Delta y = y_2 y_1$.
The contribution to the integral from this segment is $frac{y_1+y_2}{2} Delta x + frac{x_1+x_2}{2} Delta y$.
This sum over all boundary segments can be shown to be equal to $I + B/2 1$.
This proof is more advanced and relies on discrete calculus.
Conclusion:
The beauty of Pick's Theorem lies not only in its simple statement but also in the diverse mathematical tools that can be employed for its proof. From fundamental geometric decomposition and induction, to the abstract power of shear transformations and discrete calculus, each proof offers a unique perspective on the theorem's validity. The most intuitive for many might be the triangulation approach combined with proving it for basic triangles, or the geometric transformation argument which shows that all lattice polygons can be reduced to a unit square, thus inheriting its properties.
While the shear transformation proof is arguably the most rigorous and general, the idea of building up from simple shapes and proving invariance under operations is a cornerstone of mathematical discovery. The theorem stands as a testament to the deep connections between algebra, geometry, and combinatorics.