当前课程知识点:计算几何 > 07. Geometric Range Search > E. kd-Tree: Performance > 07-E-03. Beyond 2D
当然作为一种非常有名的数据结构
kd树自然有它的存在价值
它有它更擅长的应用范围
range query
并不是它最擅长的部分
那么我们这里需要指出的是
kd树实际上
本身就不限于是二维的
在很多文献中
你可能会看到类似于这样的描述
也就是k-D kd-tree
或者具体的来说
它会用到2D kd-tree
我们说其实这种表述方法
固然没有错
但是不够简洁
其实最开始在命名这个kd-tree的时候
这个kd其实指的就是k dimensional
所以如果是二维的情况
你不妨简单的就称之为
2d-Tree就可以了
当然三维的情况
你也知道叫做3d-Tree
我们说这样的描述会更简洁一些
回到range query的问题
实际上运用kd树来求解range query
也是一种解决的方案
至少是一种很可行的方案
它只不过是对刚才二维的情况
做一个变形和推广而已
区别就在于
我们刚才只是在x轴和y轴
两个方向来回的切换
切分的方向
如果是有k维
或者叫做一般来讲
有d维的话
我们的切分的方向
依然是轮回
只不过这个轮值的周期
会长一些
如果你有d维的话
那么它的轮值的周期就是d
首先对于第一维
再做第二维划分
再做第三维划分
如此下去
当所有的d维都穷举过一遍以后
又重新的回到1 2再做轮回
仅此区别而已
那么这里需要提的一条是
它的复杂度
也就是说
我们如果用这么样一种数据结构
kd-tree来求解这个range query
尤其是高维的问题的话
那么它的时间复杂度
尤其是我们看中的query time
是n^((d-1)/d)
没错
也就是说这个时候的查询的时间
主体是n^((d-1)/d)
这个结果虽然是
总是小于n的一次方的
但是随着维度d的增加
尤其是快速的增加到一定数量之后
它就很快的会趋近于多少呢
你已经看出来了
就是n的一次方
什么意思
当维度足够之高以后
用kd树来求解range query
这个问题的算法的时间效率
几乎退化为最初的蛮力算法
-Before we start
--html
-Evaluation
--html
-Online Judge
--html
-Lecture notes
--html
-Discussion
--html
-A. History of This Course
--00-A. History of This Course
-B. What's Computational Geometry
--00-B. What's Computational Geometry
-B. What's Computational Geometry--作业
-C. How to Learn CG Better
--00-C. How to Learn CG Better
-C. How to Learn CG Better--作业
-D. Why English
-A. Convexity
-A. Convexity--作业
-B. Extreme Points
-B. Extreme Points--作业
-C. Extreme Edges
-C. Extreme Edges--作业
-D. Incremental Construction
--01-D-01. Decrease and Conquer
--01-D-02. In-Convex-Polygon Test
--01-D-03. Why Not Binary Search
-D. Incremental Construction--作业
-E. Jarvis March
--01-E-06. Lowest-Then-Leftmost
-E. Jarvis March--作业
-F. Lower Bound
--01-F-02. CAO Chong's Methodology
-F. Lower Bound--作业
-G. Graham Scan: Algorithm
-G. Graham Scan: Algorithm--作业
-H. Graham Scan: Example
-H. Graham Scan: Example--作业
-I. Graham Scan: Correctness
-I. Graham Scan: Correctness--作业
-J. Graham Scan: Analysis
-J. Graham Scan: Analysis--作业
-K. Divide-And-Conquer (1)
-K. Divide-And-Conquer (1)--作业
-L. Divide-And-Conquer (2)
--01-L-03. Topmost + Bottommost ?
--01-L-07. More Considerations
-L. Divide-And-Conquer (2)--作业
-M. Wrap-Up
-0. Introduction
-0. Introduction--作业
-A. Preliminary
-A. Preliminary--作业
-B. Interval Intersection Detection
-B. Interval Intersection Detection--作业
-C. Segment Intersection Reporting
-C. Segment Intersection Reporting--作业
-D. BO Algorithm: Strategy
--02-D-01. Proximity & Separability
--02-D-02. Comparability & Ordering
-D. BO Algorithm: Strategy--作业
-E. BO Algorithm: Implementation
--02-E-03. Events & Operations
-E. BO Algorithm: Implementation--作业
-F. BO Algorithm: Analysis
--02-F-04. Complexity of Event Queue
--02-F-05. Complexity of Status Structure
-F. BO Algorithm: Analysis--作业
-G. Convex Polygon Intersection Detection
--02-G-01. Problem Specification
--02-G-02. Monotone Partitioning
--02-G-04. Decrease-And-Conquer
-G. Convex Polygon Intersection Detection--作业
-H. Edge Chasing
--02-H-01. Eliminating Sickles
-H. Edge Chasing--作业
-I. Plane Sweeping
-I. Plane Sweeping--作业
-J. Halfplane Intersection Construction
-J. Halfplane Intersection Construction--作业
-0. Methodology
-0. Methodology--作业
-A. Art Gallery Problem
--03-A-02. Lower & Upper Bounds
--03-A-04. Approximation & Classification
-A. Art Gallery Problem--作业
-B. Art Gallery Theorem
--03-B-01. Necessity of floor(n/3)
--03-B-02. Sufficiency by Fan Decomposition
-B. Art Gallery Theorem--作业
-C. Fisk's Proof
--03-C-04. Pigeon-Hole Principle
-C. Fisk's Proof--作业
-D. Orthogonal Polygons
--03-D-01. Necessity of floor(n/4)
--03-D-02. Sufficiency by Convex Quadrilateralization
-D. Orthogonal Polygons--作业
-E. Triangulation
-E. Triangulation--作业
-F. Triangulating Monotone Polygons
--03-F-02. Monotonicity Testing
--03-F-04. Stack-Chain Consistency
-F. Triangulating Monotone Polygons--作业
-G. Monotone Decomposition
-G. Monotone Decomposition--作业
-I. Tetrahedralization
--03-I-01. Polyhedron Decomposition
--03-I-02. Schonhardt's Polyhedron
-I. Tetrahedralization--作业
-A. Introduction
--04-A-02. Dining Halls on Campus
--04-A-03. More Analogies & Applications
-A. Introduction--作业
-B. Terminologies
--04-B-02. Intersecting Halfspaces
--04-B-04. Planar Voronoi Diagram
-B. Terminologies--作业
-C. Properties
--04-C-03. Nearest = Concyclic
--04-C-04. Number of Nearest Sites = Degree
-C. Properties--作业
-D. Complexity
-D. Complexity--作业
-E. Representation
-E. Representation--作业
-F. DCEL
-F. DCEL--作业
-G. Hardness
--04-G-03. Voronoi Diagram In General Position
-G. Hardness--作业
-H. Sorted Sets
--04-H-01. Convex Hull Made Easier
--04-H-02. Convex Hull As A Combinatorial Structure
--04-H-03. Voronoi Diagram As A Geometric Structure
-H. Sorted Sets--作业
-I. VD_sorted
--04-I-06. Sorting Not Made Easier
-I. VD_sorted--作业
-J. Naive Construction
-J. Naive Construction--作业
-K. Incremental Construction
-K. Incremental Construction--作业
-L. Divide-And-Conquer
--04-L-09. Intersecting with Cells
-L. Divide-And-Conquer--作业
-M. Plane-Sweep
--04-M-09. Circle Event: What, When & Where
-M. Plane-Sweep--作业
-A. Point Set Triangulation
-A. Point Set Triangulation--作业
-B. Delaunay Triangulation
-B. Delaunay Triangulation--作业
-C. Properties
-C. Properties--作业
-D. Proximity Graph
--05-D-02. Relative Neighborhood Graph
-D. Proximity Graph--作业
-E. Euclidean Minimum Spanning Tree
-E. Euclidean Minimum Spanning Tree--作业
-F. Euclidean Traveling Salesman Problem
-G. Minimum Weighted Triangulation
-G. Minimum Weighted Triangulation--作业
-H. Construction
--05-H-03. Maximizing The Minimum Angle
--05-H-04. Evolution By Edge Flipping
-H. Construction--作业
-I. RIC With Example
-I. RIC With Example--作业
-J. Randomized Incremental Construction
--05-J-01. Recursive Implementation
--05-J-02. Iterative Implementation
-J. Randomized Incremental Construction--作业
-K. RIC Analysis
--05-K-04. Types Of Edge Change
--05-K-05. Number Of Edge Changes
--05-K-07. Number Of Rebucketings
--05-K-08. Probability For Rebucketing
--05-K-10. Further Consideration
-0. Online/Offline Algorithms
--06-0. Online/Offline Algorithms
-0. Online/Offline Algorithms--作业
-A. Introduction
--06-A-03. Assumptions For Clarity
--06-A-05. Performance Measurements
-A. Introduction--作业
-B. Slab Method
--06-B-02. Ordering Trapezoids
-B. Slab Method--作业
-C. Persistence
--06-C-01. Ephemeral Structure
--06-C-02. Persistent Structure
-C. Persistence--作业
-D. Path Copying
--06-D-03. Storage Optimization
-D. Path Copying--作业
-E. Node Copying
-E. Node Copying--作业
-F. Limited Node Copying
-G. Kirkpatrick Structure
--06-G-01. Optimal And Simpler
--06-G-06. The More The Better
--06-G-07. The Fewer The Better
--06-G-09. Existence Of Independent Subset
--06-G-10. Construction Of Independent Subset
-G. Kirkpatrick Structure--作业
-H. Trapezoidal Map
--06-H-03. Properties & Complexity
--06-H-04. Search Structure: Example
--06-H-05. Search Structure: Nodes
--06-H-06. Search Structure: Performance
-H. Trapezoidal Map--作业
-I. Constructing Trapezoidal Map
--06-I-04. Case 1: Two Endpoints
--06-I-05. Case 2: One Endpoint
--06-I-06. Case 3: No Endpoints
-J. Performance Of Trapezoidal Map
--06-J-03. Number Of Ray Trimmed
--06-J-04. Number Of Trapezoidals Created (1)
--06-J-05. Number Of Trapezoidals Created (2)
--06-J-06. Time For Point Location
--06-J-07. Size Of Search Structure
--06-J-08. Fixed Query Point + Randomly Created Maps
--06-J-10. Probability Of Enclosing Trapezoid Changed
-A. Range Query
--07-A-01. 1-Dimensional Range Query
-A. Range Query--作业
-B. BBST
--07-B-02. Lowest Common Ancestor
-B. BBST--作业
-C. kd-Tree: Structure
-C. kd-Tree: Structure--作业
-D. kd-Tree: Algorithm
-D. kd-Tree: Algorithm--作业
-E. kd-Tree: Performance
--07-E-01. Preprocessing Time + Storage
-E. kd-Tree: Performance--作业
-F. Range Tree: Structure
--07-F-03. x-Query * y-Queries
-F. Range Tree: Structure--作业
-G. Range Tree: Query
-G. Range Tree: Query--作业
-H. Range Tree: Performance
-H. Range Tree: Performance--作业
-I. Range Tree: Optimization
--07-I-04. Fractional Cascading
-A. Orthogonal Windowing Query
-A. Orthogonal Windowing Query--作业
-B. Stabbing Query
-C. Interval Tree: Construction
-C. Interval Tree: Construction--作业
-D. Interval Tree: Query
-D. Interval Tree: Query--作业
-E. Stabbing With A Segment
--08-E-03. Query Algorithm (1)
--08-E-04. Query Algorithm (2)
-F. Grounded Range Query
--08-F-03. 1D-GRQ Using Range Tree
--08-F-04. 1D-GRQ By Linear Scan
-G. 1D-GRQ Using Heap
-G. 1D-GRQ Using Heap--作业
-H. Priority Search Tree
--08-H-03. Sibling Partitioning
-H. Priority Search Tree--作业
-I. 2D-GRQ Using PST
-I. 2D-GRQ Using PST--作业
-J. Segment Tree
--08-J-01. General Windowing Query
--08-J-02. Elementary Interval
--08-J-06. Solving Stabbing Query
--08-J-11. Constructing A Segment Tree
--08-J-12. Inserting A Segment (1)
--08-J-13. Inserting A Segment (2)
--08-J-14. Inserting A Segment (3)
-K. Vertical Segment Stabbing Query