gdsu — Generic Disjoint Set Union (Union-Find)

gdsu is a modern, type-safe, composable Disjoint Set Union (DSU) / Union-Find Go library.
It provides a clean DSU interface with two interchangeable implementations:

• Sparse DSU — generic, map-backed, supports any comparable type, grows dynamically
• Compact DSU — high-performance integer-indexed version using slices, fixed capacity

This library is designed for algorithmic workloads, data pipelines, graph theory, clustering, and any use-case involving connectivity queries.

---

1. DSU Interface

Located in: gdsu.go

type DSU[T comparable] interface {
    Find(x T) T            // Find returns the representative element (root) of x
    Union(x, y T) bool     // Union merges the sets containing x and y
    Connected(x, y T) bool // Connected reports whether x and y are in the same set
    IsMember(x T) bool     // IsMember reports whether x is present (no mutation)
    Groups() map[T][]T     // Groups returns all connected components
}

Both implementations — sparse and compact — satisfy this interface.

---

2. Sparse DSU (Generic, Map-Based, Dynamic)

Located in: sparse/sparse.go

Key Features

• Works with any comparable type: strings, integers, structs, user-defined keys
• Supports dynamic growth — no fixed initial capacity
• Backed by Go maps (map[T]T)
• Ideal for:
  • Arbitrary keys
  • Sparse connectivity
  • Unpredictable element ranges
  • Rapid prototyping with rich data types

Example

package main

import (
    "fmt"
    "github.com/arunksaha/gdsu/sparse"
)

func main() {
    dsu := sparse.New[string]()

    dsu.Union("apple", "banana")
    dsu.Union("banana", "cherry")

    fmt.Println(dsu.Connected("apple", "cherry")) // true
    fmt.Println(len(dsu.Groups()))                // 1
}

---

3. Compact DSU (Int-Indexed, Slice-Backed)

Located in: compact/compact.go

Key Features

• Very fast, minimal overhead
• Uses contiguous slices for parent and rank
• Requires fixed capacity at initialization (New(size))
• Ideal for:
  • Graph algorithms
  • Tight inner loops
  • Integer node IDs (0..N-1)
  • Performance-critical workloads

Example

package main

import (
    "fmt"
    "github.com/arunksaha/gdsu/compact"
)

func main() {
    dsu := compact.New(10) // supports 0..9

    dsu.Union(1, 2)
    dsu.Union(2, 3)

    fmt.Println(dsu.Connected(1, 3)) // true
    fmt.Println(dsu.Groups())
}

---

4. Benchmark Overview

Benchmark files:

• sparse/sparse_benchmark_test.go
• compact/compact_benchmark_test.go
• comparison/comparison_benchmark_test.go

Benchmarks include:

• Union performance
• Find performance
• Connected queries
• Mixed-operation simulations
• Sparse vs compact comparison
• Memory profiling

Compact is optimized for pure speed; sparse is optimized for flexibility.

---

5. Package Structure

.
├── compact
│   ├── compact_benchmark_test.go
│   ├── compact_example_test.go
│   ├── compact.go
│   └── compact_test.go
├── comparison
│   └── comparison_benchmark_test.go
├── gdsu.go
├── gdsu_test.go
├── go.mod
├── LICENSE
├── Makefile
├── README.md
└── sparse
    ├── sparse_benchmark_test.go
    ├── sparse_example_test.go
    ├── sparse.go
    └── sparse_test.go

---

6. Unique aspects of gdsu

Often (Go) DSU implementations:

• Work only with int
• Require fixed capacity specified at construction
• Do not support generic types
• Do not provide an interface abstraction

gdsu improves on all of these:

Unified interface

Both implementations satisfy the same DSU[T] interface.

Generic sparse DSU

Supports any comparable type, grows dynamically.

High-performance compact DSU

Optimized for numeric workloads.

---

7. Installation

go get github.com/arunksaha/gdsu

---

8. Getting Started

Choose sparse or compact depending on your needs.
Example usage is provided in each subpackage’s *_example_test.go.

The Makefile provides several convenience targets, namely:

• lint to run various formatting and static analysis tools, e.g., fmt, vet, staticcheck, tidy

• test to run unit tests

• coverage to test coverage

• covhtml to view test coverage results in pretty color-coded html

• doc to generate documentation for local viewing

• benchmark to run all the benchmark tests

---

9. Contributions

Issues and PRs are welcome!