barguzin / spatial_heuristics

Toy examples for heuristics used in locational science

Spatial Heuristics

prepared by Evgeny Noi

Week 5 of GEOG 294 course at UC Santa Barbara

Toy examples for heuristics used in locational science.

---

Installation and usage

Installation:

1. Clone repo via git or Github Desktop git clone https://github.com/barguzin/spatial_heuristics.git
2. Navigate to the cloned repo directory cd spatial_heuristics
3. Create a Python environment (conda or pip): conda create -n env_name
4. Activate the newly created environment: conda activate env_name
5. Install requirements using pip: pip install -r requirements.txt

Usage:

1. Run check_funs.ipynb notebook to see usage examples

---

The functions are located in spam/core.py:

make_points(n) - Generate set of points (sites) on a 2d plane using a normal distribution. Where $n$ is the number of points. Results are written into demand.csv in a root folder.

makde_grid(q) - Generate regular square grid over a 2d plane. Where $q$ is the number of cells along an axis. Yields q*q potential sites. Results are written into facility.csv in a root folder.

calculate_dist - Calculate distance via scipy.distance.cdist and store results into distance_matrix.csv in a root directory.

get_covered(r) - For each facility find the demand points that are covered given a radius ($r$). Saves pairwise_dist.csv (pairwise distance between facilities and demand points), total_pop.csv (total covered population for each facility), covered.pickle (id of covered demand for each facility_id)

naive_greedy(n_sited, sort_par='cnt') - Implements naive Greedy Addition algorithm for MCLP with $n_cited$ denoting the number of facilities to be sited and $sort_par$ defining what variables to use for sorting (either number of facilities covered by each facility or total covered population). Saves sited_id (id of sited facilities) and covered_id (id of covered demand). The output is also provided via logger available at ./out.txt.

plot_solution - plots solution

---

Implemented:

1. Greedy Addition for maximal covering

To be implemented:

1. Teitz & Bart

---

References:

1. Teitz, M. B., & Bart, P. (1968). Heuristic methods for estimating the generalized vertex median of a weighted graph. Operations research, 16(5), 955-961.