Difference between revisions of "TADM2E 1.26"

From Algorithm Wiki
Jump to: navigation, search
(Undo revision 851 by Mgits (talk))
(Blanked the page)
Line 1: Line 1:
Tests show that closest pairs heuristic generally performs better than nearest neighbour heuristic.<br/>
 
Here's python implementation for <b>nearest neighbour</b> heuristic:
 
<pre>
 
import random
 
import matplotlib.pyplot as plot
 
import matplotlib.cm as cm
 
import numpy as np
 
import math
 
  
 
def draw_arrow(axis, p1, p2, linecolor, style='solid', text="", radius=0):
 
    """draw an arrow connecting point 1 to point 2"""
 
    axis.annotate(text,
 
              xy=p2,
 
              xycoords='data',
 
              xytext=p1,
 
              arrowprops=dict(arrowstyle="-", linestyle=style, linewidth=0.8, color=linecolor,
 
                                                connectionstyle="arc3,rad=" + str(radius)),)
 
 
 
#nearest neighbour heuristic
 
def nearest_neighbour(datapoints):
 
    x, y = 0, 1
 
    #pick random starting point and add it to path
 
    i = random.randint(0, len(datapoints) - 1)
 
    path = [datapoints[i]]
 
    del datapoints[i]
 
    i = 0
 
    # while there are points find the closest one to datapoints[i], add it to path
 
    while(len(datapoints) != 0):
 
        minlen = 1e124
 
        minind = -1
 
        for k in range(len(datapoints)):
 
            dist = math.hypot(datapoints[k][x] - path[i][x], datapoints[k][y] - path[i][y])
 
            if minlen > dist:
 
                minlen = dist
 
                minind = k
 
        path.append(datapoints[minind])
 
        del datapoints[minind]
 
        i += 1
 
    return path
 
 
# MAIN SCRIPT
 
random.seed()
 
figure = plot.figure()
 
axis = figure.add_subplot(111)
 
 
n = 6
 
points = [(random.uniform(0.01, 0.99), random.uniform(0.01, 0.99)) for i in range(n)]
 
# points for line
 
points = [(0.3, 0.2), (0.25, 0.2), (0.5, 0.2), (0.7, 0.2), (0.6, 0.2), (0.8, 0.2)]
 
 
# find shortest path
 
path_points = nearest_neighbour(points)
 
 
# draw path
 
colors = cm.rainbow(np.linspace(0, 1, len(path_points)))
 
plot.scatter([i[0] for i in path_points], [i[1] for i in path_points], color=colors)
 
# draw shortest path from point[0] to point[n-1]:
 
draw_arrow(axis, path_points[0], path_points[1], colors[0], style='solid', radius=0.3)
 
for i in range(1, len(path_points)-1):
 
    draw_arrow(axis, path_points[i], path_points[i + 1], colors[i], radius=0.3)
 
draw_arrow(axis, path_points[n - 1], path_points[0], colors[n-1], style='dashed', radius=0.3)
 
 
plot.show()
 
</pre>
 
 
<br/>Python implementation of <b>closest pair</b> heuristic<br/>
 
<pre>
 
import random
 
import matplotlib.pyplot as plot
 
import matplotlib.cm as cm
 
import numpy as np
 
import math
 
 
 
def draw_arrow(axis, p1, p2, linecolor, style='solid', text = "", radius = 0):
 
    """draw an arrow connecting point 1 to point 2"""
 
    axis.annotate(text,
 
              xy=p2,
 
              xycoords='data',
 
              xytext=p1,
 
              arrowprops=dict(arrowstyle="-", linestyle=style, linewidth=0.8, color=linecolor,
 
                                                connectionstyle="arc3,rad=" + str(radius)),)
 
 
 
#closest pair heuristic
 
def closest_pair(points):
 
    distance = lambda c1p, c2p:  math.hypot(c1p[0] - c2p[0], c1p[1] - c2p[1])
 
    chains = [[points[i]] for i in range(len(points))]
 
    edges = []
 
    for i in range(len(points)-1):
 
        dmin = float("inf")  # infinitely big distance
 
        # test each chain against each other chain
 
        for chain1 in chains:
 
            for chain2 in [item for item in chains if item is not chain1]:
 
                # test each chain1 endpoint against each of chain2 endpoints
 
                for c1ind in [0, len(chain1) - 1]:
 
                    for c2ind in [0, len(chain2) - 1]:
 
                        dist = distance(chain1[c1ind], chain2[c2ind])
 
                        if dist < dmin:
 
                            dmin = dist
 
                            # remember endpoints as closest pair
 
                            chain2link1, chain2link2 = chain1, chain2
 
                            point1, point2 = chain1[c1ind], chain2[c2ind]
 
        # connect two closest points
 
        edges.append((point1, point2))
 
 
        chains.remove(chain2link1)
 
        chains.remove(chain2link2)
 
        if len(chain2link1) > 1:
 
            chain2link1.remove(point1)
 
        if len(chain2link2) > 1:
 
            chain2link2.remove(point2)
 
        linkedchain = chain2link1
 
        linkedchain.extend(chain2link2)
 
        chains.append(linkedchain)
 
    # connect first endpoint to last one
 
    edges.append((chains[0][0], chains[0][len(chains[0])-1]))
 
    return chains[0], edges
 
 
 
# MAIN SCRIPT
 
random.seed()
 
figure = plot.figure()
 
axis = figure.add_subplot(111)
 
 
n = 6
 
points = [(random.uniform(0.01, 0.99), random.uniform(0.01, 0.99)) for i in range(n)]
 
# six points for a rectangle
 
points = [(0.3, 0.2), (0.3, 0.4), (0.501, 0.4), (0.501, 0.2), (0.702, 0.4), (0.702, 0.2)]
 
 
#find shortest path
 
path_points, edges = closest_pair(points)
 
 
#draw path
 
colors = cm.rainbow(np.linspace(0, 1, len(path_points)))
 
plot.scatter([i[0] for i in points], [i[1] for i in points], color=colors)
 
# draw shortest path from point[0] to point[n-1]:
 
for i in range(len(edges)):
 
    draw_arrow(axis, edges[i][0], edges[i][1], 'black', radius=0.)
 
 
plot.show()</pre>
 
<br/>The <b>minimum angle with randomised centroid heuristic</b> solves both cases closest pair and nearest neighbour can't handle.<br/>
 
1. Calculate the centroid (geometric mean of x and y coordinates) of all points given. Add a little offset to centroid, that allows to solve cases when points form a line.<br/>
 
2. Find the point that is furthermost from centroid. Let's call it point1<br/>
 
3. Find point2 that comprises the smallest angle point1-centroid-point2<br/>
 
4. Connect point1 and point2 with an edge.<br/>
 
5. Repeat from step 3 with point2.<br/>
 
<pre>
 
import random
 
import matplotlib.pyplot as plot
 
import matplotlib.cm as cm
 
import numpy as np
 
import math
 
 
 
def draw_arrow(axis, p1, p2, linecolor, style='solid', text = "", radius = 0):
 
    """draw an arrow connecting point 1 to point 2"""
 
    axis.annotate(text,
 
              xy=p2,
 
              xycoords='data',
 
              xytext=p1,
 
              arrowprops=dict(arrowstyle="-", linestyle=style, linewidth=0.8, color=linecolor,
 
                                                connectionstyle="arc3,rad=" + str(radius)),)
 
 
 
def angle_degrees(p1, center, p2):
 
    """angle in radians"""
 
    distance = lambda c1p, c2p:  math.hypot(c1p[0] - c2p[0], c1p[1] - c2p[1])
 
    a = distance(center, p1)
 
    b = distance(center, p2)
 
    c = distance(p2, p1)
 
    cosine = (a**2 + b**2 - c**2) / (2*a*b)
 
    return math.acos(min(max(cosine, -1), 1))
 
 
 
def centroid(points):
 
    center = (sum([point[0] for point in points])/len(points) + random.uniform(0.010, 0.015),
 
              sum([point[1] for point in points])/len(points) + random.uniform(0.010, 0.015))
 
    edges = []
 
    # remember how many connections a point has. start with 0 connections
 
    uses = dict([(point, 0) for point in points])
 
    # start with the furthermost point
 
    point1 = points[0]
 
    longest = math.hypot(center[0] - point1[0], center[1] - point1[1])
 
    for pt in points:
 
        dist = math.hypot(center[0] - pt[0], center[1] - pt[1])
 
        if dist > longest:
 
            longest = dist
 
            point1 = pt
 
    # for every point find the other one that comprises the SMALLEST angle poin1-center-point2
 
    while True:
 
        if uses[point1] < 2:
 
            min_angle = 1e34
 
            point_to_connect = None
 
            # point must not be used more than twice!
 
            for point2 in [item for item in points if item is not point1]:
 
                angle = angle_degrees(point1, center, point2)
 
                if not (point1, point2) in edges and not (point2, point1) in edges and uses[point2] < 1 and\
 
                        angle < min_angle:
 
                    min_angle = angle
 
                    point_to_connect = point2
 
            if point_to_connect is not None:
 
                edges.append((point1, point_to_connect))
 
                uses[point1] += 1
 
                uses[point_to_connect] += 1
 
                point1 = point_to_connect
 
            else:
 
                break
 
        else:
 
            break
 
    # connect the last two points
 
    last_points = [k for k, v in uses.iteritems() if v == 1]
 
    assert(len(last_points) == 2)
 
    edges.append((last_points[0], last_points[1]))
 
 
    return edges, center
 
 
 
# MAIN SCRIPT
 
random.seed()
 
figure = plot.figure()
 
axis = figure.add_subplot(111)
 
 
n = random.randint(6, 10)
 
points = [(random.uniform(0.01, 0.99), random.uniform(0.01, 0.99)) for i in range(n)]
 
# points for line
 
#points = [(0.3, 0.2), (0.4, 0.2), (0.5, 0.2), (0.7, 0.2), (0.6, 0.2), (0.86, 0.2)]
 
# six points for a rectangle
 
#points = [(0.3, 0.2), (0.3, 0.4), (0.501, 0.4), (0.501, 0.2), (0.702, 0.4), (0.702, 0.2)]
 
 
edges, center = centroid(points)
 
 
# draw points
 
colors = cm.rainbow(np.linspace(0, 1, len(points)))
 
plot.scatter([i[0] for i in points], [i[1] for i in points], color=colors)
 
 
# draw lines from centroid to points
 
plot.scatter(center[0], center[1], color='green')
 
for point in points:
 
    draw_arrow(axis, center, point, 'red', radius=0)
 
 
# draw edges of shortest path
 
for i in range(len(edges)):
 
    draw_arrow(axis, edges[i][0], edges[i][1], 'black', radius=0.)
 
 
plot.show()</pre>
 

Revision as of 02:10, 2 August 2020