<?xml version="1.0" encoding="utf-8"?>
<rss version="2.0"
    xmlns:dc="http://purl.org/dc/elements/1.1/"
     xmlns:admin="http://webns.net/mvcb/"
     xmlns:content="http://purl.org/rss/1.0/modules/content/"
     xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
	<channel> 

	<title>Comments on: How to plan the most efficient route?</title>
	<link>http://ask.metafilter.com/34027/How-to-plan-the-most-efficient-route/</link>
	<description>Comments on Ask MetaFilter post How to plan the most efficient route?</description>
	<pubDate>Thu, 09 Mar 2006 07:15:36 -0800</pubDate>
	<lastBuildDate>Thu, 09 Mar 2006 07:15:36 -0800</lastBuildDate>
	<language>en-us</language>
	<docs>http://blogs.law.harvard.edu/tech/rss</docs>
	<ttl>60</ttl>

	<item>
		<title>Question: How to plan the most efficient route?</title>
		<link>http://ask.metafilter.com/34027/How-to-plan-the-most-efficient-route</link>	
		<description>I&apos;m planning to visit every stop on a major metropolitan subway system in a single day (long story), and need to plan the most efficient route. Anyone know of useful technology/mathematical theories that help with route planning?</description>
		<guid isPermaLink="false">post:ask.metafilter.com,2006:site.34027</guid>
		<pubDate>Thu, 09 Mar 2006 06:59:31 -0800</pubDate>
		<dc:creator>jasonsmall</dc:creator>
		
			<category>route</category>
		
			<category>planning</category>
		
			<category>subway</category>
		
			<category>riding</category>
		
	</item> <item>
		<title>By: orthogonality</title>
		<link>http://ask.metafilter.com/34027/How-to-plan-the-most-efficient-route#530504</link>	
		<description>Google &quot;Travelling salesman problem&quot;.</description>
		<guid isPermaLink="false">comment:ask.metafilter.com,2006:site.34027-530504</guid>
		<pubDate>Thu, 09 Mar 2006 07:15:36 -0800</pubDate>
		<dc:creator>orthogonality</dc:creator>
	</item><item>
		<title>By: zpousman</title>
		<link>http://ask.metafilter.com/34027/How-to-plan-the-most-efficient-route#530532</link>	
		<description>This is um, not possible. &lt;a href=&quot;http://www.pcug.org.au/~dakin/tsp.htm&quot;&gt;It&apos;s NP-complete&lt;/a&gt;. That&apos;s actually true only if there are a lot of nodes though. It&apos;s probably doable in Los Angeles for example. Or Atlanta (few nodes / stations, and also simple paths between).</description>
		<guid isPermaLink="false">comment:ask.metafilter.com,2006:site.34027-530532</guid>
		<pubDate>Thu, 09 Mar 2006 07:34:14 -0800</pubDate>
		<dc:creator>zpousman</dc:creator>
	</item><item>
		<title>By: santry</title>
		<link>http://ask.metafilter.com/34027/How-to-plan-the-most-efficient-route#530553</link>	
		<description>This is a pretty hard problem. Computational complexity theorists would call it &lt;a href=&quot;http://en.wikipedia.org/wiki/NP-hard&quot;&gt;NP-hard&lt;/a&gt; (err ... &lt;a href=&quot;http://en.wikipedia.org/wiki/NP-complete&quot;&gt;NP-complete&lt;/a&gt;). &lt;a href=&quot;http://en.wikipedia.org/wiki/Traveling_salesman_problem&quot;&gt;Wikipedia&lt;/a&gt; gives some good background.&lt;br&gt;
&lt;br&gt;
As you&apos;ll find on the Wikipedia page, it&apos;s common to use a weighted graph to conceptualize the problem. Each station would be a vertex, connected by edges that represent subway routes, which edges are assigned costs based on the resources (time, money, total distance traveled, etc.) you&apos;d like to conserve.&lt;br&gt;
&lt;br&gt;
Aside from the difficulty in solving the general case, you have some specific limitations to overcome. Calculating the &quot;cost&quot; for any particular segment of the route seems difficult. Unless you know what the typical travel times are on each route, how long typical delays are at various stations, the prices of fares and transfers, etc., you probably don&apos;t have enough information to come up with a reasonable heuristic, let alone solve the problem.</description>
		<guid isPermaLink="false">comment:ask.metafilter.com,2006:site.34027-530553</guid>
		<pubDate>Thu, 09 Mar 2006 07:47:23 -0800</pubDate>
		<dc:creator>santry</dc:creator>
	</item><item>
		<title>By: Leon</title>
		<link>http://ask.metafilter.com/34027/How-to-plan-the-most-efficient-route#530568</link>	
		<description>NP complete doesn&apos;t mean impossible, it just means it (probably) can&apos;t be solved in polynomial time. Finding a near-optimal route through a mass transit system (the London Underground only has 275 stations) should only take a few minutes on a modern PC.&lt;br&gt;
&lt;br&gt;
What you have here isn&apos;t exactly the classical Travelling Salesman Problem, as you have to visit some nodes more than once. and you probably have asymmetric edge costs. I&apos;d bet that you could solve this in Mathematica, or find some commercial route-planning software. I don&apos;t have any recommendations, though.</description>
		<guid isPermaLink="false">comment:ask.metafilter.com,2006:site.34027-530568</guid>
		<pubDate>Thu, 09 Mar 2006 08:06:14 -0800</pubDate>
		<dc:creator>Leon</dc:creator>
	</item><item>
		<title>By: cadastral</title>
		<link>http://ask.metafilter.com/34027/How-to-plan-the-most-efficient-route#530570</link>	
		<description>More background and info:  &lt;a href=&quot;http://en.wikipedia.org/wiki/Dijkstra&apos;s_algorithm&quot;&gt;Dijkstra&apos;s algorithm&lt;/a&gt;.</description>
		<guid isPermaLink="false">comment:ask.metafilter.com,2006:site.34027-530570</guid>
		<pubDate>Thu, 09 Mar 2006 08:08:23 -0800</pubDate>
		<dc:creator>cadastral</dc:creator>
	</item><item>
		<title>By: dirtynumbangelboy</title>
		<link>http://ask.metafilter.com/34027/How-to-plan-the-most-efficient-route#530590</link>	
		<description>I know SFA about the math involved, but I think I speak for a lot of people when I say... tell us the long story!  You can&apos;t just say you&apos;re going to do something odd and not tell us &lt;i&gt;why&lt;/i&gt; :P</description>
		<guid isPermaLink="false">comment:ask.metafilter.com,2006:site.34027-530590</guid>
		<pubDate>Thu, 09 Mar 2006 08:28:47 -0800</pubDate>
		<dc:creator>dirtynumbangelboy</dc:creator>
	</item><item>
		<title>By: Leon</title>
		<link>http://ask.metafilter.com/34027/How-to-plan-the-most-efficient-route#530661</link>	
		<description>[Assuming a subway system] It occurs to me that you can prune the problem down to only the stations that allow connections, by making the cost between each station the sum of the costs of the &quot;inbetween&quot; stations. That makes the problem a lot smaller (it might cut the number of nodes in half?)</description>
		<guid isPermaLink="false">comment:ask.metafilter.com,2006:site.34027-530661</guid>
		<pubDate>Thu, 09 Mar 2006 09:24:10 -0800</pubDate>
		<dc:creator>Leon</dc:creator>
	</item><item>
		<title>By: joegester</title>
		<link>http://ask.metafilter.com/34027/How-to-plan-the-most-efficient-route#530691</link>	
		<description>I think to do any optimal planning, you&apos;d need to know all about the timing of the trains and stuff.  I don&apos;t think it&apos;s practical.  Besides in these sorts of situations the naive solution is usually not too much worse than the optimal one.&lt;br&gt;
&lt;br&gt;
Posting flyers?</description>
		<guid isPermaLink="false">comment:ask.metafilter.com,2006:site.34027-530691</guid>
		<pubDate>Thu, 09 Mar 2006 09:48:51 -0800</pubDate>
		<dc:creator>joegester</dc:creator>
	</item><item>
		<title>By: Johnny Assay</title>
		<link>http://ask.metafilter.com/34027/How-to-plan-the-most-efficient-route#530729</link>	
		<description>Dijkstra&apos;s algorithm isn&apos;t quite applicable here &amp;mdash; it would only be useful if you were going from one point to another and needed the quickest route.  The classical &quot;traveling salesman&quot; problem isn&apos;t quite applicable here, either, since it assumes that you visit each vertex only once (in fancy jargon, it tries to find the most efficient Hamiltonian circuit.)&lt;br&gt;
&lt;br&gt;
Can you freely choose your starting and ending locations, or is there a constraint on them?</description>
		<guid isPermaLink="false">comment:ask.metafilter.com,2006:site.34027-530729</guid>
		<pubDate>Thu, 09 Mar 2006 10:21:24 -0800</pubDate>
		<dc:creator>Johnny Assay</dc:creator>
	</item><item>
		<title>By: drstein</title>
		<link>http://ask.metafilter.com/34027/How-to-plan-the-most-efficient-route#530738</link>	
		<description>Uh, is this part of a Geocaching puzzle? &lt;br&gt;
&lt;br&gt;
(We have some of those out here.. they involve blowing 2 days on local transit systems)</description>
		<guid isPermaLink="false">comment:ask.metafilter.com,2006:site.34027-530738</guid>
		<pubDate>Thu, 09 Mar 2006 10:24:19 -0800</pubDate>
		<dc:creator>drstein</dc:creator>
	</item><item>
		<title>By: jellicle</title>
		<link>http://ask.metafilter.com/34027/How-to-plan-the-most-efficient-route#530809</link>	
		<description>I don&apos;t think you need theory here - you need experience.  A friend who rides the subway and can tell you how to go and where to switch trains.  Also - can you leave the subway system, travel to another stop, and reenter?  I think little bits of knowledge that a local subway rider will have will decrease the total time taken by more than any theoretical analysis will.&lt;br&gt;
&lt;br&gt;
My advice, without knowing which system you intend to do: most subway systems look like a spider.  You will spend a lot of time riding out to the end of each arm and back into the middle.  Look for places where you can ride out on arm A, take a cab, and ride back in on arm B.</description>
		<guid isPermaLink="false">comment:ask.metafilter.com,2006:site.34027-530809</guid>
		<pubDate>Thu, 09 Mar 2006 11:19:11 -0800</pubDate>
		<dc:creator>jellicle</dc:creator>
	</item><item>
		<title>By: jasonsmall</title>
		<link>http://ask.metafilter.com/34027/How-to-plan-the-most-efficient-route#530849</link>	
		<description>What I&apos;m trying to do may be pratically impossible, but I lost a bet and now have to attempt to ride the entire 656 miles of the New York subway in a day, passing through every one of the ~468 stations at least once. I live in Brooklyn and I know the subway pretty well (it&apos;s not like I ride the JMZ all that often), but the sheer size of the system makes planning tough.&lt;br&gt;
&lt;br&gt;
Obviously, I&apos;ll have to pass through a number of stations more than once, and there is clearly no perfect solution, but I&apos;d like to make up a route more efficient that what I&apos;d come up with just using a large map and markers on my own. Given the complexity of the system, coming up with exact travel times for each leg is an impossibility, but I want to work up a rough estimate -- if only to see if the idea is feasible&lt;br&gt;
&lt;br&gt;
jellicle, the cab idea is great, though I have to check if this violates the spirit of the bet.</description>
		<guid isPermaLink="false">comment:ask.metafilter.com,2006:site.34027-530849</guid>
		<pubDate>Thu, 09 Mar 2006 12:16:15 -0800</pubDate>
		<dc:creator>jasonsmall</dc:creator>
	</item><item>
		<title>By: mendel</title>
		<link>http://ask.metafilter.com/34027/How-to-plan-the-most-efficient-route#530921</link>	
		<description>&lt;i&gt;656 miles of the New York subway in a day&lt;/i&gt;&lt;br&gt;
&lt;br&gt;
Ok, I&apos;ll bite. The average speed of New York subways is &lt;a href=&quot;http://www.bettertransit.com/forum/topic.asp?TOPIC_ID=53&quot;&gt;apparently&lt;/a&gt; under 20 mph; how do you cover 656 miles of anything, regardless of route, at 20 mph in a day?</description>
		<guid isPermaLink="false">comment:ask.metafilter.com,2006:site.34027-530921</guid>
		<pubDate>Thu, 09 Mar 2006 13:05:31 -0800</pubDate>
		<dc:creator>mendel</dc:creator>
	</item><item>
		<title>By: generichuman</title>
		<link>http://ask.metafilter.com/34027/How-to-plan-the-most-efficient-route#531088</link>	
		<description>If I can chime in, don&apos;t disregard using surface transit as connections unless it violates the rules of whatever you&apos;re trying to do by riding the whole system.   &lt;br&gt;
&lt;br&gt;
Some friends attempted a similar challenge in Toronto, (much smaller system, though.)  They used surface routs like buses/streetcars/LRT&apos;s to connect the extreme points of the subways.  Although the arrangement of the NYC subway might make that harder, it&apos;s something to factor into any calculations you make.</description>
		<guid isPermaLink="false">comment:ask.metafilter.com,2006:site.34027-531088</guid>
		<pubDate>Thu, 09 Mar 2006 14:57:28 -0800</pubDate>
		<dc:creator>generichuman</dc:creator>
	</item><item>
		<title>By: jellicle</title>
		<link>http://ask.metafilter.com/34027/How-to-plan-the-most-efficient-route#531298</link>	
		<description>NYC subway in a day?  Hmm.  I think the total mileage is &lt;a href=&quot;http://www.nycsubway.org/faq/factsfigures.html&quot;&gt;230&lt;/a&gt;, because if there are five sets of track going through a station, you only have to go through it once.  (I&apos;m assuming that one pass through a station is fully-sufficient for your wager.)  It should be theoretically possible, though you&apos;re talking a really long day.  I&apos;d be surprised if you could do it in much under 24 hours.&lt;br&gt;
&lt;br&gt;
Getting the edges is going to be lots harder than getting the middle.  The Bronx will take, say, 4 hours, maybe less if you can cab or bus between arms.  Manhattan will take about 4-5 hours.  Do Queens, end up at Jamaica Center, down into Brooklyn and start working your way through the arms of the A, S, L, 3....  Brooklyn is easily going to take ten hours or more.  You might be able to cut that down by hopping between lines at the end of the various Coney Island arms.  Add 6 hours or more for friction - waiting in stations, back-tracking to get odd stations.  Wear comfortable shoes.&lt;br&gt;
&lt;br&gt;
If you do this, be sure to bring a camera and extra memory card and lots of extra batteries, and put up a webpage with Google ads.  Battery-powered clock, take a picture of every station name through the train window with the clock in the shot?  Might be hard.</description>
		<guid isPermaLink="false">comment:ask.metafilter.com,2006:site.34027-531298</guid>
		<pubDate>Thu, 09 Mar 2006 18:25:59 -0800</pubDate>
		<dc:creator>jellicle</dc:creator>
	</item><item>
		<title>By: boudicca</title>
		<link>http://ask.metafilter.com/34027/How-to-plan-the-most-efficient-route#531600</link>	
		<description>You might want to look at Geoff Marshall&apos;s &lt;a href=&quot;http://www.geofftech.co.uk/tubechallenge/&quot;&gt;Tube Challenge &lt;/a&gt;site. He&apos;s made 8 attempts at visiting all the stations on the London Underground which are well documented on his site and you might find some ideas which transfer to New York. Geoff is the current World Record Holder for visiting all the London Underground stations in the fastest time.</description>
		<guid isPermaLink="false">comment:ask.metafilter.com,2006:site.34027-531600</guid>
		<pubDate>Fri, 10 Mar 2006 02:09:12 -0800</pubDate>
		<dc:creator>boudicca</dc:creator>
	</item>
	</channel>
</rss>
