amyoxen / k-shortest-paths

Automatically exported from code.google.com/p/k-shortest-paths
0 stars 0 forks source link

running time is longer longer than expected time claimed by the reference paper.... #13

Open GoogleCodeExporter opened 8 years ago

GoogleCodeExporter commented 8 years ago
What steps will reproduce the problem?
1. run the program normally using provided test file test_500
2.
3.

What is the expected output? What do you see instead?
the k (3000) shortest paths from node 1 to node 729

What version of the product are you using? On what operating system?
last version (1.0.3) with the updated (corrected) header file 
(QYDirectedPath.h)...boost library Ver 1.35.0 on windows XP SP2

Please provide any additional information below.
I have ran the program normally (environment and parameters are given 
above) for the test file provided (test_500) on e-machines desktop T6414 
AMD Athlon 64 Processor (3200++)...memory size = 512 MB...I halted the 
program after more than 2.5 hours of running time....but in the original 
paper....the reference paper... the expected running time would be less 
than this..... I would kindly ask for the reason of that...thnx for ur time

Original issue reported on code.google.com by k.aln...@gmail.com on 31 Dec 2009 at 12:07

GoogleCodeExporter commented 8 years ago
Did you try the old version without the updated header file? 
On the other hand, did you some other examples? Like using smaller size of data 
set,
or trying the queries starting from the different nodes? 

Original comment by yan.qi....@gmail.com on 6 Jan 2010 at 12:17

GoogleCodeExporter commented 8 years ago
I have tried these alternatives...

1. I ran the program (Ver. 1.0.3) using different pair (70 to 420) ... same 
problem.
2. I ran the program (Ver. 1.0.3) with the older version of header file 
(QYDirectedPath.h)  from node 0 to 420...same problem.
3. I ran the older version (1.0.2) from node 70 to 420...same problem.....

for all attempts above the value of K used was 300...I stoped all programs 
after 1.5 
hours of running time...

I wonder if Authors of the paper were very optimistic!!!

Original comment by k.aln...@gmail.com on 14 Jan 2010 at 2:19

GoogleCodeExporter commented 8 years ago
I have recently tried the same experiment (v 2.0) 
*using provided test file test_500
*from node 1 to node 729

yenAlg.has_next() now returns false and I get no shortest paths, this runs very 
quickly.

The same is true from node 1 to node 7, where it is easy to trace paths by 
looking at the file.

I would like this to work for 10k nodes, I get the same results, when I try it 
with my own adjacency file.  I suspect it has something to do with scale since 
test_50 works fine.

Any ideas?

Original comment by clairedo...@googlemail.com on 30 Oct 2011 at 1:54