------------------------------------------------------------------------
Accumulated Comments about Project 1.
  Important queries/clarification on the project 1
  will be made here.   --hon-wai
------------------------------------------------------------------------

10-Sept-2002
--------------
  0. Thanks to all who mailed me with comments...

  1. Please see announcement page on *CPU TIME LIMIT* error message
     when running LEDAdijkstra2 for large graphs.

  2. Errors encountered with sparse large graphs...
     See note below.




----------------------------------------------------------------------------
From: Mohammad Awrangjeb [mailto:mohamma1@comp.nus.edu.sg]

Dear sir,
Changed (on 10 sept) LEDAdijkstra2.cc give Error for the following inputs:

n=2000, m= 5*n
n=4000, m= 5*n
n=8000, m= 5*n
n=16000, m= 5*n, 10*n
n=32000, m= 5*n

What should we do?

Thanks.

Sincerely Yours
Mohammad Awrangjeb

----------------------------------------------------------------------------
My Reply:

Please refer to the web-site.
The "apparent error" occurs sometimes when the graph is too sparse and so
the graph is not connected. This means that some of the nodes
are NOT reachable....

It is OK since we are interested only in measuring the running times.

Regards, --hon-wai
----------------------------------------------------------------------------



----------------------------------------------------------------------------
From: Zhu Yi [mailto:zhuyi@comp.nus.edu.sg]
Sent: 11 September 2002 15:37
		  
Dear Prof. Leong,
I am doing the LEDA project 1. When I run the program dijkstra.cc and
dijkstra2.cc, many error information occurs in the output. It is because
that the default LEDA DIJKSTRA algorithm will set the distance to be 0 if
two nodes are not connected, while the algorithm implemented in the
sample programs will set it to be MAXINT. So there is many errors in
sparse graphs but no error in very dense graphs.

Could I change the implementation of the sample programs to do the
analysis? Thanks.
			 
yours,
Zhu Yi
			     
----------------------------------------------------------------------------
My Reply:
  Thanks for your mail and your sharp observation on the
  cause of the error. I am aware of the error, but chose
  NOT to change it as its effect is minor. Then, everyone
  have to download again..... We are interested only in 
  the running time and this does not affect it.
	    
  Since you have already gone into the code and find out 
  about it, why don't you change the code and mail the
  modified code to me. I will use it for future versions.
	  
Regards, --hon-wai
----------------------------------------------------------------------------
		  

Started: LeongHW; 10-September-2002
--------------------------------------