[Lemon-user] Fwd: About Lemon Library

[Péter Kovács]

Dear Hakan Akyuz,

I don't know the details of the matching algorithms, so let me forward 
your email to the LEMON user mailing list. I hope that someone can 
answer your questions.

Best regards,
Péter Kovács



-------- Forwarded Message --------
Subject: 	About Lemon Library
Date: 	Thu, 14 Jun 2018 14:04:56 +0300
From: 	hakan akyuz <mhakyuz at gmail.com>
To: 	kpeter at inf.elte.hu
CC: 	HAKAN AKYUZ <mhakyuz at gsu.edu.tr>



Dear Peter Kovacs,

I have been working with LEMON Library to solve a variant of Maximum 
Weight Perfect Matching Problem. To my experience, the performance of 
the LEMON is excellent and I'm thankful for the efforts you and your 
group put into this project.

I have two questions about LEMON. All is related to the header "matching.h":

1) Our problem requires that we resort to the maximum weight perfect 
matching algorithm repetitively. I'm trying to initialize the maximum 
weight perfect matching algorithm with a given matching. This is similar 
to the existing jump start initialization "fractionalInit()". However, 
our initial matching does not have to be a perfect one or even a 
matching with only one edge is welcome. Our motivation is to initialize 
the algorithm with a matching to increase the speed of the algorithm for 
solving the overall problem.

How can we achieve such an initialization?

2) In "createStructures()" function the following line # 2278

" _blossom_num = _node_num * 3 / 2; "

sets the number of blossoms to 3*n/2 where n is the number of nodes in 
the graph. Is there a specific reason or a theoretical property for this 
setting? I'm very curious about this.

We would be thankful if you could help.

Sincerely,

Hakan Akyuz

Assistant Professor
Department of Industrial Engineering
Galatasaray University
Istanbul

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lemon.cs.elte.hu/pipermail/lemon-user/attachments/20180614/f37c0c3a/attachment.html>