[Frodo2-users] Releasing FRODO version 2.15 (Max-Sum perturbations + non-omniscient controller)

[Thomas L. <tho...@a3...>]

Dear all, 

To celebrate the new year, and following a user support request on how to deploy FRODO in the real world without an omniscient controller that knows the whole problem and extracts each agent's subproblem from it, a new version of FRODO (Version 2.15) has just been released, as described in the following article on the (new) FRODO website: 
https://frodo-ai.tech/index.php/news/6-version-2-15 

The most noticeable changes affect FRODO's distributed submode, which now supports non-omniscient controllers, and include an improvement to the Max-Sum algorithm, which now optionally introduces random perturbations to the problem to break symmetries and improve solution quality. Experimental results on graph coloring and meeting scheduling are reported in the above article on the FRODO website. 

This release also includes a number of other, more minor changes, as listed in the release notes below. 

Please do not hesitate to get in touch with us if you have any questions, or any further improvement suggestions, which we will gladly consider. 

Best regards, and all the best for 2017, 

Thomas Léauté 
https://frodo-ai.tech 

***************
CHANGES IN VERSION 2.15 SINCE VERSION 2.14

Major changes: 
	- MaxSum now optionally introduces perturbations to the problem to improve solution quality. 
	- The distributed submode now support a non-omniscient controller, in which case the configuration files are fed directly to the daemons 
	(support request #10 by Andrea). 
	- The XCSPparser's main function can now split an overall problem into each agent's respective subproblem (support request #10 by Andrea). 
	
Minor changes: 
	- FRODO's website is now hosted at https://frodo-ai.tech. FRODO's SourceForge project site is still used to distribute binaries, 
	and to host the public SVN repository, the ticketing system, and the mailing list. 
	- The CSV files produced by the benchmarking scripts now use semicolons rather than tabs as separators. 
	- It is now possible to control whether the y axis in benchmarking result graphs should use natural or log scale. 
	- Further improvements to the way benchmarking result graphs are plotted. 
	- Various minor improvements to the Controller and Daemon consoles in the advanced mode. 
	- The name of the agent is now saved in its XCSP subproblem as an attribute "self" of the <agents> tag. 
	- StatsReporter modules now truly only report statistics when reportStats = true in the agent configuration file. 
	- Fixed a bug in resetting the stats gatherers between consecutive experiments in advanced mode. 
	- Fixed a bug in the way Max-Sum displays the solution. 
	- Added performance regression tests for meeting scheduling problems.