Yesterday I found a bug in one of the programs for the chemlambda visualiser, namely an unexpected functioning of the priority choice part when applied to a particular mol file (graph).
Today I took it methodically and found two bugs which create that behaviour. One was that the priority choice was not covering all the possibilities in a correct way, the other was a pure programming bug.
So I corrected everything and checked that on the examples from the gallery it works as expected, and also I checked it on many other examples which I have and it works well.
Of course, only to check a program on examples means nothing without a proof that the algorithm works well. I have done that, now the priority choice is indeed well implemented.
What is funny is that by trying to correct the priority choice part I have found the other stupid bug.
OK, so now all the new tar files are marked with 06_10_2014.
And anyway, as everything is open here, you may compare the versions and arrive to your own conclusions.
Mind that in a sense there is still a bug in the main_viral from the “play” archive: it may happen that at some point the json file which is seen with look.html may be empty. This happens because in the “play” algorithm the loops are erased, and it is possible that a molecule reduces to a collection of loops, hence an empty json file eventually.
No problem with that, by using the “add_new” tar you have an algorithm which does not remove the loops (and also has the FOE node and its new moves). This one works perfectly now.
On a different subject: the artificialagora.wordpress.com will launch soon. I don’t know yet if chorasimilarity and artificialagora will coexist, or if artificialagora will be the luxury version of chorasimilarity (hope not).
But soon enough we’ll see this.
Chorasimilarity site is still read in a strange way for a blog, because old posts receive comparable hits with new ones. Not all posts are good, for example posts as this one, where I indulge into telling about updates and preparations.