Phase III ?

I was reading some thread and stumpled upon this remark

Quote:

---

Originally posted by Stardragon
:scratch: Actually, we're not quite at the genetic algorithm stage, although we are using parts of that approach, and will be reverting to full-scale genetic algorithm in Phase III :idea:

---

I think I missed something. I have the feeling we've only just begun with Phase II and already Phase III is being mentioned.

Did I miss something or can somebody please explain when this is goint to happen, if stats will be reset AGAIN, what purpose scientifically it server etc.

Or am I jumping to conclusions and should I shut up :bonk: don't answer that :blush:

Quote:

---

Originally posted by [DPC]Mobster
I was reading some thread and stumpled upon this remarkI think I missed something. I have the feeling we've only just begun with Phase II and already Phase III is being mentioned.

Did I miss something or can somebody please explain when this is goint to happen, if stats will be reset AGAIN, what purpose scientifically it server etc.

Or am I jumping to conclusions and should I shut up :bonk: don't answer that :blush:

---

The reason for a stats reset was that we switched from a system where each structure took an equal time to process to the current one where the difference in the time it takes to process single structures can be huge. Since the goal was to have people complete a 250 structure set in order to potentially produce accurate protein predictions, the point system needed to be adjusted to give more points for the higher sets in order to prevent people from simply resetting after processing 10,000 structures in order to raise their point total more quickly. Phase One's database only kept track of the number of total structures processed, so converting it to the Phase Two point system would have been extremely difficult to impossible. The change we are talking about would not require a stats reset or even be as radical as the shift from Phase One to Two. Basicly it would be a further revision to the algorithem in order to produce even more accurate folded protein structure predictions.

I know why the stats were reset, but thanks anyway :)

It was/is just my limited knowledge of the thing called Phase III that got me interested...

it's not like howard is just going to pull the phase three alogorithim out of his A$$ one day. I'm sure he's had an idea of where he expects this project to head. I know when I was working on large research projects I had planned a couple of stages in advance so I would be able to keep myself on track.

Just because it was mentioned doesn't mean it's going to happen next month. How long was phase 2 tossed around before it finally went to the general crunching body?

I just hope howard corrects some of the issues that were brought up yesterday in the team free-dc DF forum. Especially the timestamp errorlog issue.

Not to worry, Phase III is still very far away - I was just mentioning it as it was relevant to the genetic algorithm topic that was mentioned. We will be happily crunching and ironing out Phase II for a long solid block of time. :D

Yes, if and when such an update to the algorithm were to take place it would be much smaller than the change from phase I to II (phase II was designed with this in mind). In fact, on the user side, you probably wouldn't notice any difference! The only real difference would be how it goes from one generation to the next, how it generates the trajectory distributions basically. But again, this is all just speculative for now :D