10/09/2002
- As we have now reached 10 billion structures, the next protein update will occur on the following Tuesday, Oct. 15, at around noon EDT.
Jetzt ist es offiziell !!!!
Proteinwechsel am 15. Oktober!!!!
-
Pascal
-
Michael H.W. Weber
- Vereinsvorstand
- Beiträge: 23165
- Registriert: 07.01.2002 01:00
- Wohnort: Marpurk
Die Asiaten haben jetzt noch ca. 74 Millionen Strukturen Vorsprung und wir halten Weltrangplatz Nr. 3 in der Wochenproduktion. Wir liefern momentan ca. 34 Millionen Struturen pro Woche mehr ab als das Team aus Singapur und könnten bei unserem jetzigen Durchsatz zumindest den Abstand knapp halbieren.
Michael.
Michael.
Fördern, kooperieren und konstruieren statt fordern, konkurrieren und konsumieren.
-
Matthias
-
cannibal
- PDA-Benutzer
- Beiträge: 48
- Registriert: 30.06.2002 08:49
-
Rechtsconsulter
- Rechenkraft.net-Sponsor
- Beiträge: 178
- Registriert: 14.06.2002 22:17
- Wohnort: Dortmund
Re: Proteinwechsel am 15. Oktober!!!!
Arnim Weber hat geschrieben:10/09/2002
- As we have now reached 10 billion structures, the next protein update will occur on the following Tuesday, Oct. 15, at around noon EDT.
Die ändern aber schnell ihre Ankündigungen:"10/09/2002
- As we have now reached 10 billion structures, the next protein update will occur on the following Wednesday, Oct. 16, at around noon EDT (it is Thansgiving long weekend here in Canada)."
Mal schauen, obs dann was mit dem Wechsel wird.
Wenn alle Stricke reißen - dann hänge ich mich auf!
-
Flumsi
- PDA-Benutzer
- Beiträge: 33
- Registriert: 20.09.2002 16:56
Hallo,
kann mir mal jemand erklären ob wir nun bis am 16. Oktober vergebens rechnen wie bei RC5-64? Die 10 Bio Structuren sind ja erreicht und das nächste Protein kommt erst am 16.
Zudem möcht ich noch wissen was mit den Stats passiert, im DF-Forum habe ich was gelesen doch ich kann nicht so gut english
Gruss Flumsi
kann mir mal jemand erklären ob wir nun bis am 16. Oktober vergebens rechnen wie bei RC5-64? Die 10 Bio Structuren sind ja erreicht und das nächste Protein kommt erst am 16.
Zudem möcht ich noch wissen was mit den Stats passiert, im DF-Forum habe ich was gelesen doch ich kann nicht so gut english
Gruss Flumsi
-
bluumi
- Number-Cruncher
- Beiträge: 865
- Registriert: 28.06.2001 01:00
- Wohnort: Neuenhof
Nix vergebens.
DF ist ein RANDOM-Projekt ... d.h. auch oberhalb ihrer selbstgewählten limite kann immernoch ein viel besseres zufällig auftauchen...
Stats wird einfach BEST transfered auf 99.999 zurück gesetzt..
Gruss
DF ist ein RANDOM-Projekt ... d.h. auch oberhalb ihrer selbstgewählten limite kann immernoch ein viel besseres zufällig auftauchen...
Stats wird einfach BEST transfered auf 99.999 zurück gesetzt..
Gruss
SwissTeam.NET , the better .NET... dist. computing-Team of Switzerland
-
Pascal
Hier noch ein paar aktuelle Infos vom Brian the Fist (Howard Feldman alias der Chief von DF):
hi,
The update will actually be on Wed. the 16th (as now correctly stated on the web site). I did not say anything before because I did not decide yet We will likely at this time also complete the maintenance that was started a week ago as that did not get finished, so it may be down for several hours. We will also add the UID to uniquely identify users at this time, though it may not immediately show up on the stats pages until several days later.
At to the policy of deleting info on Dec. 2003 etc., that will obviously be revised if the project is still running then, as it may indeed change direction before that time. It is meant more to assure you that when the project IS complete, we will destroy all personal info and not sell it to others, etc.
As it stands, we will proceed with the current phase, sampling 10 billion structures, but at the same time as I have mentioned elsewhere we will be trying to change the sampling to work more like a genetic algorithm, and when this is tested and complete we will instead of doing 10 billion random samples switch to the genetic algorithm. This will have little impact on the end user, running the client, and it will appear to you to still be doing pretty much the same thing. The biggest difference is every once and a while it when it uploads data to the server it will also download a small (digitally signed, of course) packet of data which it will then make use of to bias its sampling. We are very excited about trying out this new method, but expect it to be at least several months until it is all worked out and functional. We expect it will result in significantly lower RMSDs in much shorter time frames, and in the worst case it should not do any worse than the current approach. But the only way to find out for sure is to try it.
To answer one other question, any structures produced between now and when we change proteins (even though we've filled the thermometer) will still be collected and of interest so please keep folding, it is nto going to waste!
__________________
Howard Feldman
hi,
The update will actually be on Wed. the 16th (as now correctly stated on the web site). I did not say anything before because I did not decide yet We will likely at this time also complete the maintenance that was started a week ago as that did not get finished, so it may be down for several hours. We will also add the UID to uniquely identify users at this time, though it may not immediately show up on the stats pages until several days later.
At to the policy of deleting info on Dec. 2003 etc., that will obviously be revised if the project is still running then, as it may indeed change direction before that time. It is meant more to assure you that when the project IS complete, we will destroy all personal info and not sell it to others, etc.
As it stands, we will proceed with the current phase, sampling 10 billion structures, but at the same time as I have mentioned elsewhere we will be trying to change the sampling to work more like a genetic algorithm, and when this is tested and complete we will instead of doing 10 billion random samples switch to the genetic algorithm. This will have little impact on the end user, running the client, and it will appear to you to still be doing pretty much the same thing. The biggest difference is every once and a while it when it uploads data to the server it will also download a small (digitally signed, of course) packet of data which it will then make use of to bias its sampling. We are very excited about trying out this new method, but expect it to be at least several months until it is all worked out and functional. We expect it will result in significantly lower RMSDs in much shorter time frames, and in the worst case it should not do any worse than the current approach. But the only way to find out for sure is to try it.
To answer one other question, any structures produced between now and when we change proteins (even though we've filled the thermometer) will still be collected and of interest so please keep folding, it is nto going to waste!
__________________
Howard Feldman