1n0u__topology_sample_00151 ??

"Curiouser and curiouser!Ó Cried Alice (She was so much surprised, that for the moment she quite forgot how to speak good English).

Happens quite a lot to me. Seriously, is this just a new naming convention or are we doing something completely different now ? The new units also seem to be pretty fast compared to the last (*filter*) ones. Didn't notice any dramatic change in the list of command line parameters, though (at least from the laypersons point of view).

Yes--this is something a bit different. In analyzing the results over the last several weeks, I've found that after clustering the lowest energy 1% of the structures, the correct topology is almost always the largest or one of the largest clusters. the "topology resamplling" runs come with a list of these low energy topologies, and at the beginning of each run a variant of one of the topologies is randomly selected and used to guide the trajectory. the idea, which has been discussed in previous message board discussions, is to focus sampling on these promising regions,

---

---

Regards,
Bob P.

Wonderful. My very first unit was that same one or similar. But I forgot to set settings to 'leave in memory' and when it switched to another project it died. How often do these topology units get issued? Ballpark figure?

---

Founder of BOINC GROUP - Objectivists - Philosophically minded rational data crunchers.

Wonderful. My very first unit was that same one or similar. But I forgot to set settings to 'leave in memory' and when it switched to another project it died. How often do these topology units get issued? Ballpark figure?

All the currently issued work units have names of the form *topology_sample* (the ones David described in his post). So the next one you get will be one of these.