Results for the #DFT13poll
Last week we finally managed to find the time to finish the news-item for the #DFT2013poll, i.e. the online popularity poll of density functionals.
As can be seen in the Figure above, the number of entries again has increased (to 194, but see below), and hence the total number of points of the winner (PBE) has increased steadily. However, it should be mentioned at the same time that over these four years there has been somewhat of a fluctuation in the average number of points given (Navg = Ntotal_primera / 20 / Nentries). This number has gone from 0.79 (2010), 0.82 (2011), 0.69 (2012) to 0.77 (2013). Apart from the year 2012, where it was substantially lower, it has been more or less constant at around 0.8.
A shocking experience. We have had to make a drastic decision, because of "attempts to bias the outcome of the poll through repetitive single-answer entries (with only 1 option liked: optB88-vdW), which often were added at a pace of one per 8 seconds; a total of 56 of these single-answer optB88-vdW entries had been entered" (a quote from the 2013 news-item). Just to give an example of such a series of single-answer entries:
1/6/2013 10:38:24
1/6/2013 10:38:31
1/6/2013 10:38:39
1/6/2013 10:38:47
1/6/2013 10:38:54
1/6/2013 10:39:02
1/6/2013 10:39:10
1/6/2013 10:39:18
1/6/2013 10:39:26
1/6/2013 10:39:33
Note that we do not have any information about from where these entries were made (nor do we want to know: we want the poll to be as anonymous as possible). And we must add that we are sincerely sorry for the authors of the optB88-vdW functional, who have fallen victim to this, but we have seen no other way out than to disqualify the functional. We counted on the integrity and fairness of the people in the computational chemistry community to not try to bias the outcome of the poll, but simply state their preference for density functionals only once. Apparently, we will have to take measures to try to avoid this situation from happening again.
Summarizing:
PBE has once again been the winner of the #DFT2013 poll, followed by PBE0 (a.k.a. PBE1PBE) and B3LYP. Range-separated hybrids seem to gain in popularity, while dispersion-corrected functionals remain less popular than their non-corrected versions (surprisingly).
 |
| Summary of results for density functionals polls (2010-2013) |
A shocking experience. We have had to make a drastic decision, because of "attempts to bias the outcome of the poll through repetitive single-answer entries (with only 1 option liked: optB88-vdW), which often were added at a pace of one per 8 seconds; a total of 56 of these single-answer optB88-vdW entries had been entered" (a quote from the 2013 news-item). Just to give an example of such a series of single-answer entries:
1/6/2013 10:38:24
1/6/2013 10:38:31
1/6/2013 10:38:39
1/6/2013 10:38:47
1/6/2013 10:38:54
1/6/2013 10:39:02
1/6/2013 10:39:10
1/6/2013 10:39:18
1/6/2013 10:39:26
1/6/2013 10:39:33
Note that we do not have any information about from where these entries were made (nor do we want to know: we want the poll to be as anonymous as possible). And we must add that we are sincerely sorry for the authors of the optB88-vdW functional, who have fallen victim to this, but we have seen no other way out than to disqualify the functional. We counted on the integrity and fairness of the people in the computational chemistry community to not try to bias the outcome of the poll, but simply state their preference for density functionals only once. Apparently, we will have to take measures to try to avoid this situation from happening again.
Summarizing:
PBE has once again been the winner of the #DFT2013 poll, followed by PBE0 (a.k.a. PBE1PBE) and B3LYP. Range-separated hybrids seem to gain in popularity, while dispersion-corrected functionals remain less popular than their non-corrected versions (surprisingly).
Comments
Post a Comment