tag:blogger.com,1999:blog-9633767.post8226379936374310023..comments2024-03-25T09:11:17.877-07:00Comments on The Curious Wavefunction: On reproducibility in modelingWavefunctionhttp://www.blogger.com/profile/14993805391653267639noreply@blogger.comBlogger4125tag:blogger.com,1999:blog-9633767.post-33982826265663604182011-12-22T15:30:52.249-08:002011-12-22T15:30:52.249-08:00Thanks for that interesting piece of information. ...Thanks for that interesting piece of information. You know your model is in trouble when <br />a. It can be used to justify any result.<br />b. It always predicts what you see.Wavefunctionhttps://www.blogger.com/profile/14993805391653267639noreply@blogger.comtag:blogger.com,1999:blog-9633767.post-42479676177032256222011-12-22T12:10:15.286-08:002011-12-22T12:10:15.286-08:00There has been a similar problem in functional MRI...There has been a similar problem in functional MRI work (fMRI) showing which areas of the brain were activated by particular mental activities. You never saw the original data, just the final images, which usually were quite tweaked. <br /><br />I began to suspect the results, when fMRIs always confirmed the initial hypothesis of the article. There were very few unexpected results. Scientific inquiry just isn't like that. The result was the following Science 2 August '02 p. 749 it is stated that "neuroimaging has a reputation for producing "pretty pictures" but not replicable data. It has been characterized as pseudocolor phrenology.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-9633767.post-13926554391482450422011-12-14T13:38:01.663-08:002011-12-14T13:38:01.663-08:00I remember one of the cases: running GOLD several ...I remember one of the cases: running GOLD several times for the same protein-ligand complex will give you different results. Because of the genetic algorithm used, but if you provide the seed used - it will give the almost the same result.<br /><br />So, we need standards! PMML is not the best choice...maybe some other solutions.Vladimir Chupakhinhttps://www.blogger.com/profile/14838130425318070954noreply@blogger.comtag:blogger.com,1999:blog-9633767.post-91200594464736110862011-12-13T12:41:08.674-08:002011-12-13T12:41:08.674-08:00Computational scientists serious about reproducibi...Computational scientists serious about reproducibility might benefit from this new software project, <a href="http://packages.python.org/Sumatra/index.html" rel="nofollow">Sumatra</a>Chrishttps://www.blogger.com/profile/12005597475962237728noreply@blogger.com