PDA

View Full Version : tension in supernovae databases



trinitree88
2010-Apr-08, 09:29 PM
Sometimes one database gives numbers that do not fit the model you conjure up from a different database. You have a variety of ways to address this: 1.one could construct a new model that fits all the data....or 2. one could just cut out all the data that doesn't fit the model to make the data "look" better ( a number of scientists have gotten themselves in big trouble historically, nevertheless, see:)http://arxiv.org/PS_cache/arxiv/pdf/0906/0906.0828v3.pdf

Jerry
2010-Apr-10, 10:17 PM
Based on the results of this work, we suggest two truncated versions of the Union and Constitution
datasets, namely the UnionT and ConstitutionT SNIa samples, whose behaviors are more regular.

Who gets to define 'regular'? We should be looking at these objects as something we don't understand, rather than cutting and trimming and pasting them until they conform with a pet theory.

EDG
2010-Apr-11, 01:07 AM
Who gets to define 'regular'? We should be looking at these objects as something we don't understand, rather than cutting and trimming and pasting them until they conform with a pet theory.

Who says that they're not doing that? At least this way the outliers can be isolated for later examination.

Jerry
2010-Apr-12, 12:26 AM
Cutting out the outliers is always a dangerous way to improve the uniformity, the fit, of data being bumped up against a hypothesis. This is especially important when you realise that the original hypothesis; that the universe is experiencing an accelerating expansion, was based upon a much smaller subset of the same type of data. Also, many outlying supernova observations were cut from the sample during the construction of the gold, union and constitution sets.

It would be equally valid to leave the outliers in the sample and use the tension in the data to demonstrate that the hypothesis has failed.