A group of astrophysicists has notched a pair of corrections for papers on galaxy clusters, thanks to an error that affected several figures in the papers, but not the overall conclusions.
The errors came in the catalog of “mock” galaxies that first author Fabio Zandanel, a postdoc at the University of Amsterdam, created to model features that are found in clusters of galaxies. Two mistakes canceled each other out “almost perfectly,” says Zandanel, making the changes that resulted from them subtle.
Zandanel explained the errors to us:
I have a quite long code that does all the needed calculations that are common for both papers and create these mock catalogues of objects (clusters of galaxies). The problem was that there were two missing factors in the code: one in a mass conversion from one definition to another, and one in the assignation of the gas-fraction quantity for each of the objects in the mock catalogues. These two factors were redshift-dependent (i.e., distance-dependent) scaling factors and, incidentally, their final effect cancelled out almost perfectly such that I did not realise the mistake earlier.
Those errors affected two papers, both published by the Monthly Notices of the Royal Astronomical Society.
First up, “A phenomenological model for the intracluster medium that matches X-ray and Sunyaev-Zel’dovich observations,” which has been cited seven times, according to Thomson Scientific’s Web of Knowledge.
The correction — the second part of which is behind a paywall, tsk tsk — has to do with the “mock” galaxies that they made for the model. It fixes:
a mistake in the mass conversion and gas-fraction assignation of the multifrequency mock catalogues.
The correction changes several figures and tables, though it doesn’t affect the conclusion:
This mistake was redshift dependent and affected in some amount most tables and figures except for fig. 1 and table 1 of the original paper. We provide here updated figures and tables (see Fig. 1, 2 and 3, and Table 1, 2 and 3). We emphasize that these changes do not affect the overall conclusions of the paper, and that the publicly available version of the catalogues (Multi-Dark database: www.multidark.org; Riebe et al. 2013) have also been corrected.
Part of the discussion is affected, the note explains:
The comparison of the YX–M500 scaling relation of our sample with observations, shown in the right-hand panel of Fig. 1, is slightly worse than before, but the corresponding discussion in the original paper is still valid. Most noticeably, the correction of the redshift-dependent mistake in the catalogues improved the comparison of the YSZ–M scaling relation of our sample with the Atacama Cosmology Telescope (ACT) sample at z = 0.53 (Hasselfield et al. 2013) as shown in the right-hand panel of Fig. 3, and therefore the corresponding discussion on the possible reasons for the larger discrepancy in the original paper should be disregarded.
The correction notice also provides new figures.
The second paper is “On the physics of radio haloes in galaxy clusters: scaling relations and luminosity functions,” which has been cited 10 times. It has its own correction note — also partly paywalled — which starts out the same as the one above, and then describes some specific changes in the paper, before continuing on to new figures:
We provide the updated values of parameters needed for the analytical model of the luminosity function n(L, z) at 120 MHz (see footnote 8 of the original paper): A0,0 = −484.74, A0,1 = 141.50, A1,0 = 32.66, A1,1 = −9.07, A2,0 = −0.55 and A2,1 = 0.14. It is worth mentioning that while Figs 3 and 4 of the original paper are insignificantly changed and, therefore, not shown here, the luminosity functions of Figs 1 and 2 are slightly shifted towards lower luminosities and with a mildly steeper slope with respect to the previously published version. More noticeably, the cumulative number of radio haloes (RHs) above a given flux and the ones detectable by Low Frequency Array (LOFAR) Tier 1, both at 120 MHz, of Fig. 3 are slightly more than before due to the steeper slope of the luminosity functions. In particular, the total number of detectable RHs at 120 MHz by LOFAR Tier 1 is of about 1400 (instead of the ∼600 originally stated).
Zandanel told us that though the correction affects a lot of figures, it doesn’t affect them much:
As it is written in the errata, these mistakes does not impact at all the conclusions of the papers. A large part of the second paper is even totally independent from this. The errata give a percentage correction to the final numbers in the original papers. Most of the figures are basically indistinguishable by eye with respect to the old version. However, I wanted to make the community aware of the problem in the original papers and of the correction in the clearer way.
He told us that the mistake was an accident:
I believe that during the work, that was quite long in time, these two factors somehow slipped out of the final version of code that I used for the generation of the results for the papers.
And how he found it:
I was doing new calculations that had several parts in common with this older work and when looking back at the code, to re-use some of it, I noticed it.
The editor in chief of the journal, David Flower, echoed what Zandanel told us:
The author informed us that he had discovered an error in the computation of certain quantities of the catalogue. The error affected some of the results in two papers, where use of the catalogue had been made, but did not affect the overall conclusions of those papers. Accordingly the author submitted (two) errata, which is the customary and appropriate procedure in such circumstances; retraction would not have been justified.
Like Retraction Watch? Consider making a tax-deductible contribution to support our growth. You can also follow us on Twitter, like us on Facebook, add us to your RSS reader, and sign up on our homepage for an email every time there’s a new post. Click here to review our Comments Policy.