cosmo workshop Fri 15:00 12.03.2010: Li and Liu arXiv:1003.1073: WMAP quadrupole = antenna direction interpolation error
witam cosmo-torun
Cosmo workshop WHEN: Fri 15:00 12.03.2010 WHO: boud TITLE: Liu and Li arXiv:1003.1073 ("The origin of the WMAP quadrupole") ABSTRACT: The WMAP quadrupole is apparently mostly a pointing error. The true quadrupole is much closer to zero. The missing fluctuations problem has just got much worse for the infinite-flat-universe hypothesis.
Liu and Li arXiv:1003.1073 ("The origin of the WMAP quadrupole") have posted a very interesting article (presumably submitted to Nature, guessing from the style and length). The best estimate of the CMB quadrupole is apparently... zero!
They claim that the WMAP quadrupole comes from a single error - an antenna direction representation error by half of an observational angular interval. The discussion concerns quaternions, but don't be frightened - it presumably uses the imaginary part only, to represent X, Y, Z directions. You can think of them as vectors in R^3 if you like.
Order of magnitude check of their calculation:
Liu and Li say that the error is by 7' and that this causes incorrect subtraction of the dipole in the time-ordered-date (TOD). The direction of the error varies as the direction of observation varies, so it's reasonable that it doesn't give a simple offset detectable by post-processing in the analysis pipeline, IMHO (though Bartek may have another opinion).
The dipole is about 3.3mK (e.g. section 7 Bennett et al 2003). sin(7') * 3.3mK = 0.002 * 3.3mK = 6.7 microK
Liu and Li say 10-20 microK, just slightly higher, but their Eq.(1) is a motivational equation - it is not used as an entry to their data analysis pipeline.
Figures 1 and 2 should the effect very dramatically. Rephrasing what they've said:
Figure 1 left: difference between Liu Li analysis, using the same wrong method that the WMAP team used. This is a check that their pipeline does the same thing that the WMAP team does. |Difference| < 2 or so microK.
Figure 1 right: Liu Li using correct antenna directions on WMAP Q1 3yr data. They get a strong quadrupole!
Figure 2 left: The effect of using a wrong dipole, calculated using *only the directional information* from the time-ordered-data from the spacecraft, with *no CMB data*. See paragraph 2, page 4: "only the spacecraft attitude information is used to compute d' ."
Figure 2 right: official WMAP5 V+W quadrupole.
Figures 1 right, and 2 left look very, very similar to Figure 2 right. There seems to be a slight difference in angular position visible by eye, but the coincidence is striking.
Can we really believe that the CMB quadrupole just by chance happens to be very strongly aligned with and of nearly identical amplitude to a map made using only the time-ordered-data of the spacecraft direction ("attitude") and the spacecraft orbit around the Sun, and the Sun's motion with respect to the CMB?
Since this is a Nature-type article, the authors are forced to exclude almost all interesting details from the article, but info like the spacecraft attitude quaternions and interpolation method should not be too difficult to find.
* Bennett et al 2003: arXiv:astro-ph/0302207 http://arXiv.org/abs/astro-ph/0302207 * Liu, Li 2010: arXiv:1003.1073 http://arXiv.org/abs/1003.1073
The good thing about this is that it shows that a lot of very intelligent people - those in a mega-collaboration plus many outside of the official group - can spend 7 years looking closely at an important set of observational data without finding what appears to be a very elementary error in the analysis details, with a very fundamental consequence.
The fundamental consequence is that a lower quadrupole makes Infeld's (1949) prediction of the equivalent of the low l cutoff due to the shape of the Universe even stronger! Liu & Li suggest an inflationary argument, but their interpretation is not what is important - they don't want the referee to make a fuss about interpretations, since their observational argument is so strong.
pozdr boud
witam
This is being discussed on cosmocoffee as well: http://cosmocoffee.info/viewtopic.php?t=1537 Liu is participating in the discussion. He contacted Hinshaw but yesterday said that he had not yet received a reply. The software for analysing the TOD is also linked in the discussion.
Looking forward to this afternoon's discussion, :) pozdr boud
On Mon, 8 Mar 2010, Boud Roukema wrote:
witam cosmo-torun
Cosmo workshop WHEN: Fri 15:00 12.03.2010 WHO: boud TITLE: Liu and Li arXiv:1003.1073 ("The origin of the WMAP quadrupole") ABSTRACT: The WMAP quadrupole is apparently mostly a pointing error. The true quadrupole is much closer to zero. The missing fluctuations problem has just got much worse for the infinite-flat-universe hypothesis.
Liu and Li arXiv:1003.1073 ("The origin of the WMAP quadrupole") have posted a very interesting article (presumably submitted to Nature, guessing from the style and length). The best estimate of the CMB quadrupole is apparently... zero!
They claim that the WMAP quadrupole comes from a single error - an antenna direction representation error by half of an observational angular interval. The discussion concerns quaternions, but don't be frightened - it presumably uses the imaginary part only, to represent X, Y, Z directions. You can think of them as vectors in R^3 if you like.
Order of magnitude check of their calculation:
Liu and Li say that the error is by 7' and that this causes incorrect subtraction of the dipole in the time-ordered-date (TOD). The direction of the error varies as the direction of observation varies, so it's reasonable that it doesn't give a simple offset detectable by post-processing in the analysis pipeline, IMHO (though Bartek may have another opinion).
The dipole is about 3.3mK (e.g. section 7 Bennett et al 2003). sin(7') * 3.3mK = 0.002 * 3.3mK = 6.7 microK
Liu and Li say 10-20 microK, just slightly higher, but their Eq.(1) is a motivational equation - it is not used as an entry to their data analysis pipeline.
Figures 1 and 2 should the effect very dramatically. Rephrasing what they've said:
Figure 1 left: difference between Liu Li analysis, using the same wrong method that the WMAP team used. This is a check that their pipeline does the same thing that the WMAP team does. |Difference| < 2 or so microK.
Figure 1 right: Liu Li using correct antenna directions on WMAP Q1 3yr data. They get a strong quadrupole!
Figure 2 left: The effect of using a wrong dipole, calculated using *only the directional information* from the time-ordered-data from the spacecraft, with *no CMB data*. See paragraph 2, page 4: "only the spacecraft attitude information is used to compute d' ."
Figure 2 right: official WMAP5 V+W quadrupole.
Figures 1 right, and 2 left look very, very similar to Figure 2 right. There seems to be a slight difference in angular position visible by eye, but the coincidence is striking.
Can we really believe that the CMB quadrupole just by chance happens to be very strongly aligned with and of nearly identical amplitude to a map made using only the time-ordered-data of the spacecraft direction ("attitude") and the spacecraft orbit around the Sun, and the Sun's motion with respect to the CMB?
Since this is a Nature-type article, the authors are forced to exclude almost all interesting details from the article, but info like the spacecraft attitude quaternions and interpolation method should not be too difficult to find.
- Bennett et al 2003: arXiv:astro-ph/0302207 http://arXiv.org/abs/astro-ph/0302207
- Liu, Li 2010: arXiv:1003.1073 http://arXiv.org/abs/1003.1073
The good thing about this is that it shows that a lot of very intelligent people - those in a mega-collaboration plus many outside of the official group - can spend 7 years looking closely at an important set of observational data without finding what appears to be a very elementary error in the analysis details, with a very fundamental consequence.
The fundamental consequence is that a lower quadrupole makes Infeld's (1949) prediction of the equivalent of the low l cutoff due to the shape of the Universe even stronger! Liu & Li suggest an inflationary argument, but their interpretation is not what is important - they don't want the referee to make a fuss about interpretations, since their observational argument is so strong.
pozdr boud
Cosmo-torun mailing list Cosmo-torun at cosmo.torun.pl http://cosmo.torun.pl/mailman/listinfo/cosmo-torun
hi cosmo-torun,
Following our March discussion, i checked the timing offset error suggested by Liu, Xiong & Li (ArXiv:1003.1073).
The pipeline for creating maps out of raw WMAP time-ordered-data (TOD, i.e. individual observations) is:
(1) calibration: convert uncalibrated TOD to calibrated TOD (2) mapmaking: put the calibrated TOD together in pixels to make maps
The 25.6 ms timing offset did not cause an error during step (2). This is shown in http://arXiv.org/abs/1004.4506 , A&A in press.
However, it did cause an error during step (1). This is shown in http://arXiv.org/abs/1007.5307 , A&A submitted. The analyses in both cases use step (2).
So, the maps at http://lambda.gsfc.nasa.gov/product/map/current/ are wrongly calibrated. The quadrupole from W band maps made from the wrongly calibrated TOD is about 15-60% overestimated, based on an approximate correction for the wrong calibration.
On Mon, 8 Mar 2010, Boud Roukema wrote:
The fundamental consequence is that a lower quadrupole makes Infeld's (1949) prediction of the equivalent of the low l cutoff due to the shape of the Universe even stronger! Liu & Li suggest an inflationary
i found the paper - Infeld did *not* talk about a cutoff in perturbation statistics at low l. He did say that solutions to Maxwell's equations must take into account the whole manifold. This could be called an infrared cutoff argument, even though in the article, Infeld didn't think of it physically in terms of a fluctuation spectrum cutoff that gives an observational sign of cosmic topology, and didn't make a prediction of an observational effect. Probably this should remain credited to Starobinsky 1993 and Stevens et al 1993.
pozdr boud
On Fri, 12 Mar 2010, Boud Roukema wrote:
witam
This is being discussed on cosmocoffee as well: http://cosmocoffee.info/viewtopic.php?t=1537 Liu is participating in the discussion. He contacted Hinshaw but yesterday said that he had not yet received a reply. The software for analysing the TOD is also linked in the discussion.
Looking forward to this afternoon's discussion, :) pozdr boud
On Mon, 8 Mar 2010, Boud Roukema wrote:
witam cosmo-torun
Cosmo workshop WHEN: Fri 15:00 12.03.2010 WHO: boud TITLE: Liu and Li arXiv:1003.1073 ("The origin of the WMAP quadrupole") ABSTRACT: The WMAP quadrupole is apparently mostly a pointing error. The true quadrupole is much closer to zero. The missing fluctuations problem has just got much worse for the infinite-flat-universe hypothesis.
Liu and Li arXiv:1003.1073 ("The origin of the WMAP quadrupole") have posted a very interesting article (presumably submitted to Nature, guessing from the style and length). The best estimate of the CMB quadrupole is apparently... zero!
They claim that the WMAP quadrupole comes from a single error - an antenna direction representation error by half of an observational angular interval. The discussion concerns quaternions, but don't be frightened - it presumably uses the imaginary part only, to represent X, Y, Z directions. You can think of them as vectors in R^3 if you like.
Order of magnitude check of their calculation:
Liu and Li say that the error is by 7' and that this causes incorrect subtraction of the dipole in the time-ordered-date (TOD). The direction of the error varies as the direction of observation varies, so it's reasonable that it doesn't give a simple offset detectable by post-processing in the analysis pipeline, IMHO (though Bartek may have another opinion).
The dipole is about 3.3mK (e.g. section 7 Bennett et al 2003). sin(7') * 3.3mK = 0.002 * 3.3mK = 6.7 microK
Liu and Li say 10-20 microK, just slightly higher, but their Eq.(1) is a motivational equation - it is not used as an entry to their data analysis pipeline.
Figures 1 and 2 should the effect very dramatically. Rephrasing what they've said:
Figure 1 left: difference between Liu Li analysis, using the same wrong method that the WMAP team used. This is a check that their pipeline does the same thing that the WMAP team does. |Difference| < 2 or so microK.
Figure 1 right: Liu Li using correct antenna directions on WMAP Q1 3yr data. They get a strong quadrupole!
Figure 2 left: The effect of using a wrong dipole, calculated using *only the directional information* from the time-ordered-data from the spacecraft, with *no CMB data*. See paragraph 2, page 4: "only the spacecraft attitude information is used to compute d' ."
Figure 2 right: official WMAP5 V+W quadrupole.
Figures 1 right, and 2 left look very, very similar to Figure 2 right. There seems to be a slight difference in angular position visible by eye, but the coincidence is striking.
Can we really believe that the CMB quadrupole just by chance happens to be very strongly aligned with and of nearly identical amplitude to a map made using only the time-ordered-data of the spacecraft direction ("attitude") and the spacecraft orbit around the Sun, and the Sun's motion with respect to the CMB?
Since this is a Nature-type article, the authors are forced to exclude almost all interesting details from the article, but info like the spacecraft attitude quaternions and interpolation method should not be too difficult to find.
- Bennett et al 2003: arXiv:astro-ph/0302207 http://arXiv.org/abs/astro-ph/0302207
- Liu, Li 2010: arXiv:1003.1073 http://arXiv.org/abs/1003.1073
The good thing about this is that it shows that a lot of very intelligent people - those in a mega-collaboration plus many outside of the official group - can spend 7 years looking closely at an important set of observational data without finding what appears to be a very elementary error in the analysis details, with a very fundamental consequence.
The fundamental consequence is that a lower quadrupole makes Infeld's (1949) prediction of the equivalent of the low l cutoff due to the shape of the Universe even stronger! Liu & Li suggest an inflationary argument, but their interpretation is not what is important - they don't want the referee to make a fuss about interpretations, since their observational argument is so strong.
pozdr boud
Cosmo-torun mailing list Cosmo-torun at cosmo.torun.pl http://cosmo.torun.pl/mailman/listinfo/cosmo-torun
Cosmo-torun mailing list Cosmo-torun@cosmo.torun.pl http://cosmo.torun.pl/mailman/listinfo/cosmo-torun
-
Boud Roukema