Open Access
Table 4.
Numerical stability of each statistic.
Statistics | No | 100 | 1000 |
---|---|---|---|
(Results are log10) | noise | hours | hours |
![]() |
−12.25 | −12.20 | −12.51 |
![]() |
−11.87 | −9.24 | −10.14 |
![]() |
−11.57 | −9.13 | −9.11 |
![]() |
−10.66 | −8.92 | −7.97 |
![]() |
−11.20 | −8.64 | −9.14 |
![]() |
−10.28 | −7.17 | −10.03 |
![]() |
−10.08 | −7.73 | −7.38 |
![]() |
−10.23 | −9.16 | −9.70 |
![]() |
−10.40 | −8.88 | −9.80 |
Notes. Looking at the numerical stability of the different statistics, we multiply the inverse of the covariance by the covariance, which should be an identity matrix. We show the largest off-diagonal term, as a test for numerical stability.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.