journal menu
![[Figure 5]](ba5308fig5mag.jpg)
|
|
Figure 5
Contour plots illustrating, for the acentric case, the errors in likelihood scores for a perfect model arising from using French and Wilson amplitudes in the inflated-variance Rice likelihood target as a function of ZO and ![[\sigma_{Z_{\rm O}}]](teximages/ba5308fi2.gif) . (a) The expected r.m.s. error in the likelihood score (converted to bits for comparison with information gain), averaged over the calculated structure factors consistent with the measurement, with contour lines drawn from the blue region through orange to yellow at 0.5, 1, 1.5, 2 and 2.5. (b) The ratio of the r.m.s. error from (a) and the information gain from Fig. 1![[link]](../../../../../../logos/arrows/d_arr.gif) (a), with contour lines drawn from the blue region through orange to yellow at 0.1, 1, 10 and 100. Only well measured intensities have likelihood errors that are smaller than the information content gained (points below the second contour line). |
![[Figure 5]](ba5308fig5.jpg)
 | STRUCTURAL BIOLOGY |
ISSN: 2059-7983
access
