aboutsummaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
2013-10-09Integration verbosity fixesThomas White
2013-10-09Set matrix components to zero before calculating bg matrix integralsThomas White
2013-10-08Read scalar values from HDF5 files as 0d arraysThomas White
Historically, scalar values have been represented in our data files as 1d arrays with size=1. Now they are (more correctly) written as 0d arrays. This change allows CrystFEL to read both.
2013-10-07Update configure.acThomas White
2013-10-03Update man pageThomas White
2013-09-28configure.ac: Fix test for gdk-pixbufThomas White
2013-09-20partial_sim: Fix racy lockingThomas White
2013-09-19hdfsee: Use beam parameters file (--beam option)Thomas White
2013-09-17Tweak Rsplit_surfaceThomas White
2013-09-16Update Makefile.amThomas White
2013-09-16Add Rsplit_surface and clean-stream.pyThomas White
2013-09-13Monoclinic changes for MOSFLMThomas White
2013-09-13Stop PR when mean change in partiality is smallThomas White
2013-09-13Debugging/verbosity tweaksThomas White
2013-09-13Add tests/pr_pl_gradient_check for testing gradient of total correction factorsThomas White
2013-09-13partialator: Debugging stuffThomas White
2013-09-13Fix incorrect handling of monoclinic C cellsThomas White
2013-09-13partialator: Use read/write lockThomas White
2013-09-13partial_sim: Use read/write lockThomas White
2013-09-13Add INDEXING_SIMULATION for streams made by partial_simThomas White
2013-09-05Remove vestigial get_wavelength()Thomas White
2013-09-05configure: check for curses, and compile even if it's not availableThomas White
2013-09-04hdfsee: Add accelerators for "Show features" and "Resolution rings"Thomas White
2013-08-26Fix previous commitThomas White
2013-08-26Display indexing yield as a percentageThomas White
2013-08-20Re-order peak testsThomas White
The old version led to a curious situation where the number of saturated peaks could be much greater than the number of peaks, even without --no-use-saturated.
2013-08-20process_hkl: Fix --histogramThomas White
List of values now expands as necessary Values are no longer double counted when using --scale
2013-08-20get_hkl: Add --cutoff-angstromsThomas White
2013-08-12Improve show_matrix_eqn()Thomas White
2013-08-08render_hkl: Fix it once and for allThomas White
2013-08-08Add show_matrix()Thomas White
2013-08-08Show determinant when complaining about a non-integer matrix resultThomas White
2013-08-01Count filtered eigenvaluesThomas White
2013-08-01Use a struct to contain information for scaling reportThomas White
2013-08-01Count filtered eigenvaluesThomas White
2013-08-01Fix Lorentz gradientThomas White
2013-08-01Add test for gradient of Lorentz factor (as well as partiality)Thomas White
2013-08-01Fix previous commitThomas White
2013-07-31Flag crystals as "not refined" if SVD failsThomas White
2013-07-31Refine divergence as wellThomas White
2013-07-31Rescale the normal equations before attempting to solveThomas White
This makes the equations insensitive to the units of the parameters being refined, e.g. divergence is measured in radians whereas cell parameters are measured in m^-1, so their magnitudes are very different.
2013-07-31Set C=0 for gsl_cblas_dgemm()Thomas White
2013-07-31Use the correct reference list for selectionThomas White
2013-07-31Tweak rejection criteriaThomas White
2013-07-31Count reflections gained and lost during initial partiality calculationThomas White
2013-07-31Count number of "dud" crystals each iteration, not just at the endThomas White
2013-07-31Tweak error pathThomas White
2013-07-31Require scaling convergence to 5%, not 1%Thomas White
2013-07-31Revert refinement step if too many reflections are lostThomas White
2013-07-31partialator: New convergence criterion for scalingThomas White
Instead of stopping iteration when the absolute value of any scaling factor changes by more than a certain (small) amount, calculate the ratios of the scaling factors to their previous values, and stop when no scaling factor changes by more than 1% compared to the mean ratio. This method is robust against "drifting" of the scale factors when the partiality estimates are poor.