Age | Commit message (Collapse) | Author |
|
|
|
|
|
|
|
The reindexing is really going to create a whole world of pain. Even if
we can make it work correctly, it still might come as a surprise to
people when the exported data is indexed differently (e.g. to data
exported to some other format). Let's just forget the whole thing, and
refuse to export to formats which don't have a way to represent the data
we have. Perhaps revisit this later if it becomes a big problem and we
want to put in the effort required.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
If it's needed, we'll replace it with a better piece of API (...which
doesn't abort the program in the event of a typo!)
|
|
It would've been difficult to calculate this in the new unified FoM API,
and the figure is completely useless anyway.
If you disagree with both of the above statements, please get in touch!
|
|
Reasons for differences:
1. Resolution shells slightly different
The binning calculation needs to take into account small rounding errors
in the resolution calculation, when not using an explicit resolution
range (--highres). The old version did this by taking a min/max
resolution range slightly larger than the resolution of the data. The
new version handles the rounding errors explicitly, so does not need
this.
2. Number of reflections with infinite/invalid I/sigI values halved
The number reported for this count was twice what it should have been,
due to a bug in the old check_hkl.
3. Overall SNR is different
When the above warning message applies, the old version still allowed
the reflections with invalid I/sigI values to contribute to the
denominator of the mean SNR calculation. The new version does not
include them in the SNR calculation at all. Note that the reflections
contribute to the other figures of merit unless otherwise stated.
4. Standard deviation of intensity is not calculated
It would've been a lot of work to include this in the new version, and
it's a totally useless number. If you disagree, please get in touch!
|
|
|
|
|
|
This prevents an out-of-memory situation (due to ulimit) from
manifesting as a segfault.
|
|
|
|
No-one uses it, it doubles the complexity of the code, and the manual
even warns not to use it.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Bad regions specified in terms of x/y still require an iteration over
all pixels of the detector, but I don't see an easy way around that.
Avoiding x/y bad regions will give best performance.
|
|
Previously, the "getter" functions would re-calculate the requested
representation every time they were called. This could mean doing a
matrix inversion in the middle of a tight loop, wasting loads of time.
Now, it stores the calculated values and returns them directly next
time. Setting the parameters invalidates the values for all
representations other than the one given.
The cost of doing this is that the cell can no longer be "const" in the
getter functions. This tracked through some other code, but nothing too
severe.
|
|
Lots of improvements still to be made here, such as using the "real"
Cairo text API as well as using proper overlines for negative indices.
|
|
The criterion for "too large" is 20% of the 1/d value for the lowest
reflection which is not systematically absent according to the
centering.
A profile radius larger than the 1/d value for a reflection will crash
the xsphere partiality model, and some visualisation shows that this is
a clearly non-physical situation. The profile radius shouldn't be
anywhere near the inter-Bragg spacing for reasonable data.
However, feedback shows that this is happening quite often in real data,
probably due to bad indexing.
|
|
|
|
|
|
|
|
Because shortly we will have merging_result as well
|
|
This makes the interface consistent between the indexing options and the
merging options.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|