O'Grady, Paul Christopher   
Fri 2/24/2017 12:16 PM
To:   Liang, Mengning;  
Cc:   Dubrovin, Mikhail;  
Suggested Meetings

Hi Meng,

I talked with Mikhail about the fiber-diffraction “sprint” to get 
results for Germany that you have strong confidence in.  We’ll come by 
your office around 4 today to launch.

We’ll also touch base Monday, Tuesday, Wednesday, Thursday next week at 
3pm to make sure we’re headed in the right direction.  Those are allowed
 to be brief (more important to do real work).

chris

O'Grady, Paul Christopher   
Fri 2/24/2017 5:06 PM 
To: Liang, Mengning <mliang@slac.stanford.edu>; Dubrovin, Mikhail <dubrovin@slac.stanford.edu>;  

meng provides "100 events” (good and bad) in run 169
meng provides her current psocake-peak-params
meng provides problematic l=0,l=1 indexing output, mikhail fixes code

mikhail: move to psocake-consistent parameters to make life easier for meng
mikhail: do fraser_xyz and fraser have same conventions and yield same results?
mikhail: peak-fit needs to output qh and qv for the *peak* (not lattice) for images that don't index for the 100 bad events
mikhail: check after event-selection most things should fit (70-80%) and most should index
mikhail: as a longer-term target modify the pipeline so that meng can look at outputs at various stages more easily

all: tweak peak-finding parameters for the 100 events
all: ask chuck what should we use for background

Liang, Mengning   
Mon 2/27/2017 3:14 PM
To:   Dubrovin, Mikhail;  
Cc:   O'Grady, Paul Christopher;  
        
Peakfinding parameters : 
For arc:
peaks_arc = alg_arc.peak_finder_v1(nda, thr_low=40, thr_high=160, radius=8, dr=0.5)

For equator:
peaks_equ = alg_equ.peak_finder_v4(nda, thr_low=60, thr_high=100, rank=4, r0=4, dr=0.5)
-----------------------------------------------------------------------------
Peak selection:
For arc:
alg_arc = PyAlgos(windows=winds_arc, mask=mask_arc, pbits=0)
alg_arc.set_peak_selection_pars(npix_min=20, npix_max=150, amax_thr=100, atot_thr=2000, son_min=5)

For equator:
alg_equ = PyAlgos(windows=winds_equ, mask=mask_equ, pbits=0)
alg_equ.set_peak_selection_pars(npix_min=5, npix_max=150, amax_thr=100, atot_thr=600, son_min=4)

O'Grady, Paul Christopher   
Mon 2/27/2017 3:46 PM
  
Adding notes from today:
for peak-finding we will standardize on:
- median filter.
- peakfinderv4

good events should have 2-4 equatorial peaks, 0-2 arc peaks, OR 2 arc peaks and 1-3 eq peaks
80% of the good events should be found, with tuned peak-finding and the above event-selection criteria
switch to mode 5 common-mode (unbonded pixels)
meng file with equatorial peaks incorrectly indexed as l=1:
~mliang/analysis/rel-cxif5315/work1/index3d/test1/peak-idx-cxif5315-r0169-2017-01-23T10:39:03.txt

Liang, Mengning   
Mon 2/27/2017 3:48 PM
To:   Dubrovin, Mikhail;  
Cc:   O'Grady, Paul Christopher;  

Here is a list of good events (psocake event numbers)
They will have 2-4 peaks in the equatorial region 
or 
2 peaks in the arc region + 1-4 peaks in the equatorial region.
Thanks,
Meng