aglie/yell

The program for PDF analysis of diffuse scattering from single crystals

yell's Issues

Check there is enough operating memory

Proper filenames for weights etc

Introduce proper files for LSQ weights, file that scales reciprocal space, the one for resolution effects. Names should be clear so that it is obvious what to write there.

Properly set up openmp

Implement periodic boundary conditions switches for fft-algorithm

Redefine variables

Now redefinition of variables in input file does not work.

Check there are enough InitialParameters

Sometimes ~i eats more than one variable from stack.

example:

Cell 4 4 1 90 90 90
DiffuseScatteringGrid -2 -2 0 0.1 0.1 1 40 40 1
InitialParams 1 1 2 3 4 5 6 7 8 9
PointGroup -1
CalculationMethod direct
Refine false
RecalculateAverage true
DumpPairs true

UnitCell
[
Var=Variant[
(p=0.5)

C1 1 ~i 0 0 1 1 1 0 0 0

[
C2 1 1 0 0 1 1 1 0 0 0
]
]
]
Modes
[]
Correlations
[
[(0,0,0)
substitutional_correlation(Var,Var,0.5)
]
]

Dump refined pairs

In a separate file.

Problem with 3-fold laue symmetry

It only averages a hexagon, not the whole map.
Options:

• force weights to zeros where it does not average,
• apply it only on the level of pairs

Make class for output handling

Program a nice set of functions for handling multiple levels of output.
Make it testable.
Also make it shut up for tests!

Check laue groups

reject say m3mk

Read the parameter values from a separate file

The separate file could have a fixed name like 'variables.txt' or be called by a keyword like variables = variables.txt

Call rotational mode shear mode

Multiple grids

Automatic multiplicity

Now we have a toggle between automatic multiplicity for caleidoscopic Laue Groups. It is little bit inconsistent for rotational axes and center of inversion. Should we turn it off whatsoever?

Rotational modes should be in radians

or degrees.

"fft" algorithm inconsistent results compaired to "direct" algorithm

a=8.247;
b=2.9345;
c=4.6361;
Cell a b c 90. 90. 90.
step_a=0.065976;
step_b=1./3.;
step_c=0.0370888;
start_a=-9.038712;
start_b=-8./3.;
start_c=-5.0811656;
DiffuseScatteringGrid start_a start_b start_c step_a step_b step_c 274 16 274

InitialParams 0.0362431 0.392103 0.254694 0.174224 0.0649597 0.111

InitialParams 0.0450704 0.366663 0.255732 0.167755 0.0652468 0.0996195 0.111

InitialParams 0.0450072 0.366135 0.255362 0.167407 0.0646357 0.0657025 0.0726069 0.111

InitialParams 0.0449814 0.365235 0.255255 0.166843 0.0643781 0.0670622 0.0501103 0.0722321

InitialParams 0.055715 0.38668 0.244463 0.163527 0.027291 0.0418286 0.0270311 0.030538
horiz_0p5_0_0=~i;
horiz_1_0_0=~i;
horiz_1p5_0_0=~i;
horiz_0_0_1=~i;
horiz_0p5_0_1_short=~i;
horiz_0p5_0_1_long=~i;
horiz_1_0_1=~i;

PointGroup mmm

CalculationMethod direct

Refine false

FFTGridSize 16 16 16

As_x=0.25;
As_y=0.44774;
As_z=0.03948;

As_ADP_iso=0.0;

As_ADP_iso=0.01140;

As_m_y=1.-As_y;

z_shift=2*As_z;

DumpPairs true

UnitCell
[

As = Variant
[
(p=0.333333333333)
As1=[
As 1 As_x As_y As_z As_ADP_iso As_ADP_iso As_ADP_iso 0 0 0
As 1 As_x As_m_y+1 As_z As_ADP_iso As_ADP_iso As_ADP_iso 0 0 0
]

(p=0.333333333333)
As2=[
As  1  As_x As_y+1   As_z As_ADP_iso As_ADP_iso As_ADP_iso 0 0 0
As  1  As_x As_m_y+2 As_z As_ADP_iso As_ADP_iso As_ADP_iso 0 0 0
]

(p=0.333333333333)
As3=[
As  1  As_x As_y+2   As_z As_ADP_iso As_ADP_iso As_ADP_iso 0 0 0
As  1  As_x As_m_y   As_z As_ADP_iso As_ADP_iso As_ADP_iso 0 0 0
]

]

As_b = Variant
[
(p=0.333333333333)
As1b=[
As 1 As_x+1/2 As_y -As_z As_ADP_iso As_ADP_iso As_ADP_iso 0 0 0
As 1 As_x+1/2 As_m_y+1 -As_z As_ADP_iso As_ADP_iso As_ADP_iso 0 0 0
]

(p=0.333333333333)
As2b=[
As  1  As_x+1/2 As_y+1   -As_z As_ADP_iso As_ADP_iso As_ADP_iso 0 0 0
As  1  As_x+1/2 As_m_y+2 -As_z As_ADP_iso As_ADP_iso As_ADP_iso 0 0 0
]

(p=0.333333333333)
As3b=[
As  1  As_x+1/2 As_y+2   -As_z As_ADP_iso As_ADP_iso As_ADP_iso 0 0 0
As  1  As_x+1/2 As_m_y   -As_z As_ADP_iso As_ADP_iso As_ADP_iso 0 0 0
]

]

]

Modes
[
]

Correlations
[

[
(0,0,0)
multiplicity 1.
substitutional_correlation(As, As,
1/3, 0,
0, 1/3)
]

[
(0,3,0)
multiplicity 1.
substitutional_correlation(As, As,
1/3, 0,
0, 1/3)
]

shift_0p5_0_0=(1/3-horiz_0p5_0_0)/2.;

[

(0,0,0)

multiplicity 1.

substitutional_correlation(As, As_b,

horiz_0p5_0_0, shift_0p5_0_0,

shift_0p5_0_0, horiz_0p5_0_0)

]

shift_1_0_0=(1/3-horiz_1_0_0)/2;

[

(1,0,0)

multiplicity 1.

substitutional_correlation(As, As,

horiz_1_0_0, shift_1_0_0,

shift_1_0_0, horiz_1_0_0)

]

shift_1p5_0_0=(1/3-horiz_1p5_0_0)/2.;

[

(1,0,0)

multiplicity 1.

substitutional_correlation(As, As_b,

horiz_1p5_0_0, shift_1p5_0_0,

shift_1p5_0_0, horiz_1p5_0_0)

]

shift_0_0_1=(1/3-horiz_0_0_1)/2;

[

(0,0,1)

multiplicity 1.

substitutional_correlation(As, As,

horiz_0_0_1, shift_0_0_1,

shift_0_0_1, horiz_0_0_1)

]

shift_0p5_0_1_short=(1/3-horiz_0p5_0_1_short)/2.;

[

(0,0,1)

multiplicity 1.

substitutional_correlation(As, As_b,

horiz_0p5_0_1_short, shift_0p5_0_1_short,

shift_0p5_0_1_short, horiz_0p5_0_1_short)

]

shift_0p5_0_1_long=(1/3-horiz_0p5_0_1_long)/2.;

[

(0,0,1)

multiplicity 1.

substitutional_correlation(As_b, As,

horiz_0p5_0_1_long, shift_0p5_0_1_long,

shift_0p5_0_1_long, horiz_0p5_0_1_long)

]

shift_1_0_1=(1/3-horiz_1_0_1)/2;

[

(1,0,1)

multiplicity 1.

substitutional_correlation(As, As,

horiz_1_0_1, shift_1_0_1,

shift_1_0_1, horiz_1_0_1)

]

]

print "a " a

print "b " b

print "c " c

print "step_a " step_a

print "step_b " step_b

print "step_c " step_c

Detect and report incorrect refinements

Pij>0 Uij positively defined, constraints on SE are satisfied

Check lower limits and stepsize before doing fft

Grids with arbitrary steps

(not only along main axes)

FIX THE BUG IN ADPs beta = G'B/4

not sure what that means though

Document h5 format.

Write matlab scripts so that users could modify the .h5 files.

Change direct and fft names

to something reasonable

Variant should not introduce void silently

Symmetry operators only allow displacement of 1/2

Check that all pairs are within PDF grid

If not show warning or error, depending on periodic boundary conditions

Bug in Laue Symmetry for reciprocal space calculation

It now uses matrices for real space even in reciprocal space and screws up the six-fold rotation in FFT calculation method

Report pixelsize

both in direct and reciprocal space. also grids etc

Atom occupancy should work as multiplicity

Now atom probability does not change anything since it is redefined by the Variant.

special funcitons in expressions

Add log sqrt pow sin cos exp mod abs

Allow initial parameters to be multiline

Initialize parameters before average structure calculation

Now average structure is initialized with all-zeroes

Translational modes should be in angstroems

Laue symmetry bug in left-bottom line

Now the symmetry does not expand there

Check how symmetry for six-fold rotation works

Tests show that it is ok, but it seems like I was using the wrong algorithms up till now. Check on some example.

Check Void

Program accepts Void as an atomic type. Check that this atom does what expected.

Empty RefinableVariables block causes segmentation fault

as a consequence refinement of only the scale factor is not possible
semms to be difficult to reproduce

Allow formulas in InitialParams

Output R-factors in Yell

Also iteration-wise

fft mode when only hk0 layer provided

Right now fft mode will show a cut through PDF space, not a projection!!

Document clearly how occupancy correlation matrix runs

what P(1,3) means exactly? is it V1 V3 pair or V3V1 pair?

Neutron scattering factors

Check special functions are ok

sqrt is not complex and log is non zero. In case this happens, report error.

Accept data in cartesian coordinates

Allow colon in laue group definition

Namely, allow -3m:H as alias for -3mH. Also 2/m:b instead of 2/mb

Variant should fail, if sum of the chemical units does not equal to one.

Normalize calculated intensities

Should do it like in say shelx. Once the scale factor is known, divide experimental data by scale and save to experiment.h5 or create "scaled_experiment.h5"

report direct and reciprocal metrics

a,b,c, angles in both spaces and metric tensors

Variant should check occupancies

Parallelization for direct algorithm