Richard Cooper

Nov 132017
 

The CRYSTALS v6720 installer is now available.

This update uses a new 64-bit Fortran compiler, fixes a number of bugs and eliminates a few mysterious crashes. Please report problems. The version number now refers to a specific snapshot of the source code enabling us to better identify and fix bugs.

Key changes between v14.6237 and 14.6720
64-bit compiler, bundle mkl math libraries – improved stability. Updated icon.
Improved SHELX import for atoms on special positions, SADI, DFIX and MPLA restraints, long form factors.
Spot is_nan in plot data.
New set of displacement parameter restraints, including TLS.
Better #CHECK output for restraints – including individual leverages.
Fix misreport of symmetry operators in CIF for C2/m
Dynamic threshold of Q peak height in Fourier peak viewer script.
Allow paths containing single quotes.
Add greek symbols to text and graphs for more concise display.
Leverages calculation via #sfls, refine punch=leverages, end
Changes related to getting consistent results for Thmax in CIF
Fixed listbox operation clicking in list now generates appropriate feedback for scripts.
Least squares matrix condition number output now included for problem refinement diagnosis.
Avoid almost infinite loop looking for bonds if atom coords have gone bad. Side effect – bonds outside +/- 50 unit cells of the origin won’t be found.
New choice of matrix invertors for SFLS. Default can be set in L23.
Fix default hkl format when re-reading ‘from-cif.hkl’ files (3F4.0,2F10.0).
Add a few more instrument/diffractometer types (synchrotron, neutron sources, etc.)
Modify platanom calculation: (i) use more d.p for improved precision; (ii) use c and d terms to correct data with all anomalous scattering effect removed as detailed in final version of ‘HUG and SQUEEZE’ manuscript.
Compute dispersion correction terms and photon interaction cross section for non-standard (=Mo,Cu) wavelength data. Alter xinlist3 to streamline experience.
Undo dialog: fix situation where current L5 is marked as error list.
Check for isotropic / unknown atoms before attempting to SPLIT in xsplits.ssc
Make L11 depend on L28, now if L28 filters change the VCV becomes invalid, require re-refinement. Avoids publishing CIF after omitting reflections without extra cycles.
Fix storage of converge checkbox state so it isn’t overridden when called from xwrite5.
Edlist3: revamp user interface to simplify. Added get/set mu values from L29 for each element.
Add Helium to properties file.
LIST 6 version incremented to 124. NEW DSC files will not be backwards compatible with older versions of CRYSTALS.
Better treatment of twins in Fourier maps.
Fix output of RIDE constraints (better refinement stability).
Reduce decimal places for angles with no esd (e.g. by symmetry constraint, or not refined) from 3 to 2 in CIF output. Avoids reporting e.g. 179.995 degress for exactly 180 degree angles.
Add internal vs external variance plot
Fix switch on of extinction from the Guide ‘Add extinction parameter’ dialog.

 

Oct 232017
 

Acta Crystallographica, 2017, C73, 845–853. [ doi:10.1107/S2053229617013304 ]

Using an approximate correction to the X-ray scattering from disordered, resonantly scattering regions of crystal structures we have developed and tested a procedure (HUG) to recover the absolute structure using conventional Flack x refi nement or other post-re finement determination methods.

Oct 102017
 

Lewis is working on a project to improve our understanding of modulation in molecular materials by studying analogues of Barluenga’s Reagent.

Away from the lab Lewis enjoys sports including football, rugby, and American football, as well as the odd game of FIFA.

He was inspired to join the group after attending the Structural Methods Options Course. His favourite space group is I-43d and he is equally happy programming in Fortran or Python.

Oct 082017
 

Gwenno is investigating methods for determination of flexible organic molecular structures using the atomic pair distribution function.

Around labs, Gwenno manages to fit in singing, acting, dancing and painting, and enjoys skiing.

 She still prefers programming in Python 2 over Python 3, and her favourite XC functional is LDA (who knew?).
Oct 082017
 

Oli Bar is investigating geometric parameters of metal organic frameworks using data analysis tools and machine learning.

To unwind after a hard day of Python algorithms he plays tennis and badminton, and enjoys swimming. He prefers the non-standard setting of space group 14, P21/n, and at the time of writing he is undecided on his favourite exchange correlation functional.

Oct 082017
 

Kiaora’s research project is to determine the structures of molecules which are liquids at room temperature or otherwise hard-to-crystallize using crystalline host frameworks and/or low temperature in-situ crystallization techniques.

While away from the lab Kiaora indulges in the incompatible sports of baking and swimming.

Her favourite space group is P21/c and she prefers the hydrogen bond over other types of intermolecular interaction.

Oct 082017
 

James is carrying out a research project using molecular dynamics to improve our models of disorder in crystal structures.

James’ favourite undergraduate course was Prof Ritchie’s Quantum Mechanics stuff, and his preferred exchange correlation functional is the classic PBE.

 

Sep 082017
 

The triennial congress of the International Union of Crystallography was held in Hyderabad over 8 days in August 2017.

During the meeting Richard Cooper presented recent work with Jerome Wicker:
Optimizing co-crystal  screens using a data-driven machine learning method

We also took an opportunity to present recent developments in the CRYSTALS software at the IUCr parallel programme Software Fayre:
Advanced restraints in CRYSTALS

Prior to the IUCr meeting, Richard Cooper was a tutor at the Crystallographic Computing School at IISc in Bangalore, organised by the IUCr Commission on Crystallographic Computing, and gave a lecture:
Recent advances in small molecule refinement

 

Photograph by @LouiseDawe

 

 

 

 

 

 

Jul 062017
 

CrystEngComm, 2017, 19, 5336 – 5340 [ doi:10.1039/C7CE00587C ]

A data-driven approach to predicting co-crystal formation reduces the number of experiments required to successfully produce new co-crystals. A machine learning algorithm trained on an in-house set of co-crystallization experiments results in a 2.6-fold enrichment of successful co-crystal formation in a ranked list of co-formers, using an unseen set of paracetamol test experiments.