Warning: This page uses Cascading Style Sheets. We recommend you upgrade to a current version of your favourite browser.

cposs logo

Control and Prediction of the Organic Solid State

CPOSS Home

Project Background

Project publications

Other project publicity

DMACRYS

The Computational Database

The Experimental Database

Personnel

PDRA & PhD Opportunities

UCL Home

UCL Chemistry Home

 

System Requirements and Compiler issues

Please check our recommendations for the computer system and compilers.

Known limitations of programs

  1. NEIGHCRYS copes with only some Z'<1 input files.
  2. Automated symmetry reduction is not currently possible with anisotropic repulsion potentials, because of the need to generate the atoms typing and local axes for the new independent molecule in the asymmetric unit cell.
  3. Only one Tang-Toennies damping parameter can be specified for the entire molecule, whereas more recent research (A J Misquitta, 2013, J Chem Theory Comput, 9, 5313) suggests that induction damping parameters may be more accurately defined by atomic type.
  4. Linear molecules cannot be dealt with by NEIGHCRYS.
  5. In the case of symmetry reduction, the output energy is calculated per formula unit of the original cell, so as to keep it directly comparable with non-symmetry-reduced structures. Z in the output fort.16 file is crystallographically incorrect, but will allow repeated calculations to keep directly comparable energies.
  6. The programs run very slowly for larger molecules, and some values defaults are too small. We are working on a new release that will be setup to run with larger (1000 atoms) molecules, which can still be used for smaller molecules, albeit some functions may need to be turned off for smoother running.

Current release versions

DMACRYS version 2.0.8 and NEIGHCRYS version 2.0.8

These versions of the codes contain the following improvements on previous versions.

  1. All 2.x.x versions of the codes are not compatible with 1.x.x versions.
  2. Addition of splines to smooth off potential at cutoff distances, and hence give better numerical derivatives with moderate cutoffs
  3. Inclusion of induction energy using distributed dipolar polarisability models
  4. Include (damped) C6 C8 and C10 isotropic dispersion

Reported bugs

  1. Although the code should not allow zero to appear as a negative number, this still seems to be occuring with some compilers. This leads to differences in the NEIGHCRYS generated *.dmain, and subsequent differences in the *.dmaout files and fort.12 output in the case of properties calculations. This causes the test suite to appear to fail on tests 11, 12 and/or 13. Detailed analysis of the fort.12 (run diff on the file provided and the user-generated output) will only show changes in the signs of some zeros.
  2. The DMACRYS output fort.16 has a ZERR line reporting that the numbers are only accurate to 1 decimal place ("0.0" accuracy in each parameter), which was intended to flag up that this is a computer-generated structure. If this input is used in some crystallographic programs that use ZERR, for example the Mercury conversion of the .res file to .cif format, it leads to inaccurate cell parameters rounded to this accuracy. This can be avoided by removing all the 0.0 error estimates.  In the next release, this will be the default output.
  3. When NEIGHCRYS is used to reduce symmetry, the output dmacrys.new.axis has the inverted molecules included. This causes some problems with running NEIGHCRYS on the subsequently reduced symmetry structure. These inverted molecule definitions will be removed from this file in the next version of NEIGHCRYS.
  4. When DMACRYS is run without symmetry (SYMM is removed and information in fort.20 is not used), the resulting fort.16 is missing information on atom types (in the SYMM and atom lines).
  5. When DMACRYS is minimizing a cell which is a long needle or a thin plate, the program assumes that you will always need to sum more than 1 reciprocal lattice vector in each direction, but for plates or needles it sometimes works out a value of 1. If you encounter this problem, please email Louise and registered users will be sent a new version of the program.
  6. In the development version of the program, 2.2.1.0, the splines are incorrectly summed over the charge-dipole and dipole-dipole terms.  This can lead to incorrect imaginary phonon frequencies.

Ongoing developments

  1. NEIGHCRYS 2.1.0 and DMACRYS 2.1.0 have been changed to reduce time spent in repeated setting up of intramolecular connectivity data for very large molecules. (This was implemented incorrectly in 2.1.0, with the scaling from lattice units to Angstroms not carried out, but is fixed in 2.2.0.)
  2. DMACRYS 2.2.0 writes out the symmetric component of the hessian, and can read this in at the start of subsequent runs to reduce time.

Bugs in early versions of 2.0.8 that appear to have been eradicated from final 2.0.8 version

  1. NEIGHCRYS outputs the new punch file with only 2 characters allowed for the atoms, which limits the number of atoms in independant molecules to 99.
  2. DMACRYS outputs atom numbers incorrectly in fort.16 if there are more than 100 of a single species.
  3. NEIGHCRYS has a problem with R-3c structures.
  4. NEIGHCRYS does not transfer the charge of a spherical ion from the dma to the .dmain file.
  5. NEIGHCRYS reads in bromine as boron in fdat files (related to issue with 1.0.8?)

Please email DMACRYS Support if there is anything you want to add to this page. If you spot a new problem, let her know, or if you have anything further to add to an existing problem. Also, if you have solved the problem, email DMACRYS Support to remove it from the list.


Previous versions

Known problems with DMAREL

  1. d packing types from MOLPAK have density out by a factor of 2.

Known problems with previous release version (1.11) of DMACRYS

  1. There are some problems with symmetry reduction.
  2. DMACRYS does not work for spherical ions (eg Cl-).

Known problems with previous release version (1.10) of NEIGHCRYS

  1. It standardises all OH bond lengths to 0.976 Å (neutron value from Allen, F. H.; Kennard, O.; Watson, D. G.; Brammer, L.; Orpen, A. G.; Taylor, R. Tables of bond lengths determined by x-ray and neutron diffraction. Part 1. Bond lengths in Organic compounds. Journal of the Chemical Society-Perkin Transactions 2 1987, (12), S1-S19.), which is not appropriate for carboxylic acids or water.
  2. An extra line, reporting the number of times the primitive cell appears in the unit cell, was introduced, for use with DMAflex2
  3. NEIGHCRYS assigns S as silicon for .fdat files, although .res files it is correctly assigned as sulfur.

Enter secure pages (For project members only - password required)

© UCL Chemistry Department 2003. This page was last updated on 23 March, 2017. If you have any problems with this page please email the WebMaster