- PCMSolver is now exported as a proper CMake target thanks to the work of @loriab.
- The Python tools distributed with the library are Python 2 and Python 3 compatible.
- Catch and Eigen have been updated to their latest versions.
The bug introduced in the generation of the static library
libpcm.a is fixed in this minor release.
This commit fixes issues #34, #35 and #36 (on GitHub) and #60, #61 and #62 (on GitLab)
The fix consisted in subscribing all subclasses to the factory in a top-level convenience header that gets included into the
Namespaces have finally been introduced into the project and some core abstract base classes have been renamed to signal their ABC status. Please see the change log for detailed information.
This release contains a number of fixes to the internal infrastructure. The Autocmake system was updated to its latest version. Some internal functions have been revamped. The Green's functions, boundary integral operators and solvers code have been extensively rewritten to simplify their structure.
The output flushing from library to host was managed by a function called
host_writer with signature:
void host_writer(const char * message, int message_length)
This has been changed with the current version without preserving backwards compatibility
From this version onwards, the output flushing function will have to be passed as a function pointer from the host to the library via the
pcmsolver_new function. Moreover, the expected signature for the passed function only needs one argument:
typedef void (*HostWriter)(const char * message)
A new radii set derived from Allinger's MM3 parameter set was added. This set of radii is the one used by default in ADF for the COSMO implementation. They are obtained from the data reported in the original paper divided by 1.2
The previous release had introduced a bug in the definition of the polarization weights calculated by the
IEFSolver object. This was problematic especially in response calculations and was not caught by the library own test suite.
A number of small bugs were fixed:
- Initialization of the explicit list of spheres when executing the standalone
- Memory leak in
PEDRAwhen a large number of spheres is given.
- A better log from the
CPCMSolverboth in initialization and ASC computation.
IEFSolverboth in initialization and ASC computation.
A number of small modifications, both internal and at API level.
API modifications are non-breaking. Notice that minimum required version for CMake has been pushed up to 2.8.10
Read the change log for exhaustive information. Consult also the list of issues