Version 1.1.0

Adds hydrogen bond constraints. Updates contents to SMIRNOFF 0.3 spec (spec change performed by OFF toolkit 0.4.0 reading smirnoff99Frosst-1.0.9, and then writing the ForceField object out in the 0.3 spec). Fixes several issues with FF hierarchy, SMIRKS patterns. This fixes some issues with coverage of certain areas of chemical space, and parameters which could never be utilized due to hierarchy issues. Full details available in openforcefield/openforcefield Issue 367.

Version 1.0.9

Addresses #89: Fixes torsion t56, where SMIRKS [!1:1]-[#7X4,#7X3:2]-[#6X4;r3:3]-[*:4] should be [!#1:1]-[#7X4,#7X3:2]-[#6X4;r3:3]-[*:4]. The first means "not an isotope with mass 1" (!1), but we intend for it to apply to all hydrogens, so it has been changed to !#1.

Version 1.0.8

In our paper about the SMIRNOFF format, we introduced a non-zero hydrogen parameter for polar hydrogens. Unfortunately, in the first introduction of these terms in our official force field the rmin_half and epsilon parameters were reversed. Here, we fix that small human error. This release also includes a fix for bond parameters between hydrogen and divalent carbons (issue #81) and fixed SMIRKS patterns for angle parameters around trivalent carbon in 5-membered rings (issue #84).

Version 1.0.7

Adds hydroxyl hydrogen LJ radii (in deviation with AMBER force fields, as per SMIRNOFF paper); also removes generic parameters.

Version 1.0.6

Added monovalent ion parameters (Joung/Cheatham) for TIP3P as default, explained in detail in the openforcefield repository issue #53. Added angle parameters for cyclobutyl groups as discussed in issue #65. Replaced R decorators with x to guarantee compatibility between OpenEye toolkits and RDKit SMIRKS parsing (#54).

Version 1.0.5

Version 1.0.5 adds an extensive set of new parameters (and modification of existing SMIRKS to improve coverage) to provide full coverage of our small filtered set from DrugBank, which we internally call MiniDrugBank. A full set of changes is detailed in openforcefield/smarty#232.

Version 1.0.4

Fixes a minor bug where a C-OS bond length overrode a C-OH bond length, and a major bug introduced by an earlier revision where a generic torsion had been replaced by a specialized one (to fix one bug) in a way which introduced holes in the coverage of generic torsions.

Version 1.0.3

Adds a missing bond parameter and fixes some torsion SMIRKS patterns. Now can reproduce parm@frosst energies on AlkEthOH chains except for parameters where differences were deliberate for generality.

Version 1.0.2

Fixes an incorrectly ordered generic affecting certain carbon-carbon bonds, as per #4 .

Version 1.0.1

Same as version 1.0, but triggering a Zenodo DOI.