SE operates using the same drivers as DE (GString::growth_iters and GString::opt_iters), but has some differences that are particular to SE. Some of the differences have already been discussed in Delocalized Internal Coordinates and Double-Ended GSM. On this page we will further highlight the differences.
Unlike DE, which requires a reactant and a product, SE only requires the reactant and driving coordinates which are internal coordinates e.g. add/break bond. Therefore, SE can be used to search for pruducts that haven't been found yet. But because SE doesn't have a product structure, it can't form the internal coordinate tangent in the same way as DE. Instead, SE uses the GString::tangent_1b function to form the constrain vector using the driving coordinate . As the reaction proceeds,the tangent vector is scalled appropriately using atom-connectivity measurements. Another differences between SE and DE is that SE only optimizes the frontier node during the growth phase and doesn't undergo string reparametrization until the string is done growing. Nodes are set as active using GString::set_fsm_active and the variable GString::nnR keeps track of the reactant node.
Besides those differences, SSM must keep careful track of how the string is being grown because the number of nodes it uses is not constant like DE and it needs to detect when it has passed a TS to finish growing. In GString::growth_iters, the function GString::past_ts checks if the string has grown over the TS. GString::opt_iters checks if the TS node is the second to last node and will add another node to the string. The function GString::addCNode, will add a node if more space between the TS and product is needed, i.e. if the distance from the TS to product requires more than two nodes for the overall string spacing to remain even. When the product node is found, it undergoes optimization using ICoord::opt_b. After the string is fully grown it behaves identically to DE because the string is fully grown and the product is optimized.
P. M. Zimmerman "Single-ended transition state finding with the growing string method" J. Comput. Chem. 2015, 36, 601–611. DOI: 10.1002/jcc.23833
http://onlinelibrary.wiley.com/doi/10.1002/jcc.23833/abstract
1.6.1
