greengroup / cantherm_old Goto Github PK

Some of the physical constants and conversion factors seem to be off from the values I am familiar with by a fraction of a percent. I have already adjusted R (universal gas constant) in the following edit: http://github.com/gmagoon/CanTherm/commit/2a657aea4882337e1c54c8fcf059b32e303afce4 , but various other values could be refined as well. Other examples include:
4180 and 4.18 should be 4184 and 4.184 (J/cal)
6.023e23 should be 6.022...e23 (Avogadro's number)
1.6605e-27 is used sometimes whereas 1.67e-27 is used elsewhere (amu to kg?)

CanTherm doesn't easily allow to account for changes in chirality (mirror symmetry) between reactants and TS which can cause errors of a factor of two in the computed rate constant. It could also contribute to entropy (S) being off by a factor of R*ln(2).
I think this may be addressed (in a somewhat hackish manner) in the version of CanTherm that I use as a submodule with RMG-Java.

CanTherm completes calculations in a matter of seconds, but I think there is potential for it to be done a lot faster. I'm guessing the limiting step is the diagonalization of the Hamiltonian matrix to get energy eigenvalues. Perhaps we could speed this up through cythonizing the code?

The second half of section 2.5, about the Clenshaw-Curtis method and Table 1 and the discrepancies, is not fully clear to me. Maybe I'm just not familiar enough with the literature, but if anyone is able to expand on the explanation that may help other people like me :-)