Sometimes (e.g. when designing Your laboratory or Your project budget) You meet with the difficult choice of deciding which program to buy for Your computations. Or maybe just use a freeware or some with academic licence?
This is not so easy question, because that depends on what are You going to do. There are four main pros (or cons) of any computational code:
Hence among the four main traits of QC software two or three are important for them – and these are the ones commercial institutions usually excel in (the performance and computational stability).
I checked performance for three programs (two academy-free and one commercial) for the following case:
Molecule: C34H32N2O4, mostly unsaturated, with no symmetry elements (72 atoms, 282 electrons, 210 degrees of freedom)
Computational model: B3LYP/6-31G(d,p), with B3LYP defined as in Gaussian and no solvent in the model (760 basis functions)
Computational tasks: geometry optimization and thermochemistry of the ground state.
The results are as following:
Here are the same data on pictures:
Looks impressive, doesn't it? It is this moment I realized why those commercial suites cost that much.
So, should we use only commercial software? Again, it depends.
Why are there usually more new methods in the freeware? The reason is simple: they don't need to spend their time and/or people on customer support and refining the robustness of the code of the ease of use. This is because these programs are written by people who understand what's going on for people who understand what's going on, hence both the authors and users has less urge to communicate.
The trade-off is, again, not that clear, because new methods may be computationally more effective even if the commonly used algorithms are less effective that in some commercial code which does not have this new method.
What about the customer service? Aren't there forums from which You can get the support? Well, there are, but, again. for the academic/freeware the community-based approach is invaluable. Commercial software may have some (2–4) people they are paying to reply to every message they get; for freeware, anyone who knows the answer will answer. When he or she has time. And if they know how to solve this particular problem which probably they haven't run into yet. Therefore, both cases have their advantages and disadvantages.
These are my thoughts, so feel free to disagree :-)
This is not so easy question, because that depends on what are You going to do. There are four main pros (or cons) of any computational code:
- performance
- availability of different methods
- computational stability
- ease of use.
- as soon as possible
- as easily as possible.
- the quality of results (except when they disagree with they experimental ones).
Hence among the four main traits of QC software two or three are important for them – and these are the ones commercial institutions usually excel in (the performance and computational stability).
I checked performance for three programs (two academy-free and one commercial) for the following case:
Molecule: C34H32N2O4, mostly unsaturated, with no symmetry elements (72 atoms, 282 electrons, 210 degrees of freedom)
Computational model: B3LYP/6-31G(d,p), with B3LYP defined as in Gaussian and no solvent in the model (760 basis functions)
Computational tasks: geometry optimization and thermochemistry of the ground state.
The results are as following:
| Geometry optimization time | Gibbs free energy time | |
| Commercial 1 | 1374 | 3209 |
| Academic 1 | 3815 | 143394 |
| Academic 2 | 36488 | — |
| Final SCF energy | Final Gibbs free energy | |
| Commercial 1 | -1725.348479 | -1724.833152 |
| Academic 1 | -1725.314384 | -1724.788763 |
| Academic 2 | -1725.314514 | — |
Here are the same data on pictures:
Looks impressive, doesn't it? It is this moment I realized why those commercial suites cost that much.
So, should we use only commercial software? Again, it depends.
- If You are doing many calculations routinely and either do not care about selecting the best method or want to use some of the community standards, then probably You should consider buying some commercial package.
- If You are interested in using new methods out-of-the-box, consider using academic freeware because usually they contain more new methods.
Why are there usually more new methods in the freeware? The reason is simple: they don't need to spend their time and/or people on customer support and refining the robustness of the code of the ease of use. This is because these programs are written by people who understand what's going on for people who understand what's going on, hence both the authors and users has less urge to communicate.
The trade-off is, again, not that clear, because new methods may be computationally more effective even if the commonly used algorithms are less effective that in some commercial code which does not have this new method.
What about the customer service? Aren't there forums from which You can get the support? Well, there are, but, again. for the academic/freeware the community-based approach is invaluable. Commercial software may have some (2–4) people they are paying to reply to every message they get; for freeware, anyone who knows the answer will answer. When he or she has time. And if they know how to solve this particular problem which probably they haven't run into yet. Therefore, both cases have their advantages and disadvantages.
These are my thoughts, so feel free to disagree :-)
