alt Hacker NewsI took some effort to change my research interest from computer vision to DFT calculation in quantum chemistry.
Honestly, I'm kind of frustrated now, too many work is close-source in this area. The research paper will tell you everything except how to reproduce this work in minimal effort, it's like they are hiding something.
They also using a `Origin` to plot and MS Word to write paper, which is also non-free licensed, and made them harder to collaborate and reproduce.
Replies
Very true that they're hiding things. I actually wrote some code (that strung together other people's code) to complete a simulation pipeline for non adiabatic molecular dynamics. I was tasked with writing documentation to teach the group but was instructed to not release it anywhere publicly because other groups would simply take the method and move faster since they had more money and compute.
This issue also bugged me for a while. It is more of cultural issue, and older the research group is, the less likely it is for research software to be open, in my experience.
In the area of deep learning based simulations, one good example of an open software is netket. The researcher their is pretty active in terms of github/gitlab/huggingface ecosystem.
I miss OriginPro in my undergrad when we had campus licenses for, before moving to matplotlib for data visualization. matplotlib is simply too disappointing for making publication quality figures. The most recently encountered problem is how to plot with a broken x-axis, which is one of the most basic need in physical science but requires a non-trivial amount of hacking to get with matplotlib.
Open source tool or not, I don't care at all as I get the science right. I have already enough frustration dealing with my samples, so I simply want the least frustration from the software I use to plot.
> The research paper will tell you everything except how to reproduce this work in minimal effort, it's like they are hiding something.
They are. I used to work in an adjacent field. Everyone was open about doing it - they're competing with others for grants, and worry that if they reveal the secret sauce, others will move faster than they can.
You can say you performed a DFT calculation to get the result, but anyone who's studied these types of simulations/calculations knows that it's highly nontrivial to implement, with lots of coding and numerical tricks involved. So it's extremely hard to reproduce if you don't have detailed access to the algorithms.