The half baked bits at the moment is full scale automated catalyst discovery, and a bunch of the "glue" code that pulls the different simulations together seamlessly (molecular dynamics, dimer methods, nudged elastic band, retrosynthesis, automated bulk analysis) -- they all exist in various states of development (mostly done) but getting them all to interact totally seamlessly is tricky.
Also this is quite computationally intensive hobby, I am often limited by the amount of computing power I have - even though I run a small cluster in my house (384 cpu cores, 4 GPUs, 24TB NAS)
I am super excited though, just in the last week I have started to make novel, truely unique catalyst discoveries (that are matching lab validations), so after a year of experimenting "it's working.gif"
The half baked bits at the moment is full scale automated catalyst discovery, and a bunch of the "glue" code that pulls the different simulations together seamlessly (molecular dynamics, dimer methods, nudged elastic band, retrosynthesis, automated bulk analysis) -- they all exist in various states of development (mostly done) but getting them all to interact totally seamlessly is tricky.
Also this is quite computationally intensive hobby, I am often limited by the amount of computing power I have - even though I run a small cluster in my house (384 cpu cores, 4 GPUs, 24TB NAS)
I am super excited though, just in the last week I have started to make novel, truely unique catalyst discoveries (that are matching lab validations), so after a year of experimenting "it's working.gif"