Computational Chemistry 0.1 - Introduction

Short lecture introducing the computational chemistry.
Computational chemistry is the use of computers to solve the equations of a theory or model for the properties of a chemical system. Thanks to the widespread use and rapid growth of computing power, simulation is becoming a third pillar of science, along with theory and experiment. Using theory, we develop models: simplified representations of physical systems. We test the accuracy of these models in experiment, and use simulation to help refine this feedback loop. Anything that can be measured can be simulated, including properties related to energetics, structures, and spectra of chemical systems.
This course currently includes chapters on molecular geometry, molecular mechanics, and molecular dynamics / Monte Carlo methods. Future chapters include various electronic structure theory methods, including Hartree-Fock (HF) theory, Configuration Interaction (CI), basis sets, Moller-Plesset (MP2) perturbation theory, Coupled Cluster (CC) theory, and Density Functional Theory (DFT).
Notes Slide: i.imgur.com/Wwmw75r.png
-- About TMP Chem --
All TMP Chem content is free for everyone, everywhere, and created independently by Trent Parker.
Email: [e-mail]
-- Video Links --
Course Playlist: • Computational Chemistry
Other Courses: • PChem Course Intros
Channel Info: • About TMP Chem
-- Social Links --
Facebook: / tmpchem
Twitter: / tmpchem
LinkedIn: / tmpchem
Imgur: tmpchem.imgur.com
GitHub: www.github.com/tmpchem
-- Equipment --
Microphone: Blue Yeti USB Microphone
Drawing Tablet: Wacom Intuos Pen and Touch Small
Drawing Program: Autodesk Sketchbook Express
Screen Capture: Corel Visual Studio Pro X8

[@zohrehamanollahi8818]

You are so kind, generous and the best. I really really thank you for letting us to get informed this way with all of these valuable things. I don't know how to appreciate your tries on making these helpful videos.

[@asanvigyan9275]

Such a lucid introduction. Course designed in small learnable bits. Awesome! Thank you so much for the course.

[@TMPChem]

Thanks Asan. More in the works, but it's certainly been slow going. Appreciate the support.

[@Chitaaaaaa]

I start my PhD in Theoretical Chemistry this Fall. I feel that completion of your Quantum Chemistry and Spectroscopy and Computational/Theoretical Chemistry playlists will set me up greatly for the future.

[@TMPChem]

Hi Danny. I agree, that those can form a good foundation, especially if completed prior to the beginning of the fall coursework.

[@dutonic]

Starting this playlist is like watching shadow of israphel. You know it’s never gonna be finished 😭, but you can’t stop yourself anyway

[@shrodingereli2534]

very nice TEMP chem >>> awesome .so simple and clear thanks so much and we hope more in chemistry

[@matteoorlandi2469]

Will you upload new videos about other topics like DFT,CC and post HF methods, perturbation theory or statistical mechanics? I'm studying physical chemistry and your help has been very usefull. Thanks TMP Chem

[@meryemfat]

thank you

[@liyajesus5388]

thank you and i want know about computational material science and also which is one more important or mainly used DFT, post HF Ab initio,..........

[@3Dgirlteluguvlogs]

It is very useful to me sir thank you

[@somethingfishy1088]

Thank you Trent Parker :)

[@senseidele]

I’m required to do some computational chemistry to complete my doctoral thesis. Does anyone know of any software available for use? Please help!

[@matteobruschi4886]

Hi, which free computational program do you suggest me for wave function and DFT calculations? I've tried ORCA for example

[@TMPChem]

Hi Matteo. This depends on whether you are doing small molecule / gas phase calculations, or condensed phase / periodic simulations. For the former, I'd suggest looking into PSI4, as the density fitted DFT should be suitable for your needs up to the low hundreds of atoms for standard functionals (full disclosure: this opinion is completely biased as I worked for / with the primary developers of PSI4 for years). For the latter, I'm not sure as I haven't done enough periodic simulations to have a strong / informed opinion. Of the Wikipedia list which are free and include DFT, I've really only interacted with people who use and like Quantum ESPRESSO.

[@pradiptadey2338]

Awesome

[@ahmedsalafap5898]

Thank you so much

[@TMPChem]

Thanks for watching, Ahmed.

[@gaoyongliu1783]

very useful for beginners

[@TMPChem]

Good. That's the intent. Have fun as you continue your learning.

[@emonreturns7811]

Why MM.py script not working. Any of the programme is not running in that MOLECUlAR_MECHANICS folder. Showing value error. . TMP Chem, Kindly Help.

[@ramniwasprajapat3382]

Thank you sir

[@TMPChem]

Thanks for watching, Ram.

[@Abcs-o7j]

Please make some lectures on machine learning in chemistry

[@foreverofthestars4718]

Out of curiosity, what sort of computer hardware would you recommend to have for modeling of systems? What would suggest as the bare minimum, a good midrange system, and high end stuff? Are mobile workstations worth it?

[@TMPChem]

It all depends on what type of modeling you do, on what kind of systems, and what your goals are. If you want to do large, cutting-edge computations and complex systems, that likely requires access to institutional resources such as national lab supercomputers, or dedicated large clusters at research universities. If you want to do something quick and straightforward for a class project, that might be doable on a laptop. There are projects whose ambition are everything in between those, with hardware requirements everywhere in between as well. Often, if you aim to do publishable research quality computational work, it will require either a high-end workstation, or an account and computing hours on a larger cluster. With state-of-the-art software, in my grad school research group we were often able to do useful electronic structure theory computations on our personal workstations which cost around ~$2k (6-core latest generation i7 processor, 64 GB ram, 9 TB disk). Many kinds of computations don't require such high-end single workstation specs, but require an enormous number of CPU hours, which can only be achieved by running on dozens or hundreds of machines in a cluster. Others might not be possible on a single chip, but only with hundreds of cores with highly parallel specialized software. Again, it all depends.

[@foreverofthestars4718]

@@TMPChem Hello TMP Chem, thanks for getting back to me so quickly. I have a follow up question with a bit more detail if you don't mind. I also sent the question to you on FB, though I'm not sure which you check so I'll ask it here as well. Also if it's answered here then someone else may benefit from it as well. Anyway: I am currently starting graduate school and my research will be focused on first principle simulations of atomic and molecular structures for the purposes of use in solar energy technology. While doing this research I will have access to my universities supercomputer for particularly demanding calculations. However, for smaller simulations for practice and preparation purposes I thought it may be nice to get my own workstation that can handle the more basic stuff. That way I won't need to constantly request access to the supercomputer, schedule usage times, take away from the usage time of other researchers at the university, and just generally have an easier time of it myself.... So again on that note, I was thinking of building my own workstation. I already have in mind a general idea of what I want to build. A desktop with an AMD threadripper (probably a 3960X, 24 core, to start, but hopefully upgrade to 3990X, 64 core, in the future when money isn't as tight) and 64GB of RAM (again with room to upgrade). I feel like that's probably the best bet for a consumer/private level workstation, that is still pretty beefy. However, I have yet to workout what kind of GPU's would be best. I am not sure how GPU architecture is typically implemented in these kind of atomic and molecular simulations (Also if it is relevant, I plan to mostly work with VASP and Quantum Espresso), and thus how important GPU's are for them. I've been researching these things for a while now and have read some conflicting information. A lot of sources suggest that Nvidia's Quadro line is what I should be using since they offer double precision floating points and ECC memory, but I have also read that Nvidia's new 3090 is not only adequate, but better. In either case, I ultimately plan to to connect multiple GPU's together via SLI to increase performance. So with all that said, I was wondering what your thoughts on the matter would be? Is Quadro's ECC support and double precision better or is something like a GTX 3090 better for these kinds of simulations?

[@arminbehdad6781]

Hello and thank you for these very good teaching and video I have a question. In statistical mechanics, we consider ensembles rather than a single system. In MD we also say ensemble but I can't understand the MD's ensemble?, and how average ensemble over trajectory.?is an infinite image of PBC are our ensemble?

[@TMPChem]

Hi Jhon. What let's us do this is the ergodic hypothesis, which is the central theorem of statistical mechanics. The ergodic hypothesis states that all allowable physical states are equally likely to be observed. This is equivalent to saying that average over a large ensemble of structures is the same as averaging over a trajectory covering a sufficiently large period of time. Since all states are equally likely, if we average over an infinite period of time, the number of times a state is observed is directly proportional to its probability in the ensemble.

[@arminbehdad6781]

Thank you so much. So, if we do our MD in NVE ensemble our energy is conserved and ergodic hypothesis is true. if we do our simulation in NPT ensemble still this ergodic hypothesis can be applied?still, energy is conserved?

[@TMPChem]

NPT ensembles have to go about maintaining canonical (or grand canonical) ensembles in special ways which are mathematically involved and theoretically nuanced, but there are ways to do it. There energy is not the conserved quantity, but there are other special conserved quantities to monitor depending on the type of ensemble. It's usually not T or P, as those values need to fluctuate around there specified values, but will tend toward their average over a large period of time.

[@arminbehdad6781]

Thank you so much for your help

[@aadoobhai9839]

Nice work. Any textbooks for reference ?

[@TMPChem]

Yes. kzbin.info/www/bejne/fqqTh4Gka8qqi7c

[@desgbtv5733]

Nice

[@TMPChem]

Nice.

[@hkharis07]

How can access the next lecture?

[@TMPChem]

Hi HK. Typically the next lecture will appear in the recommended section. If not, you can check the video number. This introduction video is 0.1, with the next videos being 1.1, 1.2, 1.3, ... all the way to 2.1, 2.2, ... The complete playlist can be found on the channel page as well, kzbin.info. kzbin.info/aero/PLm8ZSArAXicIWTHEWgHG5mDr8YbrdcN1K