Messages

Published Apr. 28, 2022 5:13 AM

Dear all, we hope that you are all doing well. As we discussed during last week, we will for the rest of the semester just work on finalizing project 2. It means that we will not have any lectures from 215pm to 4pm.
Rather, these lectures will function as lab sessions together with our standard lab session from 415pm to 7.
Feel free to meet at 215pm  each Thursday. This will be effective from April 28. Our last session is May 12.
Thus, our last three weeks will be devoted to project work only from 215pm to 7pm. Same weblink as usual.
Best wishes to you all,
Morten and ?yvind

Published Apr. 8, 2022 4:57 AM

Hi all, it was great pleasure to discuss with you all today. Here's btw the link to the video from April 7 and the link to the handwritten notes:

  1. Video of lecture at https://youtu.be/EXOIsfPWxgk
  2. Handwritten notes at https://github.com/CompPhysics/ComputationalPhysics2/blob/gh-pages/doc/HandWrittenNotes/2022/NotesApril7.pdf

Since the majority of you will be working on Boltzmann machine (and hopefully neural networks as well), our lectures for the rest of the semester will focus on these topics. However, for the other variants of project 2, we will discuss these during the lab sessions.
There are some minor typos in the material on Boltzmann machines and this will be updated asap. Similarly, a final version with variants of project 2, will be made available at the end of the weekend April 9-10.

Else it remains to wish you all the very best for the coming break, use the opportunity to recharge your batteries...

Published Apr. 6, 2022 6:42 PM

Dear all, here's a short update on our further plans till the end of the semester. 
Since the majority of you plan to work on neural networks as ways to solve the same type of equations as in project 1 (replacing for example the correlated part of the wave function with either a neural network or a Boltzmann machine), we will use the rest of the semester for regular lectures on this topic, in addition to our normal labgroups. As of now we have two project groups based on the selection below
  o Neural Networks and Boltzmann Machines with material at http://compphysics.github.io/ComputationalPhysics2/doc/LectureNotes/_build/html/boltzmannmachines.html
  o Path integral Monte Carlo project

Since the majority of you plan to work on neural networks and Boltzmann machines, we will have lectures on this topic for the rest of the semester.
We meet each Thursday at 2.15pm for theory lectures and discussions till approx 4pm. The remaining...

Published Mar. 24, 2022 12:40 PM

This week there is only lab and the lectures on parallelization and optimization will be offered as a video only which will be uploaded later.

Published Mar. 18, 2022 1:04 AM

Dear all, today we tried to wrap our discussion of resampling techniques with an emphasis on the bootstrap and the blocking methods. The latter is more suited for large data sets like those we will generate with Monte carlo methods for quantum mechanical studies, with millions or more of samples. 
The video of the lecture (subtitles available in few hours) is at https://www.youtube.com/watch?v=F3ePjvn1aZo
and the handwritten notes are at https://github.com/CompPhysics/ComputationalPhysics2/blob/gh-pages/doc/HandWrittenNotes/2022/NotesMarch17.pdf

Next week we will discuss how to speed up our codes via various optimization tools, vectorization and parallelization. The lecture next week ends our discussions on the necessary ingredients of project 1. 
The week thereafter we will try to wrap up and finalize project 1.

Best wishes to you all,
Morten and ?yvind

Published Mar. 9, 2022 11:22 PM

Dear all, just a short message this time. Tomorrow we will have an in-person lecture for those of you who can attend. We will also run the lecture in a hybrid mode for those of you who cannot be at the university. Note also that we have been assigned the lecture room F?397 for Thursday March 10. 
Tomorrow we will discuss how to perform a statistical analysis and obtain a reliable estimate for the standard deviation. This will lead us to resampling methods like blocking and boostrap, two widely used resampling techniques.  The material for tomorrow is available at http://compphysics.github.io/ComputationalPhysics2/doc/pub/week8/html/week8-reveal.html

Cya soon and best wishes to you all,
Morten and ?yvind

Published Feb. 24, 2022 7:29 PM

Hi all, it was a great pleasure seeing you all again.
Here are the links to the video for the lecture this (subtitles coming later) and the handwritten notes
 -Video of Lecture at https://youtu.be/hqO-lSns9jo
- Handwritten notes at https://github.com/CompPhysics/ComputationalPhysics2/blob/gh-pages/doc/HandWrittenNotes/2022/NotesFebruary24.pdf

Next week we will wrap up the discussion of gradient methods, including a derivation of the steepest descent method, the conjugate gradient descent and semi-Newton-Raphson methods, in addition to a discussion of stochastic gradient descent methods.  The lecture slides for next week will be upgraded during the weekend,  but their location is at  http://compphysics.github.io/ComputationalPhysics2/doc/pub/week7/html/week7-reveal.html

After we have concluded the optimization part we will move to a statistical analysis of our results, in particular we will discuss resampling methods l...

Published Feb. 17, 2022 8:12 PM

The video for the lecture of February 17 is at 
https://youtu.be/5lIORG7Ppss
and subtitles will be available during the evening/night (youtube uses some 10++ hours to process all the odd things I said today?!)
Else, the handwritten notes are at
https://github.com/CompPhysics/ComputationalPhysics2/blob/gh-pages/doc/HandWrittenNotes/2022/NotesFebruary17.pdf


The aim next week is to discuss optimization and gradient methods.  As background material, take a look at the slides from the text on optimization mentioned earlier today. It contains much more than we will cover. But it gives a good overview, see https://web.stanford.edu/~boyd/cvxbook/bv_cvxslides.pdf

Till now we have covered
1) Implementing the Metropolis algorithm
2) Implementing analytical expressions for the local energy
3) started to implement importance sampling. The hope is to finalize the coding of this by the end of next week or begin week...

Published Feb. 17, 2022 5:19 AM

Hi all and welcome back to FYS4411/9411. Here's a short update on our plans for  this week with a short summary from last week.
1) Lab: we continue developing our code for project with the aim to add importance sampling
2) Lecture: Last week we discussed the overarching background needed for importance sampling and discussed the basic equations which are needed to code, such as the so-called quantum force and how to efficiently compute gradients, laplacians and ratios of wave functions.  The material is available through the slides at http://compphysics.github.io/ComputationalPhysics2/doc/pub/week4/html/week4-reveal.html
3)  Lecture: This week we continue our discussion on importance sampling with the material from last week (see again slides from last week). We will also start with our next topic that we need to encode, namely how to find the optimal parameters for a variational Monte Carlo simulation. The slides for week 7 at http://compphysics...

Published Feb. 4, 2022 5:16 PM

Hi all and best wishes for the weekend. This week we discussed the basics of the Metropolis algorithm and Markov Chain Monte Carlo approaches. In the lectures next week we will discuss in more detail how we do importance sampling, in particular where does the quantum force come from!
This will lead us to some elements of transport theory and the diffusion equation (which is intimately linked with Markov Chains).  This will also lead us to the equations we need to code, based on implementations from a modified diffusion equation called the Fokker-PLanck equation and the so-called Langevin equation. 
These are elements we need to discuss next Thursday. In more details, see the weekly slides at  for example
http://compphysics.github.io/ComputationalPhysics2/doc/pub/week4/html/week4-reveal.html
we will cover

The plans for this week  are:
o First we want to give you an overarching  view on the equations we need to progr...

Published Feb. 3, 2022 4:15 AM

Hi all, we hope this week started the best possible way and welcome back to FYS4411/9411.
The lecture material this week is to discuss central elements (and how to implement them) of the Monte Carlo machinery. 
We will (for some of you it is a repetition from previous courses)  discuss the mathematics of
- Markov Chain Monte Carlo
- Metropolis-Hastings sampling and Importance Sampling (and for importance sampling the Fokker-Planck and Langevin equations).
With this background we will discuss how to implement these elements in a code. We will most likely use next week as well to discuss these topics.
In addition to the weekly lecture slides at for example http://compphysics.github.io/ComputationalPhysics2/doc/pub/week3/html/._week3-bs001.html, there is also additional material from FYS3150/4150 on Markov chains at http://compphysics.github.io/ComputationalPhysics/doc/pub/rw/html/rw-bs.html
 Look forward to see you all again and...

Published Jan. 27, 2022 5:42 PM

Hi all, it was good seeing you all again.
Here's the link to the video from today's lecture. Subtitles will be available most likely within 10-20 hours from now (YouTube uses some time to process the file).
Video at :  https://youtu.be/aR0H5mp5fVc
Handwritten notes at https://github.com/CompPhysics/ComputationalPhysics2/blob/gh-pages/doc/HandWrittenNotes/2022/NotesJanuary27.pdf

Best wishes to you all,
Morten

Published Jan. 20, 2022 12:57 AM

Dear All, first, happy new year and best wishes for 2022 and the new semester.
We hope this semester has started the best possible way.
Due to covid-19 regulations the first two weeks will be taught (labs and lectures)  in a digital mode.
From week 5, we will have both online and in-person laboratories (room F?434, Department of Physics, University of Oslo).  Lectures will however always be in a digital mode since one of us (Morten) is in the USA in the period January-June. We may schedule an intensive week for the week of March 7-11, with in-person lectures. All lectures will recorded, with subtitles, and posted online after the lectures. There will also be additional videos and there is an extensive learning material at the GitHub address of the course., see 
https://github.com/CompPhysics/ComputationalPhysics2. and 
http://compphysics.github.io/ComputationalPhysics2/doc/web/course for an outline of the weekly teaching material....

Published Jan. 9, 2022 4:48 PM

Dear all, we hope you have had an excellent winter break and are ready for gaining new and exciting knowledge.

Our first lecture is Thursday January 20, at 215pm. We will have weekly lectures (except for a  break April 11-15) 215pm-4pm every Thursday till May 12.

Depending on the capacity of the room, we will run these lectures in person. These lectures area also available via zoom and will be recorded and uploaded after the lectures. The first week is however fully digital.

The lectures are followed by an in-person and digital lab. The lab starts at 415pm and ends at 7pm.  The first lab is January 20 and the last lab is May 12. 

All teaching material is available through the github address of the course, see https://github.com/CompPhysics/ComputationalPhysics2

 

Welcome to a new semester.

Morten and ?yvind.