Access (14:00-21:00 CEST, July 25-14 August):

• Go to https://gather.town/app/V6RDCO6Ee6hWnt8D/adfocs2021.
• Confirm you are at least 13 years old (unless you aren't). Click on Sign in
• Gather now starts up
• Allow Gather to use your mic and camera.

Troubleshooting:

• Use Chrome or Firefox. Make sure no other applications (such as Skype, Teams, etc.) are using your mic and camera.
• Make sure Chrome or Firefox does not permanently block permission requests for mic and camera (see browser settings).
• Check https://gather.town/faq, and otherwise send an e-mail to support@gather.town or adfocs@mpi-inf.mpg.de.
  
  

User guide:

Everything is fairly self-explanatory once you are in Gather, but please have a look at this guide to improve your experience.

• At the start:
  • You are spawned into the Common Area. Click "x" on your keyboard to see the Map (near the map sign).
  • Use Ghostmode (hold down "g") to move through people if a room entrance is blocked.
• Lectures:
  • Guide your avatar into the Lecture Hall. Please don't stand in the doorway!
  • Press "x" on your keyboard to get the Zoom link.
  • Leave the Zoom meeting after the lecture! Upon re-entering Gather, again give access to your camera and mic feeds if needed.

Technical information:

• Gather is supported on any desktop or laptop (no tablets and phones), and any OS (Windows, Mac OS, Linux). Camera and microphone are required.
• Only supported in Chrome and Firefox.
• Make sure your browser is up to date.*
• Using a VPN is not recommended (which also applies to Zoom).*
• Make sure "x" is not a shortcut key of a browser plug-in (disable those if needed).

This year the event will take place over the span of three weeks. We will start with a primer week with the goal of learning the basics of convex optimization. This is meant to cover the necessary background for the talks given by the invited speakers, which start in the second week.

Schedule

Lecture Topics

The topics of the lectures are the following:

Primer

• Lecture I: Math background. Karl Bringmann.
• Lecture II: Introduction to Convex Optimization. Vasileios Nakos.
• Lecture III: Gradient Descent. Alejandro Cassis.
• Lecture IV: Mirror Descent. Kurt Mehlhorn.
• Lecture V: Spectral Graph Theory. Andreas Karrenbauer.

Alina Ene

• Lectures I+II: Adaptive gradient descent for convex minimization
• Lecture III: Adaptive mirror prox algorithms for variational inequalities

Rasmus Kyng

• Lectures I+II: Laplacian solvers
• Lecture III: p-norm regression with acceleration

Aaron Sidford

• Lectures I+II: First-order methods for Undirected Graphs
• Lecture III: Interior point methods for Directed Graphs

The registration for the lectures is free and open for everyone. If you want to participate in the exercise sessions, you will need to provide either a motivation letter or a recommendation letter from your advisor. If you choose to do this, please mark the check-box in the registration link! We will contact you to provide the letter at a suitable time.

Primer Week

• Lecture 1: Math background. Slides. Video. [K] 2 [V] 1,2,3
• Lecture 2: Introduction to Convex Optimization. Slides. Video [B] 1 [V] 4,5
• Lecture 3: Gradient Descent. Slides. Video [B] 3 [K] 3 [V] 6
• Lecture 4: Mirror Descent. Slides. Video [B] 4 [V] 7
• Lecture 5: Spectral Graph Theory. Slides. Video [K] 1,4,7

Reading Material

• [B] Convex Optimization: Algorithms and Complexity: monograph by Sébastien Bubeck
• [K] Advanced Graph Algorithms and Optimization: lecture notes by Rasmus Kyng and Maximilian Probst Gutenberg
• [V] Algorithms for Convex Optimization: book by Nisheeth K. Vishnoi

Alina Ene

Slides

• Lecture 1: Adaptive Gradient Descent
• Lecture 2: Accelerated Gradient Descent
• Lecture 3: Variational Inequalities

Videos

• Lecture 1
• Lecture 2
• Lecture 3

Exercises

• Exercises 1 and 2
• Solutions

Further material

• Lecture Notes

Rasmus Kyng

Slides

• Lecture 1: Graph and Matrix approximation via sampling, calculations during the lecture
• Lecture 2: Solving Laplacian Linear Equations, calculations during the lecture
• Lecture 3: p-norm Regression with Acceleration

Videos

• Lecture 1
• Lecture 2
• Lecture 3

Exercises

• Exercises 1
• Exercises 2

Aaron Sidford

Slides

• Lecture 1: Box-Simplex Games Part I
• Lecture 2: Box-Simplex Games Part II

Videos

• Lecture 1
• Lecture 2
• Lecture 3

Exercises

• Exercises 1
• Exercises 2

Websites of previous ADFOCS summer schools.

• ADFOCS 21 (2020)
• ADFOCS 20 (2019)
• ADFOCS 19 (2018)
• ADFOCS 18 (2017)
• ADFOCS 17 (2016)
• ADFOCS 16 (2015)
• ADFOCS 15 (2014)
• ADFOCS 14 (2013)
• ADFOCS 13 (2012)
• ADFOCS 12 (2011)
• ADFOCS 11 (2010)
• ADFOCS 10 (2009)
• ADFOCS 9 (2008)
• ADFOCS 8 (2007)
• ADFOCS 7 (2006)
• ADFOCS 6 (2005)
• ADFOCS 5 (2004)
• ADFOCS 4 (2003)
• ADFOCS 3 (2002)
• ADFOCS 2 (2001)
• ADFOCS 1 (2000)

ADFOCS supports the recommendations of the SafeTOC report to combat harassment and discrimination in the Theory of Computing community.

By participating in the ADFOCS program, you agree to not exercise any demeaning, discriminatory, or harassing behavior and speech. More precisely, we will adhere to the guidelines as recommended by the ACM/STOC code of conduct.

If at anytime you feel unsafe, intimidated or harrassed during ADFOCS, please contact the organizers immediately and we will start an (anonymous) investigation.

The organizers reserve the right to remove any participant from the ADFOCS summer school.