
Talks 2014  2022
2022
Differential Equations with R
R User Group, Heidelberg, March 2022
Several CRAN packages provide solvers for handling Differential Equations. We will  through examples  show how to solve ordinary differential equations (ODEs), systems of DEs, and DEs of second order, mostly with routines from the 'pracma' and 'deSolve' packages. If there is time, it will be demonstrated how the R user can apply Julia's extremely powerful 'DifferentialEquations.jl' through the 'diffeqr' R package.
ODEs with R
2021
Voting Procedures (and the 'vote' Package)
R User Group, Heidelberg, October 2021
Die Teilnehmer einer Abstimmung haben bestimmte Präferenzen bzgl. möglicher Alternativen. Darauf aufbauend gibt es verschiedene Abstimmungsverfahren (engl. voting procedures), einige davon sind auch in dem 'vote' Paket implementiert. Aber welche dieser Verfahren sind gerecht, fair oder weniger manipulierbar. Am Ende werden wir einige relativ unbekannte Verfahren und den "Arrowschen Unmöglichkeitssatz" diskutieren.
Abstimmungsverfahren / Voting Procedures
Regular Expressions in R
R User Group, Heidelberg, September 2021
Short reminder of regular expressions, their syntax and basic elements, functions in Base R and/or package 'stringr' for using regular expressions, some examples, and a Shiny application for testing regular expressions.
Regular Expressions
Keyvalue Stores
R User Group, Heidelberg, May 2021 (and Wiesbaden, February 2022)
The data structure of keyvalue stores or associative arrays (computer science) come under different names in several programming languages, such as 'dictionary' (Python, Julia), 'hash' (Perl, Ruby, JavaScript), 'map' (Java, Haskell), or 'table' (Lua), etc. We discuss some R packages that provide this functionality of keyvalue stores, like 'hash', 'filehash', 'fastmap', 'rredis' or 'RcppRedis', packages that provide keyvalue stores.
Keyvalue Stores
Linear Regression Trees
January 2021
The idea of "Linear (or local) Regression Trees" is to grow a tree, similar to a decision tree, in which every end node is associated with a linear regression for some or all of the variables in the data. The first idea and implementation was done by Ross Quinlan (of C4.5 fame) in his 'M5' program, for R users available in the 'Cubist' package. Other implementations are lmtree() in 'partykit' or M5P() in 'RWeka'; we also compare it with CART and RandomForest trees.
Linear Regression Trees
2020
Nichtlineare Regression in R
Wiesbaden R User Group, December 2020
'nls()' in Base R ist eine häufig verwendete Funktion zur "leastsquares" Schätzungen der Parameter eines nichtlinearen Modells; ihre Implementierung hat jedoch schwerwiegende Probleme und Einschränkungen. Wir werden Alternativen erörtern, die in RPaketen verfügbar sind, vor allem nlxb() im 'nlsr' Paket  und auch, wie Einschränkungen, z.B. Positivitätsbedingungen, für die gesuchten Parameter einbezogen werden können.
R Markdown notebook
R und das S4 ObjektSystem
Wiesbaden R User Group, September 2020
Beschreibt das S4 Klassensystem anhand einer Implementierung des Gauss'schen Gesetzes zur Fehlerfortpflanzung; die notwendigen arithmetischen Operationen werden als Operationen auf den S4 Objekten realisiert. Ein analoges Beispiel für "dual numbers" findet man hier:
Implementation of R Dual Numbers
The mlr3 package system
R User Group, Heidelberg, JulAug 2020
Duscussing the 'mlr3' packages with introduction, mlr3 learners (especially Random Forest), prediction and accuracy, resampling and crossvalidation, as well as hyperparameter tuning.
mlr3 examples Github repository
Subgroup Discovery mit R
Wiesbaden R User Group, June 2020
Subgroup Discovery (SD) ist eine Technik des beschreibenden und explorativen Data Minings. Es werden 'interessante' Teilmengen der Daten identifiziert, in Hinblick auf ein für die jeweilige Anwendung wichtiges Merkmal. SD ist von praktischer Relevanz für wissenschaftliche wie geschäftliche Datenanalysen und kann für Analysten wie Kunden überraschende Einblicke in die Daten liefern. Ein R Paket wird vorgestellt, das diese Methode implementiert.
Slides  the 'rsubgroup' package
2019
JuliaCall: Calling Julia from R
Github Page, 2017, Updated to Julia 1.0 in 2019
A short description of how to install JuliaCall and make use of Julia functions from R. Includes examples on plotting, for Optimization and Automatic Differentiation (AD), or calculating special functions not available in R.
Notes on JuliaCall
Random Numbers and Walks
R User Groups, February 2019
Presents and discusses several CRAN packages dealing with Random Number Generators (RNGs). Applies random numbers to the simulation of Random Walks and for calculating some of their properties. Use of JavaScript and/or Julia in simulation studies for faster runtimes is demonstrated.
Slides of the Talk
2018
Disciplined Convex Programming with CVXR
R User Group, Heidelberg, April 2018
CVXR is a new R package that provides a modeling language for Convex Programming, similar to CVX for MATLAB or CVXPY for Python. The user can formulate optimization problems following the mathematical equations describing it. The talk will give a short introduction to convex optimization, followed by several examples in regression with constraints and quadratic optimization, for instance the "smallest enclosing ball" problem.
Slides of the Talk
2017
Optimization With R  Tips and Tricks
Köln R User Group, September 2017
There are almost 100 packages listed on the CRAN optimization task view. We will discuss some of the more important packages in areas such as nonlinear optimization with and w/o constraints, leastsquares problems, nonsmooth and global optimization, and mixed integer programming. Tips and tricks will be provided for high precision or highdimensional problems, for equality constraints, for minimax problems, etc.
Slides of the talk
Can R Be Used for Numerical Mathematics?
Private Communication
R is a "software environment for statistical computing and graphics." But can R be used in courses on Numerical Mathematics, and is it possible to apply R to realworld numerical problems? Many R packages will be discussed that provide functionality in numerical analysis and linear algebra, and ways to integrate with other scientific environments such as Python, Julia, MATLAB/Octave, or even Mathematica.
Symbolic and Numeric Differentiation With R
Private Communication
Function differentiation is an important technique for optimization problems and differential equations applications, in engineering and many other areas. The talk will discuss packages for symbolic differentiation, automatic forward differentiation (implemented as S4 class), the complex step approach, central difference formulas, Richardson extrapolation for computing numerical Hessians, and higher order derivatives.
2016
Infos on new packages presented at UseR 2016
Meetup Session, August 2016
Some interesting packages presented at the UseR! 2016 at Stanford University will be described in one slide each, including profvis, covr, broom, future, feather. For data mining, packages like xgboost, ranger, or mxnet are useful new contributions. Links to Microsoft Channel 9 are included.
Short slide show (.Rmd)
Package V8: Computing With JavaScript in R
Meetup Session, April 2016
Package V8 provides an interface to Google's open source JavaScript engine and thus implements ECMAScript with compiled libraries for highperformance computing. We will present some examples of how to utilize JavaScript for quite efficient numerical computations in R.
Short slide show (.PDF)
R Training  Basic Introduction
ABB Forschungszentrum, Ladenburg, January 2016
Eine kurze Einführung in R Syntax, Einlesen von Daten, Dataframes, statistische Graphen, Programmierung mit R, Beispiele der Regression, etc.
2015
Nonlinear Optimization with R  An engineering example
Wiesbaden R Users Group, October 2015
A prediction task for gas distribution networks with storage tanks will be solved as a nonlinear optimization problem with constraints, by applying oprimization solvers available in R packages. Several different objective functions will be tested. Also shown is how this problem can be formulated in the AMPL modeling language and be sent to NEOS solvers of the COINOR project.
Jupyter notebook An Example in Constraint Optimization
Numerik mit MATLAB
DHBW Kurs, 20 Stunden, Mannheim, Herbst 2015
Siehe Unterlagen MatlabKurs
2014
Introduction to Julia for R Users
Köln R User Group, December 2014
Julia is a highperformance dynamic programming language for scientific computing, with a syntax that is familiar to users of other technical computing environments (Matlab, Python, R, etc.). It provides a sophisticated compiler, high performance with numerical accuracy, and extensive mathematical function libraries. Usercontributed packages are available for time series, statistics and machine learning, or operations research.
Notes from the Meetup in Cologne
Handout of the Talk
An introduction to the Julia language for scientific computing and its connection to Python with IJulia
Heidelberg Python Meetup Group, September 2014
Julia is a highlevel, highperformance dynamic programming language for technical computing, with syntax that is familiar to users of other technical computing environments. IJulia, a collaboration between the IPython and Julia communities, provides a powerful browserbased graphical notebook interface to Julia and enables easy use of Python within Julia.
Presentation: Julia for Python Users
Die StatistikUmgebung R: Einführung, übersicht, Anwendungen
VDI Arbeitskreis "Mess und Automatisierungstechnik"", Kassel, Februar 2014
Der Vortrag beinhaltet eine Einführung zu R, seiner Programmiersprache und seinen Entwicklungswerkzeugen. Beispiele vorhandener Methoden zu (nichtlinearen) Regression, zur Optimierung und zum Maschinellen Lernen (Data Mining). Der Vortrag richtet sich an alle, die an statistischer Datenverarbeitung und Visualisierung interessiert sind, bzw. sich über Fähigkeiten einer Open Source Software (OSS) in diesem Bereich informieren möchten.
Folien: Statistics Environment R (in English).

Andrew Gelman, Columbia University, and others
Statistical Modeling, Causal Inference, and Social Science
Last updated on June 15, 2022
"Andrew Gelman is a professor of statistics and political science and director of the Applied Statistics Center at Columbia University. His blogging platform is shared by other great authors, all with fantastic knowledge and insights."
And this blog is generally considered to be one of the best Statistics blogs on the Internet.
Click here to read the blog
Sabine Hossenfelder, Frankfurt Institute for Advanced Studies
BackReAction: Science, Physics, Philosophy of Science News
Last updated on June 11, 2022
"I am a [theoretical] physicist. People often wonder what a theoretical physicist does. You might not believe it, but most of the time I think. Sometimes, I scribble funny looking things with a pencil on a notebook. Processes like this usually involve lots of coffee and walking up and down the corridor. I am working on physics beyond the standard model, phenomenological quantum gravity, and modifications of general relativity."
Click here to read the blog
David Mumford, Brown University
Ruminations on cosmology and time
March 2021
"Like many people, I have been riveted for decades by the breathless bulletins from cosmologists describing the latest twist to their model of space and time at the largest possible scale. But recently, I have been looking more closely at these theories and, frankly, do not find them 100% convincing. Maybe it's all true but maybe in 50 years, it will all change. My biggest source of skepticism is its treatment of time: it feels as if in several ways it is trying to undo the vista that special and general relativity opened up for potential models of spacetime, that the'standard model' reverts to a very Newtonian perspective on which an extremely simple relativistic model has been foisted. Let me explain."
Click here to read the article
Sean Carroll's Video Series
The Biggest [Physics] Ideas in the Universe
Last updated on Sept. 1, 2020
A series of 48 informal videos talking about the biggest ideas we've come up with to describe our natural world. No prerequisites required!
Click here to see the videos
Alison Gopnik, Smithsonian Institute
Will A.I. Ever Be Smarter Than a FourYearOld?
February 2019
"Everyone's heard about the new advances in artificial intelligence, and especially machine learning. You've also heard utopian or apocalyptic predictions about what those advances mean. They have been taken to presage either immortality or the end of the world, and a lot has been written about both of those possibilities. But the most sophisticated AIs are still far from being able to solve problems that human fouryearolds accomplish with ease. In spite of the impressive name, artificial intelligence largely consists of techniques to detect statistical patterns in large data sets. There is much more to human learning."
Click here to read the article
