Lecture ZZ - Final Exam Review (2020-04-28)

In this lecture, we review the format and possible content for the Spring 2020 final exam for SOS 212. The exam is comprehensive, covering systems dynamics modeling from causal loop diagrams (and systems thinking) up through stock-and-flow diagrams (and system dynamics models). Example models, such as the SIR model as Bass diffusion model, are briefly discussed. Students are encouraged to revisit information on chaos, stochastic modeling, and tipping points. Questions are asked and answered.

Lecture G1 - Randomness and Chaos (2020-04-16)

In this lecture, we introduce stochastic modeling as well as the topic of chaotic behavior in complex systems. The first half of this lecture focuses on stochastic modeling and the distinction between randomness and stochastic modeling, where "stochastic modeling" is the deliberate choice to use randomness in a model to simplify the modeling process. We then transition to discussing chaos (a particularly extreme sensitivity to initial conditions in models), which gives models an appearance of random behavior despite having no randomness within them. We close with an explanation of the so-called "butterfly effect" and an explanation of "strange attractors" frequently captured in the study of chaotic systems.

Lecture F3 - Morecroft (2015, Ch. 10) - Model Validity, Mental Models, Learning (2020-04-14)

In this lecture, we cover topics related to Chapter 10 of Morecroft (2015) on model validity, mental models, and learning. We revisit the general description of a model and discuss how models can range from analogue to metaphorical, with "illustrative models" balancing the two and providing plausible scaling with generalizable insights. The discussion guided by Morecroft (2015) is interleaved with classic arguments from Frank Keil about the shallows of explanation and illusions of depth. We discuss how formal models are "transitional objects" that allow for deepening the understanding of mental models as well as joining with others in collective cognition and cognitive division of labor. We close the discussion with examples of how to build confidence in mental models through verifying correct boundaries, structure, and calibrated parameter values. Significant confidence in a validated model comes from learning something new about the system being modeled.

Lecture F2 - Morecroft (2015, Ch. 9), Public Sector Applications of Strategic Modeling (2020-04-09)

In this lecture, we cover topics brought up by Morecroft (2015, Chapter 09) is his exploration of public sector applications of strategic modeling. We review the Forrester Urban Dynamics model and how it was synthesized from dynamic hypothesis (as a causal loop diagram) to multi-sector stock-and-flow diagram to a simulation that generated predictions that could be compared to real data (for both validation and scenario testing/planning). We then pivot back to the simple harvested fishery and use it to discuss the more general dynamical systems topic of "tipping points" and the visualization tool of "bifurcation diagrams" that help visualize things like hysteresis in systems (which are often associated with tipping points). The class ends after skipping over some content from the chapter about adding an endogenous investment model to the simple harvested fishery, and so that content is covered in a separate video to be (optionally) watched outside of class.

The separate lecture content on the endogenous investment model can be found at can be found at: https://youtu.be/uXdcHXq7eeQ

Lecture F2 video:

Extra module on fishery with endogenous investment model:

Lecture F1 - Morecroft (2015, Ch. 8), Industry Dynamics - Oil Price and the Global Oil Producers (2020-04-02)

In this lecture, we cover the major modeling topics from Chapter 08 of Morecroft (2010), which covers the development of a dynamic model of oil price that explicitly models global oil producers (including independent producers, OPEC, and the effect of Russia entering the market after the collapse of the USSR). The lecture marches through the selection of major sectors to model and then drills down into each sector to discuss how to structure stocks and flows that capture the dynamics of how variables relate to each other to explain price and production fluctuations over time.

Lecture E5 - Introduction to Assignment E5 and Review of Related Concepts (2020-03-26)

This video is primarily intended to introduce Assignment E5, a system dynamics modeling assignment meant to be an example of what a model for a final project might look like. It also reviews topics such as lookup tables, shadow/ghost variables, and the use of time functions to create seasonal patterns within models. Concrete examples of these concepts are shown within Vensim and Insight Maker.

Lecture E4 - Morecroft (2015, Ch. 6), Dynamics of Growth and Diffusion (2020-03-24)

This lecture covers content from Chapter 6 of Morecroft (2016) on models for the dynamics of growth and diffusion, with a major focus on the Bass model (with advertising) and its conceptual relationship to the SIR model of infectious disease dynamics from the previous lecture. We start with a way to re-conceptualize a simple model of growth in a fishery toward something that looks more like an S-I model (the first "half" of an SIR model) and use that to motivate the Bass model. We discuss the modeling of advertising in the Bass model (related to spontaneous infections of a population) and different ways to expand the Bass model to be more specific to different problems of interest (such as durable goods, goods with repeat purchases, etc.). We close with a real-world example of a modified Bass model being used for scenario planning of a low-cost airline being introduced into an existing market.

Lecture E3 - Epidemic Modeling (SIR models and friends) and Scenario Planning with Simulation (2020-03-19)

In this lecture, we demonstrate how to use System Dynamics Modeling in the special case of modeling disease spread (epidemic dynamics). We motivate the construction and use of the SIR (susceptible-infectious-recovered) model for describing the spread of disease through a population, and then we describe a simple quarantine-planning scenario exploration that demonstrates that quarantines are only effective at minimizing disease spread if they are implemented very early and otherwise are primarily used for "flattening the curve." We then close with a very brief introduction to related compartmental models (SI, SIS, SIRS, SEIR, SEIRS, MSIR, etc.).

Lecture E2 - Simulating Fixed and Smoothing Delays in Vensim and Insight Maker (2020-03-17)

In this lecture, we define two different types of delays that are frequently used in simulation – fixed delays and smoothing (or averaging) delays. We also show how these delays can be implemented in Vensim and Insight Maker, and we cover the basics of using "step responses" to determine the "time constant" for first-order averaging delays. The lecture ends with a few more advanced delay concepts, such as higher-order smoothing delays.

Lecture E1 - Units, Sliders, and Lookup-Table Converters in Vensim and Insight Maker [PARTIAL LECTURE] (2020-03-05)

Unfortunately, due to a technical issue, a recording of the first half of this lecture was lost. Consequently, this recording skips through that content and starts with the second half of the content. If this recorded was complete, it would first cover how to configure Vensim and Insight Maker to use units in order to find formulas problems that result in incorrectly calculated units. It would then cover the use of Sliders to more quickly change and experiment with model parameters. It would then close on the use of lookup tables (or "converters" in Insight Maker) to define relationships graphically (as opposed to with formulas).

There are examples built into the lecture notes, and there are also additional examples given within Vensim and Insight Maker live during the lecture.

Fortunately, this lecture's content is covered by three independent tutorial videos (which also include advanced topics not covered in the main lecture). Students are encouraged to view those videos as well:

Video: Interacting with Vensim: Embedded Plots and Sliders

Video: Units in Vensim - Unit Synonyms and Checking Units

Video: Units in Insight Maker - Unit Synonyms, Conversions, and Checking

Lecture D-E - Midterm Review (2020-02-25)

This lecture reviews all content in Units A, B, C, and D in SOS 212 as preparation for the midterm. These topics cover modeling fundamentals, causal loop diagrams, system archetypes, and basic stock-and-flow diagrams (as well as numerical integration of dynamical system models in general). Examples are given within the slides as well as separately on top of a sample midterm and midterm review problems distributed to the students before the lecture.

Link to Spring 2019 sample midterm with annotations made while in class:

https://www.dropbox.com/s/yjlzhjxooe9oqsu/SOS212-LectureDE-2020-02-25-Annotated_MidtermB_Examples.pdf?dl=0

Link to Midterm Review problems with annotations made while in class:

https://www.dropbox.com/s/9zasvdppu0f2d2g/SOS212-LectureDE-2020-02-25-Annotated_MidtermReview_Examples.pdf?dl=0

Lecture D4 - Morecroft (2010, Chapter 3), "Modelling Dynamic Systems" (2020-02-20)

This lecture is focused primarily on the content of Chapter 3 from Morecroft (2010) on "Modeling Dynamic Systems." We start with a very brief Insight Maker tutorial and then drill down into the main content of the chapter. We introduce stocks, flows, and coordinating networks of links between them. We also motivate how stock-and-flow diagrams when combined with computer simulation (or numerical integration) provides an alternative methodology for investigating dynamical systems that does not require the formal methods of calculus. We end with a basic dynamical systems model of a city allocating police to manage the illegal drug market and discuss basic verification and validation (V&V) with respect to the graphs that come out of the simulated model.

Lecture D3 - Stock-and-Flow Diagrams in Vensim and Insight Maker (2020-02-18)

In this video, we cover more detailed descriptions of "stocks", "flows", and "converters" as well as how they are shown in stock-and-flow diagrams and how to annotate them using reasoning from causal loop diagrams (CLDs). Connections to the underlying calculus concepts (derivatives and initial-value integration problems) are also made. We then conclude with a brief introduction of how to draw and simulate stock-and-flow diagrams in Vensim and Insight Maker.

Lecture D2 - Introduction to Numerical Simulation of Dynamical Systems, Part 2 (2020-02-13)

In this lecture, we continue our discussion of basic numerical methods (implemented in Excel) to simulate (deterministic) dynamical systems. We start with the bacterial example with only births (reproduction events) and add in deaths. We then discuss the effect of changing the time step ("dt"). We then conclude with an in-class exercise simulating the trajectory of a toilet filling up after a flush.

Lecture D1 - Introduction to Numerical Simulation of Dynamical Systems, Part 1 (2020-02-11)

In this lecture, we introduce basic numerical methods for simulating (or "integrating") dynamical systems models, like those specified by ordinary differential equations (ODE's) or System Dynamics Models formulated graphically as stock-and-flow diagrams. The lecture starts with a compound interest example (a model of how money grows in a "bank" -- with the bank balance playing the role of a stock and the interest generated over a year playing the role of a flow) and demonstrates how more general continuous dynamics (like the average population trajectory of bacterial growth) can be approximated like money growing in a bank via compound interest. All examples are done in a spreadsheet program, like Microsoft Excel, with hints of how the models will later be put into a formal System Dynamics Modeling tool like Vensim or Insight Maker in the form of stock-and-flow diagrams.

Lecture C2 - "Applying Systems Archetypes" from Kim and Lannon (1997) (2020-02-06)

In this lecture, which is based on "Applying Systems Archetypes" from Kim and Lannon (1997), we discuss four different ways that CLD systems archetypes can be used when analyzing systems, recognizing problems, and developing interventions. This lecture sets us up for the development of stock-and-flow models for system dynamics simulation modeling starting in the next lecture.

Lecture C1 - Building More Complex Causal Loop Diagrams (2020-02-04)

This lecture covers the process of building more complex causal loop diagrams. We first start with behavior over time trajectories and use them to guide what loops (and then what variables) should be included in our models. We then flip to a perspective where we focus on what kind of problem we're interested in modeling and add loops until we form basic archetypes (with expected behaviors over time) that we can further customize.

Lecture B3 - Morecroft (2010), Chapter 2, Introduction to Feedback Systems Thinking (2020-01-30)

In this lecture, we discuss content from Chapter 2 out of Morecroft (2010), which introduces feedback systems thinking. We focus on including enough aspects of a system to form loops in causal loop diagrams and how certain combinations of loops often lead to certain predictable modes of behavior. We close with an example of how modeling coupled showers in a hotel gives us insight into other systems with interdependencies.

Lecture B2 - Drawing Causal Loop Diagrams in Vensim (2020-01-28)

Primarily an introduction to using the Vensim PLE software tool to draw causal loop diagrams (CLD's) that are fit to be included in reports and presentations (or as the first stage of a more-involved system dynamics modeling process leading up to stock-and-flow diagrams). Also includes a few CLD construction/repairing examples within the presentation.

Lecture B1 - Introduction to Causal Loop Diagrams (2020-01-23)

In this lecture, we motivate the use of causal loop diagrams (CLD's) to better understand how feedback loops interact in complex systems. We introduce the CLD, how to draw them, how to find and label the loops, and the basic method of identifying patterns of loops that have commonly associated behavior over time.

Lecture A3 - Morecroft (2010), Chapter 1, Appeal and Power of Strategic Modeling (2020-01-21)

This lecture covers Chapter 1 of Morecroft (2010), which focuses on the Appeal and Power of Strategic Modeling. We introduce the modeling continuum (from analog to metaphorical) and discuss how increasing realism in models can come at the cost of reducing generality. We then using the iThink software to show how simple stock-and-flow models can be simulated on a computer to reveal interesting behaviors over time.

Lecture A2 - Introduction to Modeling (2020-01-16)

Further discussion of what is a "model" (in general), highlighting a wide variety of models including those instantiated in the physical world (such as laboratory model systems) to those formed within the mind (mental models). We then discuss the strengths and weaknesses of quantitative models (relative to mental models), the three different uses of quantitative modeling, and the three different kinds of simulation modeling (with an emphasis on system dynamics modeling).

Lecture A1 - Course Introduction (2020-01-14)

Introduces the course, its structure, and its policies. Also begins to introduce the notion of a "model" and the kinds of system dynamics models that will be covered in the course.