Discrete and

Continuous Simulation
Marcio Carvalho
Luis Luna
PAD 824 Advanced Topics in System Dynamics
Fall 2002
PAD 824 Advanced Topics in System Dynamics
Fall 2002
What is it all about?
Numerical simulation approach
Level of Aggregation
Policies versus Decisions
Aggregate versus Individuals
Aggregate Dynamics versus Problem solving
Difficulty of the formulation
Nature of the system/problem
Nature of the question
Nature of preferred lenses
PAD 824 Advanced Topics in System Dynamics
Fall 2002
Basic concepts
1. Static or dynamic models
2. Stochastic, deterministic or chaotic models
3. Discrete or continuous change/models
4. Aggregates or Individuals
PAD 824 Advanced Topics in System Dynamics
Fall 2002
1. Static or Dynamic models
Dynamic: State variables change over time
(System Dynamics, Discrete Event, Agent-
Based, Econometrics?)
Static: Snapshot at a single point in time
(Monte Carlo simulation, optimization models,
etc.)
PAD 824 Advanced Topics in System Dynamics
Fall 2002
2. Deterministic, Stochastic or
Chaotic
Deterministic model is one whose behavior is
entire predictable. The system is perfectly
understood, then it is possible to predict precisely
what will happen.
Stochastic model is one whose behavior cannot be
entirely predicted.
Chaotic model is a deterministic model with a
behavior that cannot be entirely predicted
PAD 824 Advanced Topics in System Dynamics
Fall 2002
3. Discrete or Continuous
models
Discrete model: the state variables change
only at a countable number of points in time.
These points in time are the ones at which
the event occurs/change in state.
Continuous: the state variables change in a
continuous way, and not abruptly from one
state to another (infinite number of states).
PAD 824 Advanced Topics in System Dynamics
Fall 2002
3. Discrete or Continuous
models
Continuous model: Bank account
Principal
Interest
Average
Interest Rate
Noise
Simulated
Principal
Sim Interest
Estimated
Interest Rate
Noise Seed
Observed
Interest Rate
Continuous and Stochastic
Continuous and Deterministic
PAD 824 Advanced Topics in System Dynamics
Fall 2002
3. Discrete and Continuous
models
Discrete model: Bank Account
Simulated
Principal 1 0
Sim Interest 1 0
Average
Principal 0
Averaging
time 0
<Time>
<TIME
STEP>
Observed
Interest Rate 0
<Average
Interest Rate>
Simulated
Principal 1
Sim Interest 1
Average
Principal
Averaging
time
<Time>
<TIME
STEP>
Observed
Interest Rate
<Average
Interest Rate>
<Noise>
Discrete and Stochastic Discrete and Deterministic
PAD 824 Advanced Topics in System Dynamics
Fall 2002
4. Aggregate and Individual
models
Aggregate model: we look for a more distant
position. Modeler is more distant. Policy
model. This view tends to be more
deterministic.
Individual model: modeler is taking a closer
look of the individual decisions. This view
tends to be more stochastic.
PAD 824 Advanced Topics in System Dynamics
Fall 2002
The Soup of models
Waiting in line
Waiting in line 1B
Busy clerk
Waiting in line (Stella version)
Mortgages (ARENA model)

PAD 824 Advanced Topics in System Dynamics
Fall 2002
Time handling
2 approaches:
Time-slicing: move forward in our models in equal
time intervals.
Next-event technique: the model is only examined
and updated when it is known that a state (or
behavior) changes. Time moves from event to event.
PAD 824 Advanced Topics in System Dynamics
Fall 2002
Alternative views of
Discreteness
Culberstons feedback view

TOTE model
(Miller, Galanter and Pribram, 1960)
) )( ' ( ) ' (
2 1 1

t t t t t
Y Y d d Y g g a Y
Test
Operate
(Congruity)
(Incongruity)
PAD 824 Advanced Topics in System Dynamics
Fall 2002
Peoples thoughts
The system contains a mixture of discrete
events, discrete and different magnitudes, and
continuous processes. Such mixed processes
have generally been difficult to represent in
continuous simulation models, and the common
recourse has been a very high level of
aggregation which has exposed the model to
serious inaccuracy
(Coyle, 1982)
PAD 824 Advanced Topics in System Dynamics
Fall 2002
Peoples thoughts
Only from a more distant perspective in which
events and decisions are deliberately blurred
into patterns of behavior and policy structure
will the notion that behavior is a consequence
of feedback structure arise and be perceived to
yield powerful insights.
(Richardson, 1991)
PAD 824 Advanced Topics in System Dynamics
Fall 2002
So, is it all about these?
Numerical simulation approach
Level of Aggregation
Policies versus Decisions
Aggregate versus Individuals
Problem solving versus Aggregate Dynamics
Difficulty of the formulation
Nature of the system/problem
Nature of the question
Nature of preferred lenses