Download technology basics program

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts
no text concepts found
Transcript
Explorations in Computational Science:
Hands-on Computational Modeling using STELLA
Presenter:
Robert R. Gotwals (“Bob2”)
Shodor Education Foundation, Inc.
Neura l tp r
ANS i no
Humo ral tpr
Sys Art Vo l
P Ven Sys
Pul Runoff
Pul Art Vol
P Ven Pu l
Sys Run off
~
C sys
Vo l in
P Art Sys
COsys
Vo l ou t
P Pu l Art
Sys Ven Vol
P Rt Hrt
P L t Hrt
Lt Hrt Vo l
COpul
Pul Ven Vol
VR L t Hrt
VR Rt Hrt
Lt Hrt Ela stan ce
National Computational Science Leadership Program (NCSLP)
Rt Hrt Vol
Rt Hrt Elastance
1
Session Goals



First experience in computational science
» Application: computational epidemiology
» Algorithm: 1927 Kermack-McKendrick SIR algorithm
First experience with an Architecture: STELLA on a PC
» computational tool
– STELLA
Logistics
» Short overview
» Hands-on model building exercise
» Extensions as time permits
National Computational Science Leadership Program (NCSLP)
2
System Dynamics


A method of studying dynamic (time-driven) phenomena through the
use of:
» Computer simulations based on ordinary differential equations
» Development of causal mechanisms (feedback loops)
» Analysis of the factors that affect a system (a collection of
interacting elements)
Examples:
» Interactions of predators and prey in an ecosystem
» Fate, transport, and distribution of a pharmaceutical through a
patient
» Photooxidation of precursor atmospheric pollutants becoming
ozone
National Computational Science Leadership Program (NCSLP)
3
STELLA Highlights







no programming skills required
Relatively short learning-curve
icon-based: modelers need to understand functions of icons
graphs and tables easily constructed, manipulated, exported
mathematical engine underlying software fairly robust
mathematics is transparent to users
“authoring” capabilities, provides user-friendly graphical
interface to underlying models
National Computational Science Leadership Program (NCSLP)
4
Mathematical Basis of STELLA (an
intro to ODE's!)



we wish to be able to study events as they change over time.
main question: how do different elements change the event
over time?
Example: how does one's height change over time?
² height
² time
= some mathematical equation ( or function)
Suppose the height of an animal increases by 20% every year until it
is 20 years old?
National Computational Science Leadership Program (NCSLP)
5
Mathematical Basis of STELLA (an
intro to ODE's!)
²h
²t
=
0.2h
or
dh = 0.2h
dt
Do some algebraic rearrangements:
dh = 0.2 h dt
then "integrate" both sides from starting time to stopping time:
t20
 dh 0.2hdt
where h at start equals 1 inch
t 0
h = e 0.2h * (20 - 0) (by definitions from integral calculus) = 54.6 inches
National Computational Science Leadership Program (NCSLP)
6
STELLA Implementation
National Computational Science Leadership Program (NCSLP)
7
STELLA Basic Elements




Stocks
» act as "accumulators", have an initial
value
» viewed as having some unit
» are the "nouns" (things) for the system
Flows
» provide input/output to the stock
» have value of unit/time (unit same as
stock unit)
» are the "verbs" for the system
Converters
» hold constants or change units
» can be algebraic or graphical
» are the "adverbs" or "adjectives"
Connectors
National Computational Science Leadership Program (NCSLP)
8
Case Study: Simple Epidemiology Model:
Influenza Epidemic in a Boarding School


Source: Mathematical Biology, J.D. Murray, Springer-Verlag,
1989.
Background:
» In 1978, a study was conducted and reported in the British
Medical Journal (4 March 1978) of an outbreak of the
influenza virus in a boys boarding school. The school had a
population of 763 boys. Of these 512 were confined to bed
during the epidemic, which lasted from 22 January until 4
February. It seems that one infected boy initiated the
epidemic. At the outbreak of the epidemic, none of the boys
had previously had influenza, so no resistance to the
infection was present.
National Computational Science Leadership Program (NCSLP)
9
Case Study: Influenza Epidemic



Goal: create a
computational model of
the boarding school
epidemic
Algorithm: 1927
Kermack-McKendrick
SIR algorithm
Three “types” of
students:
» Susceptibles
» Infecteds
» Recovereds
dS
 rSI
dt
dI
rSIaI
dt
dR
aI
dt
where:
S= population of susceptible people
I = population of infected people
R = population of recovered people
r = probability of becoming infected
a = infectious period
National Computational Science Leadership Program (NCSLP)
10
Case Study: Influenza Epidemic

Extensions:
» What is the effect of vaccinations? Add a vaccination
algorithm. Use sensitivity analysis to analyze effect
» What is the effect of a return to susceptibility? Add a
return loop to susceptibility
» Add possibility of deaths, both as a result of disease and
natural deaths
» Add possibility of influx of new susceptibles
National Computational Science Leadership Program (NCSLP)
11