| Preventing Financial Outbreaks [message #1903] |
Sat, 04 April 2009 07:11  |
Eclipse User |
|
|
|
Hi again,
Just adding to the idea of predicting other global disasters (and I guess
the Plague or some form of Flu has caused not more victims than are
currently being fired or have already lost their jobs worldwide??) with
adapted formulas and methods yours are based on...
You sure know, some of your colleagues used money bills and how they
circulate around the world to predict the spread of air-born diseases like
flu. Not that all "pyramid games" or similar reasons for this financial
condition are transmitted from hand to hand, but some certainly have reached
other countries than those they originated from.
Maybe other core data than simple population was required here of course.
Wealth or income per capita rather than how many people live in a country.
And some parts of the world where Health-based STEM is most needed may
therefore not be prime targets for this potential variation ;-)
Werner
|
|
|
|
|
|
|
| Re: Preventing Financial Outbreaks [message #2000 is a reply to message #1986] |
Wed, 08 April 2009 09:23  |
| Eclipse User |
|
|
|
I think you're asking if STEM is "agent" based or not. Not. It is "equation
based." It would be possible to create a kind of hybrid version of STEM in
which particular nodes in the graph had labels that were updated by the
results of separate agent based simulations.
--
Daniel Ford
IBM Almaden Research Center
San Jose, CA
"Miles Parker" <milesparker@gmail.com> wrote in message
news:grh0g2$o66$1@build.eclipse.org...
>
> Yes, there has also been quite a bit of work in financial contagion and
> similar effects in the ABM community. Here we'd model companies, traders,
> countries, etc.. as seperate actors with realtionship graphs and a set of
> boudedly rational rules. I'm not exactly sure how this docks to the STEM
> setup but it is beggining to sound similar. It is interesting how people
> in different communities slice a problem up in the same way but call it
> something different. Ecologists for example say "Individual-Bsed
> Modeling". And I sometimes I think a better name for Agent-Based modelling
> would be "disaggregate modelling"
>
> Daniel and others, would you characterize a typical STEM model as an
> equation-based vs. algorithmic-based (individual level) model? Is this a
> useful distinction to make? Is it somewhere in between? IOTW what does a
> typical graph node represent?
>
> On 2009-04-07 08:05:47 -0700, "Daniel Ford" <daford@almaden.ibm.com> said:
>
>> Nope, it's definitely not a "weired" idea. The way STEM maintains it's
>> model state in a graph and the way it sequences changes to the graph it
>> is
>> quote possible to "layer" different kinds of models with the lower level
>> models not being "aware" of the ones "above" them. For instance, it
>> would
>> be possible to create an economic model on top of an existing disease
>> model.
>> The economic model could reference the computations performed by the
>> disease
>> model and factor them into its own calculations.
>
>
|
|
|
|
| Re: Preventing Financial Outbreaks [message #560614 is a reply to message #1903] |
Mon, 06 April 2009 09:46 |
| Eclipse User |
|
|
|
Using STEM for things other than disease modeling is somethign we've
considered for quite a while. We just recently completed a major
refactoring of the data sets to separate the different parts into separate
plug-ins to make this much easier. This is in the current HEAD of the STEM
CVS. For instance, the geographic data, disease data, population data and
instructure data are all separated out; they had been combined into a single
plug-in previously.
This should make it easier to make models for other domains like finance or
disaster planning.
Daniel Ford
IBM Almaden Research Center
San Jose, CA
"Werner Keil" <werner.keil@gmx.net> wrote in message
news:gr7f94$m92$1@build.eclipse.org...
> Hi again,
>
> Just adding to the idea of predicting other global disasters (and I guess
> the Plague or some form of Flu has caused not more victims than are
> currently being fired or have already lost their jobs worldwide??) with
> adapted formulas and methods yours are based on...
>
> You sure know, some of your colleagues used money bills and how they
> circulate around the world to predict the spread of air-born diseases like
> flu. Not that all "pyramid games" or similar reasons for this financial
> condition are transmitted from hand to hand, but some certainly have
> reached other countries than those they originated from.
>
> Maybe other core data than simple population was required here of course.
> Wealth or income per capita rather than how many people live in a country.
> And some parts of the world where Health-based STEM is most needed may
> therefore not be prime targets for this potential variation ;-)
>
> Werner
|
|
|
| Re: Preventing Financial Outbreaks [message #560622 is a reply to message #1933] |
Mon, 06 April 2009 18:08 |
| Eclipse User |
|
|
|
Good to hear.
I may have a look at this CVS head over Easter. And either this week or
after the holiday also try to join your call.
Mike Milinkovich confirmed, the idea sounded interesting to him, and if a
project was technically open and capable for such ideas, no idea would be
considered "weird" or crazy at Eclipse ;-)
Werner
|
|
|
| Re: Preventing Financial Outbreaks [message #560639 is a reply to message #1947] |
Tue, 07 April 2009 11:05 |
| Eclipse User |
|
|
|
Nope, it's definitely not a "weired" idea. The way STEM maintains it's
model state in a graph and the way it sequences changes to the graph it is
quote possible to "layer" different kinds of models with the lower level
models not being "aware" of the ones "above" them. For instance, it would
be possible to create an economic model on top of an existing disease model.
The economic model could reference the computations performed by the disease
model and factor them into its own calculations.
--
Daniel Ford
IBM Almaden Research Center
San Jose, CA
"Werner Keil" <werner.keil@gmx.net> wrote in message
news:grduhc$qjk$1@build.eclipse.org...
> Good to hear.
> I may have a look at this CVS head over Easter. And either this week or
> after the holiday also try to join your call.
>
> Mike Milinkovich confirmed, the idea sounded interesting to him, and if a
> project was technically open and capable for such ideas, no idea would be
> considered "weird" or crazy at Eclipse ;-)
>
> Werner
|
|
|
| Re: Preventing Financial Outbreaks [message #560645 is a reply to message #1972] |
Tue, 07 April 2009 22:00 |
| Eclipse User |
|
|
|
Yes, there has also been quite a bit of work in financial contagion and
similar effects in the ABM community. Here we'd model companies,
traders, countries, etc.. as seperate actors with realtionship graphs
and a set of boudedly rational rules. I'm not exactly sure how this
docks to the STEM setup but it is beggining to sound similar. It is
interesting how people in different communities slice a problem up in
the same way but call it something different. Ecologists for example
say "Individual-Bsed Modeling". And I sometimes I think a better name
for Agent-Based modelling would be "disaggregate modelling"
Daniel and others, would you characterize a typical STEM model as an
equation-based vs. algorithmic-based (individual level) model? Is this
a useful distinction to make? Is it somewhere in between? IOTW what
does a typical graph node represent?
On 2009-04-07 08:05:47 -0700, "Daniel Ford" <daford@almaden.ibm.com> said:
> Nope, it's definitely not a "weired" idea. The way STEM maintains it's
> model state in a graph and the way it sequences changes to the graph it is
> quote possible to "layer" different kinds of models with the lower level
> models not being "aware" of the ones "above" them. For instance, it would
> be possible to create an economic model on top of an existing disease model.
> The economic model could reference the computations performed by the disease
> model and factor them into its own calculations.
|
|
|
| Re: Preventing Financial Outbreaks [message #560652 is a reply to message #1986] |
Wed, 08 April 2009 09:23 |
| Eclipse User |
|
|
|
I think you're asking if STEM is "agent" based or not. Not. It is "equation
based." It would be possible to create a kind of hybrid version of STEM in
which particular nodes in the graph had labels that were updated by the
results of separate agent based simulations.
--
Daniel Ford
IBM Almaden Research Center
San Jose, CA
"Miles Parker" <milesparker@gmail.com> wrote in message
news:grh0g2$o66$1@build.eclipse.org...
>
> Yes, there has also been quite a bit of work in financial contagion and
> similar effects in the ABM community. Here we'd model companies, traders,
> countries, etc.. as seperate actors with realtionship graphs and a set of
> boudedly rational rules. I'm not exactly sure how this docks to the STEM
> setup but it is beggining to sound similar. It is interesting how people
> in different communities slice a problem up in the same way but call it
> something different. Ecologists for example say "Individual-Bsed
> Modeling". And I sometimes I think a better name for Agent-Based modelling
> would be "disaggregate modelling"
>
> Daniel and others, would you characterize a typical STEM model as an
> equation-based vs. algorithmic-based (individual level) model? Is this a
> useful distinction to make? Is it somewhere in between? IOTW what does a
> typical graph node represent?
>
> On 2009-04-07 08:05:47 -0700, "Daniel Ford" <daford@almaden.ibm.com> said:
>
>> Nope, it's definitely not a "weired" idea. The way STEM maintains it's
>> model state in a graph and the way it sequences changes to the graph it
>> is
>> quote possible to "layer" different kinds of models with the lower level
>> models not being "aware" of the ones "above" them. For instance, it
>> would
>> be possible to create an economic model on top of an existing disease
>> model.
>> The economic model could reference the computations performed by the
>> disease
>> model and factor them into its own calculations.
>
>
|
|
|
|
Powered by
FUDForum. Page generated in 0.15863 seconds
Home