INSTALLATION

Type the following command to unpack the prometheus archive.

tar -zxf prometheus.tar.gz

A new directory entitled 'prometheus' will be created within your
current working directory.


EXECUTION

You must be in the 'prometheus' directory to run the program.  Start
the program using the following command.

./prometheus.sh


INPUT

Model:
The format for model files is described in X

Simulation Data: 

The first line must consist of the space-delimited names of each input
variable.  The first variable must be t, which represents time.  The
second line must consist of initial values for time and each input
variable.  These values must also be delimited by a single space.  The
remaining lines in the file should contain a value for t indicating
the time step at which the values on that line were observed.  Values
for each variable at that time step can be provided.  Missing values
are allowed.  Values for all exogenous variables should be provided at
each time step.  For intermediate time steps, the prior value of the
exogenous variable will be assumed.

Ensure that there are no extra spaces at the end of lines or in
between values.

Ex/
X and Y are input variables, Z is exogenous
t X Y Z
0 4 2 5.3
1 3 1 1.2
2   9.4

Note that in the last line, values for X and Y are missing.  Thus
there are no numbers between the delimiting spaces, but the delimiting
spaces must still exist.

Revision Data (from RPMDocumentation.txt in the RPM directory):

This file contains the time series data which will be used by RPM as
measure of quality of the models.

The first column in DATAFILE represents the time points and is tagged
by "t" at the top. Each of the other columns contains the observed
values of a single observed variable at the corresponding time points
given in the first column. Also each column is tagged by the variable
name at the top. You may use spaces or tabs as delimiter.

In addition, the first two rows in the DATAFILE can optionally include
some desired initial values for an arbitrary set of variables. This is
particularly useful to manually set initial values for hidden
variables. If this option is not used, the system will assume an
initial value of zero for all of the hidden variables in the model.

(For an example refer to RPM/examples/rossdatatrain.txt)


OUTPUT

Simulation:

Simulation of a model produces a file called 'simulation.out' in the
'prometheus' directory.  This file can be parsed to extract the values
of each variable at each sampled time step.

Revision:

You will generally not need to work with the output files produced by
revision, however, the file <model_name>.data.rev.out will contain the
initial conditions for each model produced during revision.  The file
<model_name>.rev.out contains the list of models produced by the
revision.  The file <model_name>.rev.in contains the input to the
revision subsystem.


EXAMPLES

Example models can be found in the 'prometheus/models' directory.
Some of these models contain data for revision and simulation.  The
models predator_prey and ross_sea are suggested for exploring the
capabilities of the environment.


SIMULATION

To simulate a model, you must provide the program with data (see
above), the size of the time steps and the number of time steps.


REVISION

To revise a model you must provide a data file (see above),
constraints on the search space, and a type definition file.
Currently the revision subsystem uses a format for declaring types
that differs from that used by the GUI environment.  By default the
revision subsystem assumes that each type in the model is a direct
subtype of 'number'--the most general type.  To specify a type
hierarchy for revision purposes, use the following format.

<type_name> is a <type_name>;

For example, we can create two subtypes of the type species by writing

predator is a species;
prey is a species;

Here, species is assumed to be a subtype of number.

The above types are defined in predator_prey.type in the
'prometheus/models/predator_prey' directory.