 |
|
| |
|
|
|
| |
New
Features
-
The Flowchart Editor enables
you to create models graphically using icons.
-
The Parameter Plot feature replaces the Parametric Run
feature in previous versions of Madonna. Now you only have to
work with a single integrated dialog to specify the parameter
to vary and the variables to plot.
-
The Optimize feature searches for parameter values that
minimize an arbitrary expression. This expression is evaluated
at the end of each run (i.e., when TIME equals STOPTIME), so variables
referenced in this expression will have their final values.
-
The Chemical Reactions feature
generates kinetic equations from a set of chemical reaction equations.
-
The curve fitter has been enhanced to support simultaneous fitting
to more than one dataset. For example, you can now fit three variables
in your model to three different external datasets. You can even
specify the relative weight of each variable-dataset pair.
Import
Data Improvements
-
Imported data (now referred to as "datasets") are stored in your
model file. You no longer need to import data each time after
you open your model.
-
Each dataset is assigned a name of your choice when it is first
imported. The name can be changed at any time.
-
The new Datasets window lists the datasets in your model
and allows them to be renamed and discarded.
-
Contents of an existing dataset can be replaced with new data
by importing a new data file and assigning it the same dataset
name.
-
Datasets can be referred to in your model's equations as piecewise-linear
functions.
-
You can import two-dimensional (matrix) datasets and refer to
them in your equations as piecewise-linear functions taking two
arguments.
Slider
Improvements
-
The position and visibility of the slider window is now remembered
separately for each model. A model's slider window is visible
only when one of its windows is active. If you have two or more
models open, switching between their windows may cause the the
sliders to appear, disappear, or move.
-
Double-clicking the text area of a slider opens the Define
Sliders dialog and selects that slider. This makes is easy
to quickly modify a slider's limits.
-
Increment/decrement arrows have been added to each slider. They
are similar to arrows on scroll bars in that clicking them will
change the slider's value by one unit as specified in the Define
Sliders dialog.
-
Checkboxes labelled 10x have been added to each slider.
When checked, the slider's range is reduced by a factor of ten
and adjusted so that the center of the range corresponds to the
slider's current value. The slider's increment is also divided
by ten. This feature makes it easier to fine-tune a parameter's
value with the sliders.
Numerical
Engine Changes
-
Parameters controlling variable-stepsize integration methods (Auto
and Stiff) have been changed somewhat. When a variable-stepsize
method is chosen, a new parameter called DTMIN appears
in the parameter window. DTMIN specifies the minimum
and initial stepsize. The DT parameter is not used by
variable-stepsize methods and is not shown in the parameter window
when such a method is selected.
-
The new STEPSIZE built-in function returns the amount
of time from the previous step to the current step. It returns
zero during the initialization step. A common use of this function
is to assign it to a variable so the stepsize can be plotted when
using variable-stepsize integration methods.
-
The DELAY built-in function has been rewritten so that
it works properly for all integration methods including variable-stepsize
ones. This new implementation requires that the stepsize never
exceed the delay time. If this happens, a "stepsize exceeded delay"
error message appears and the run is aborted.
-
The new SQUAREPULSE built-in function generates a single
square pulse with a specified start time and duration.
-
The interface for plug-in integration methods has changed. If
you have written plug-in integration methods for previous versions
of Madonna, they must be modified and recompiled for version 7.0.
Request the plug-in development kit for version 7.0 (specify Macintosh
or Windows).
-
The random number built-in functions (RANDOM, NORMAL,
and BINOMIAL) use an improved random number generator.
Specifically, they use a lagged Fibonacci generator based on the
FIBMULT
program. Our implementation of this algorithm uses lags of 250
and 103 which results in a period of approximately 10^84.
-
Single-line comments can be added to your equations using the
semicolon character. This style of comment is more suitable for
short comments since the comment automatically ends at the end
of the line.
-
Comments defined with pairs of curly-bracket characters can now
be nested.
-
The default integration method and some parameter values for new
models have been changed.
Other
Changes
-
The curve fitter and boundary value solver automatically run your
model after a solution is found and plot the relevant variables
and datasets.
-
Performance of batch runs on Macintosh has been improved by reducing
the rate at which the Running dialog is updated.
-
The notes window now supports multiple text styles using the toolbar
at the top of the window.
-
Berkeley Madonna automatically creates a new graph window when
you run your model and no window exists in which the results can
be plotted. This change means that the Choose Variables
dialog is never displayed when you attempt to run your model.
-
The plotter now supports logarithmic scales. Simply check the
Log boxes in the Scales page of the Axis Settings
dialog. Note that when log scaling is used, you cannot distinguish
positive and negative values because Berkeley Madonna plots the
absolute value (magnitude) of the data.
-
The number of steps reported in the top right corner of graph
windows is now the actual number of steps taken during model execution
regardless of how many steps were stored (these differ when DTOUT
is nonzero). For averaged batch runs and parameter plots, the
number of steps reported is the sum of the individual runs.
-
[Macintosh Only] The Tile Windows and Stack Windows
commands on the Graph menu arrange the active model's graph windows.
The tile operation positions the graph windows so that they occupy
as much of the screen as possible without overlapping. The stack
operation resizes the graph windows to their default size and
overlaps them so that their title bars are visible.
-
The Duplicate Window command (Graph menu) makes a clone
the active graph window including all runs and settings.
-
Tables now display all runs and datasets just as graphs do. The
TIME column at the left reflects the actual times for the last
run (rightmost column) displayed in the table. Values for other
runs are computed using linear interpolation if their times don't
match the times in the last run.
-
Tables can be printed and copied to the clipboard just like graphs.
-
Text windows (equations, notes, and equation help) can be printed.
-
The Choose Data dialog has been renamed to Choose Variables.
-
The Multiple Runs dialog has been renamed to Batch Runs.
|
|
