3.4.2
|
Mercury uses a new solution to underflows for HMTVDD. If an
underflow is detected, we compute the total storage in the delay. If this sum is less than 0.0001, then
all elements in the array are set to zero, and no warning is signaled. If the sum is greater or equal to
0.0001, then the single underflowing element is set to zero, and a warning is signalled.
This has not yet been implemented for HM2DTVDD.
|
3.4.1
|
Bug fix to make Mercury compatible with old models in which the
dispersion property of the TVDD was a scalar, rather than a vector of six scalars.
|
3.4
|
Support for a hexagonal grid of models, aka, spatial simulation.
|
3.3.14
|
Changed the behavior for circular dependencies and state underflows of time-varying distributed delays. These now issue non-fatal warnings.
|
3.3.13
|
Fixed a problem with interpreting the Start Date parameter. (Apple stopped supporting a function I was using.)
|
3.3.12
|
Fixed problem where Outputs were not correctly read from the
Crocus file.
|
3.3.11
|
A few minor adjustments to UI elements.
|
3.3.10
|
Fixed problem with scolling of dependent variable list in QuickLook. Fixed
circular dependency, again.
|
3.3.9
|
Report location of output file.
|
3.3.8
|
Bug fixes.
|
3.3.7
|
Fixed a couple of bugs in Quick Look.
|
3.3.5
|
Fixed problem with life cycles being incorrectly flagged as ciruclar dependencies.
|
|
Added GCD check box. This turns on parallel execution.
|
3.3.2
|
Output variables can now be integrators.
|
3.3.1
|
Fixed a problem where constant arrays that were entered literally in an HTL program were not evaluated properly during a simulation.
A simulation output which is an array will appear in QuickLook as the sum of the array elements.
|
3.3
|
Mercury updated. Crocus checks if the model has unsaved changes before running a simulation.
|
3.2.0
|
Adds support for new HTL part.
|
3.1.8
|
Fixed bug in debugging code.
|
3.1.7
|
Fixed identification of object dragged from the model outline view.
Added tool tips to show full paths in Quick Look.
Added Abort button for simulation executions.
Adding a stepper makes document dirty.
|
3.1.6
|
Fixed memory leaks in the HTL executor.
Redesigned dependency graphing.
|
3.1.5
|
Fixed parsing errors in Mercury for numbers and path names.
|
3.1.4
|
Fixed problem where the dependent variable controls in Quick Look were mislabeled.
|
3.1.3
|
Fixed problem with parts that have the default name.
|
3.1.2
|
Speed improvements to Mercury.
|
3.1
|
Adds 2D plots.
|
3.0.1
|
I hope this is now a universal binary.
|
3.0
|
Major new feature provides 3D plots of stepper simulations. Substantial speed improvements in Mercury.
|
2.1.3
|
Support for 2-D time varying distributed delays.
|
2.1.2
|
A special variable "julianday" provides the Julian Day for HTL expressions.
|
2.1.1
|
The simulation start day is now a calendar date.
|
2.1
|
The simulation start day is now user settable.
|
2.0.5
|
Fixed bug: unconnected inputs were not properly assigned zero.
|
2.0.4
|
Improved search algorithm for ports. Improved error handling for unfound ports.
|
2.0.3
|
Check to see if any output is undefined after evaluation of an HTL expression.
|
2.0.2
|
Changed status display from a table to a view. Improved error reporting of HTL in Mercury.
|
2.0.1
|
Fixed double malloc free when using HMInputSplitter with arrays.
|
2.0
|
Crocus now has steppers, which allow parameters to be systematically varied across many simulations to produce sensitivity data. For a tutorial, download the movie from the Abstract Tools web site.
|
1.0.9
|
Improved error reporting for HTL: undefined variables are detected and reported.
|
1.0.8
|
Added HTL function "frac" to return the fractional part of a number. Improved error reporting for HTL.
|
1.0.7
|
Fixed bug with unconditional exception in HMExpression updateRates.
|
1.0.6
|
Mercury now terminates gracefully on end-of-file for a weather file. All abnormal terminations now use NSException.
|
1.0.5
|
Fixed a problem in which Crocus did not know if a document was changed. Fixed problem in Mercury so that files which had exhausted their data will return zero, and give an alert, rather than falling into an infinite loop. This is a short term solution. Probably this condition should cause the simulation gracefully to terminate.
|
1.0.4
|
Fixed problem with decimal numbers that caused them to be increased 10-fold when the setup file was saved---sometimes.
|
1.0.3
|
Check for unsaved changes before running the simulation.
|
1006:
|
Added automatic application updating.
|
