Phyto-IT demonstrated its PhytoSim modelling and simulation software at the HortiModel 2012 conference (Nanjing, China, November 4-8, 2012).

The HortiModel 2012 conference was the Fourth International Symposium on Models for Plant Growth, Environmental Control and Management in Protected Cultivation. It covered a wide range of topics related to horticulture:

• Crop and Climate Management
• Water, Nutrient and Energy Management
• Decision Support and Farm Management
• Product Quality
• Systems Biology, Gene-plant-crop Modelling
• 3D-models, Architectural Models
• Methodological Issues, Aggregation and Scale
• Combination of Models and Sensors

Thanks to all the participants who came by our booth!

Version 1.3 of PhytoSim has been released. This release comes with updates for all modules: Data I/O, Modelling, Simulation, Calibration, Sensitivity and Uncertainty.

Download it here (WARNING: Mac OS X Snow Leopard is no longer supported). Install the new version and update the modules in the Module Manager.

[UPDATE] If you are experiencing problems updating the Simulation module, delete the PhytoSimSimulation folder in: “My Documents/PhytoSim/Modules” and continue updating the other modules.

For this occasion, all trial accounts have been reset enabling everyone to try out the new release.

Key new features:

• User Guide search (Help > User Guide; search field is located at the top).
• The current workspace can now be closed (File > Close).

• Data I/O
• Titles and units of input columns can now also be edited by clicking a settings button in the header of a selected input column.

• Modelling:
• Added support for unit substances, e.g. cm^3[H2O].g^-1[leaf].
• New model: CarbohydrateSupplyDemand, carbohydrate supply and demand model for vegetative growth, Gent and Seginer (2012).
• New function: valueat( variable, at ), returns the value of a variable at any past time.
• New function: average( variable, from, to ), returns the average of a variable between two points in time.
• New function: integral( variable, from, to ), returns the integral of a variable between two points in time.
• New function: uniform( minimum, maximum ), returns a uniformly distributed random number.
• New function: triangular( minimum, mode, maximum ), returns a triangularly distributed random number.
• New function: normal( mean, stddev ), returns a normally distributed random number.
• New function: poisson( mean ), returns a poisson distributed random number.
• Parentheses matching in the model editor.
• Units in equations are no longer automatically converted: stricter but less confusing! As a consequence, unit conversion warnings no longer show up. WARNING: If you relied on automatic unit conversions in your models, they will not be valid anymore. Inserting the proper unit conversion factors will fix this.

• Simulation:
• New solver: CVODE stiff/non-stiff differential equation solver (Settings > Solver).
• New example: Stiff, shows off the benefits of the CVODE solver for stiff differential equations (in Examples/PhytoSimSimulation/Misc).
• New example: CarbohydrateSupplyDemand, simulation example for the new CarbohydrateSupplyDemand model (in Examples/PhytoSimSimulation/Plant Physiology).
• New example: Random, illustrates the random number generation functions (in Examples/PhytoSimSimulation/Misc).
• New example: ExpDecay, illustrates the valueat, average and integral functions (in Examples/PhytoSimSimulation/Misc).
• Option to turn off curve painting optimization (Graph Settings > Curves). Only needed in some rare cases.
• A warning is given when no plot interval (Settings > Graphs) is set: protects against too many curve points.

• Calibration:
• New global optimizer: Shuffled Complex Evolution.
• New global optimizer: Simulated Annealing.
• New global optimizer: Random Search.
• New example: SixHumpCamelBack, global optimizer test function (in Examples/PhytoSimCalibration/SixHumpCamelBack).

• Uncertainty:
• New uncertainty distribution type: normal distribution.

• Sensitivity:
• Data variables can now also be used as source components (like parameters and initial values of derived variables).
• Setting to turn off numerical error checking (workspace specific).

Updated PhytoSim Calibration (v1.2.2) and Simulation (v1.2.1) modules are available. These updates fix a problem with missing Data I/O data when starting a simulation or calibration.

To install the new modules start PhytoSim, head to the Module Manager and update the Simulation and Calibration modules. Also make sure you have upgraded to the latest PhytoSim version (v1.2.2).

PhytoSim has been updated to solve an issue that didn’t allow updating trial modules without purchasing depending modules.

The new version can be found at the PhytoSim download page.

An update of the PhytoSim calibration module is available (v1.2.1). This release fixes a bug on Windows that caused the application to become unresponsive when plotting a large number of infinite objective values on the “Objective Value vs. Evaluation Number” graph.

The calibration objective is given an infinite value when the optimizer tries to run a model evaluation (simulation) with optimizer variable values on or outside the allowed bounds. These values are plotted on the objective graph as vertical lines (see screenshot). It is normal that the optimizer occasionally ventures close to, or over the bounds. However, if the optimizer repeatedly tries to run evaluations with “illegal” optimizer variable values, it might be an indication  that the optimum is located outside the optimizer variable bounds. If that is the case, the bounds of the optimizer variables which are causing the problem should be relaxed, always keeping in mind that they should still make physical sense.

To install the new module start PhytoSim, head to the Module Manager and update the calibration module.

After a few months of beta, today the PhytoSim Uncertainty Analysis and Sensitivity Analysis reached their first stable version (v1.0). As of this release, both modules are no longer free to use, but as always, a 30-day free trial is available for both of them.

Compared to the beta version, only minor changes were made to the Uncertainty Analysis module. However, the new Sensitivity Analysis module does come with a few important new features:

• Visualisation of the simulation results used for the sensitivity calculations. Access through the ‘More Graphs’ toolbar button.
• Visualisation of the absolute sensitivities for each source and target component combination. Access through the ‘More Graphs’ toolbar button.
• The progress bar now also shows the identifiability analysis progress.
• The identifiability analysis is stopped once a subset size is found for which all subsets are unidentifiable. This greatly reduces the calculation time when a large number of model parameters and/or initial conditions are used in the analysis.

More details about each of the modules (pricing, change log, user guide) can be found on the PhytoSimUncertainty and PhytoSimSensitivity webpages.

To start using the new versions: launch PhytoSim, select the Module Manager and upgrade to the new module versions.

PhytoSim 1.2.1 has been released. This is a maintenance release which improves the stability of the login and module manager system. Make sure you upgrade to this version in order to avoid future problems logging in and installing module updates. No modules need to be re-installed.

Direct download links: Windows and Mac.

Ever wondered what the uncertainty of your model predictions is? Using the NEW PhytoSim Uncertainty Analysis module you will be able to find out just that.

An uncertainty analysis allows propagating parameter, initial condition and/or input uncertainties through the model to obtain the model output uncertainty. Setting up the analysis requires almost no work, just assign the source uncertainties, select the number of model evaluations and you are good to go. The well-known Monte Carlo technique is used to propagate the uncertainties and Latin Hypercube sampling is used to ensure that the uncertainty distributions are sampled evenly.

Screenshot of a source uncertainty while it’s being sampled:

Screenshot of the model output uncertainty (maximum, 95th percentile, mean, median, 5th percentile and minimum of the distribution):

For more information have a look at the PhytoSim Uncertainty Analysis User Guide webpage.

Pricing: Free for now (beta period will last until the end of 2011), afterwards 499 EUR.

Installation:

• Make sure you have the latest PhytoSim version.
• Select the Uncertainty Analysis module in the PhytoSim module manager and select ‘Install module’ from the toolbar.

Probably the most important aspect of any modelling study is knowing which of your model parameters (and/or initial conditions) contribute most to the model output. As of today, such analysis is possible in PhytoSim.

The PhytoSim Sensitivity Analysis module offers 3 types of information:

Dynamic sensitivity functions: Shows how small changes in model parameters change the model output. By analyzing the sensitivity functions you will get a tremendous insight into your model.

Importance ranking: Ranks the model parameters according to their influence on the model output and shows this on a nice bar chart.

Identifiability analysis: Use the intuitive user interface to find out which parameter combinations can be estimated based on a given set of measurements. You can even do the analysis before any real data is collected.

For more information have a look at the PhytoSim Sensitivity Analysis User Guide webpage.

Pricing: Free for now (beta period will last until the end of 2011), afterwards 499 EUR.

Installation:

• Make sure you have the latest PhytoSim version.
• Select the Sensitivity Analysis module in the PhytoSim module manager and select ‘Install module’ from the toolbar.

Version 1.2 of PhytoSim has been released. This release comes with updates for following modules: Data I/O, Modelling, Simulation and Calibration.

Download the Windows version here and the Mac version here.

For this occasion, all trial accounts have been reset enabling everyone to try out the new release.

Key new features:

• Work offline on the Mac.
• ‘Remember me’ option when logging in.
• Data I/O:

• Use date/time in outputs (converts from decimal time).
• New output type: matrix file. Can contain matrix output from various PhytoSim modules.
• Modelling:

• New nounit( x ) function: returns the value of x without a unit, where x is a model component.
• New pi() function: returns the number pi.
  WARNING: models having a ‘pi’ variable will now be invalid. Please replace these with ‘pi()’.
• Multi-line comments. Use: ‘/*’ and ‘*/’ to enclose multiple lines of comments.
• Added Collapse/Expand to the model navigator context menus to quickly expand/collapse the model tree.
• New model: HydR, linking diameter changes to radial water transport across a radial hydraulic resistance (Steppe et al., 2011). Only available for non-trial accounts.
• Simulation:
• Added the ability to put all model component values to their defaults and/or initial values.
• Added Collapse/Expand to the model navigator context menus to quickly expand/collapse the model tree.
• Calibration:
• Confidence information calculations: estimation errors and correlation matrix.
  
• ‘Single shot’ optimizer: evaluate the objective at the initial values.
• ‘Best’ simulation curve now has a +/- 2*stdev measurement uncertainty band on it.
• Minimum and maximum graph marker lines for optimizer variables.

For a more detailed list of all new features, changes and bug fixes see:

• PhytoSim change log
• Data I/O change log
• Modelling change log
• Simulation change log
• Calibration change log