Results »

The Results Node displays the solution summary and information about measurement deviations.

Solution Summary

This view is read-only and displays the Solution Summary. It is organized in columns as explained below:

• Weighted LSQ Error: This column shows the Weighted Least Squares (LSQ) Error, which is minimized, at the solution of the Estimation model. The Weighted LSQ Error is the sum of all the weighted component least square deviations.
• LSQ Error: This column shows the unweighted Least Square Error, and is a sum of all the component least square deviations. When Ignore Zero Measurements option is selected in the Weights node, the errors from zero experimental data values are excluded in the calculation of the LSQ Error. This LSQ Error is independent of the weighting strategy and is useful when comparing the absolute percentage deviations between different weighting methods.
• Solution Status: This column displays a message reporting the solution status of the Estimation Run. The list of possible messages is shown below:
  
  • Successful Solution: A feasible solution has been found. When doing Estimation, this message also indicates that the parameter values that minimize the objective function were found. However, some parameters may lie at their bounds. It the latter is true, you may change those bounds, initialize from the current results, and re-run the model. Examine if there is a significant improvement on the objective function that justifies doing more estimation runs.
  
  If the solution is not successful, first try running CheckModel to detect errors in the specifications.
  
  
  • Unbounded: This indicates that the solution is unbounded. For linear kinetics, this message is reliable, but occasionally it appears for difficult nonlinear problems that are not truly unbounded, but lack some bounds to limit the variables to reasonable values.
  • Infeasible: This means that the problem is infeasible. Something is probably wrongly specified in the logic or the data. You should re-examine your problem specification. Infeasibilities could arise from inconsistent bounds, or from overspecification of parameter relationships.
  • Locally Infeasible: This message means that no feasible point could be found for the nonlinear problem from the given starting point. It does not necessarily mean that no feasible point exists. Try changing some options in the Solution Options node, like the Solver used, the Number of finite elements, or the initialization values. Another option may be to run the model as Simulate Only. If that is successful, you may auto-fill the initialization values with the "Initialize Profiles" action, and then run the Estimation with the Warmstart option set to Use Provided Initial Values.
  • Intermediate Infeasible: This means that the current solution is not feasible, but the solver program stopped, either because of a limit (iteration or resource), or because of other difficulties such as arithmetic operation errors. If a specified limit is the reason for the infeasibility, you may increase the iteration or CPU time limits in the Advanced Options Node.
  • Intermediate NonOptimal: This indicates a valid feasible solution, but the local optimization cannot be established as complete.
  • Error Unknown / Error No Solution: There is no solution, and you may have to look at additional diagnostics. In these cases, it CheckModel does not indicate any errors, you may contact Optience for support.
• Solution Date: This column shows the date and time of the Estimation results.

Measurement Deviations

This view is stacked together with the Measurement Deviations by Set and Charts views.
It displays the list of components selected in the Weights → Reconciled Measurements grid. For each component, the Deviation and the Weighted Deviation are displayed. The deviation for pressure is also included when it is reconciled or controlled. When Ignore Zero Measurements option is selected in the Weights node, the zero experimental data points are excluded in the calculation of the Deviation.
The sum of these deviations should match the corresponding Weighted LSQ and LSQ shown in the Solution Summary grid.
By clicking on the header of a column, the rows can be sorted: either in ascending or descending order, by alphabetical or numerical criteria, according to the type of data in the column selected.

Measurement Deviations by Set

This view is stacked together with the Measurement Deviations and Charts views.
The information shown here is similar to the previous view, except that the values are shown individually for each set.

Charts

Four pie charts are displayed here, graphically comparing the contribution of each variable or experimental set to the total Least Squares Error(LSQ). Variables or sets whose LSQ contribution is among the seven biggest are shown, while the remaining ones are merged into the Others sector.
The following charts are shown:

• Weighted Deviation by Set: The chart is done for the weighted LSQ values of the experimental sets. The slices shows the contribution of individual sets to the total weighted LSQ.
  
• Deviation by Set: Similar to Weighted Deviation by Set, except that unweighted LSQ values are used.
  
• Weighted Deviation by Variable: The weighted LSQ values are used for the pie chart. The slices show the contribution of individual variables to the total weighted LSQ.
• Deviation by Variable: Similar to Weighted Deviation by Variable, except that the unweighted LSQ values are used to generate the chart.
  
  

Phase Values

Phase Values is used to select how the phase values of compounds should be displayed in all the child nodes.
It has up to three sub-actions:

Concentration
This is the default option. Concentration (typically Molarity) values are shown for liquid and gas phase.

Moles
The total moles in each phase for every compound is shown.

Partial Pressure for gas; Concentration for Liquid
Concentration (typically Molarity) values are shown for liquid phase and Partial Pressure for gas phase.

Project Explorer

Estimation