Here’s the page we think you wanted. See search results instead:

 

  

Main Support:       Knowledge Center  > ADS Support Home   >   ADS Documentation (all releases)
Documentation:   ADS 2009 Update 1   >  Tuning, Optimization, and Statistical Design   >  Nominal Optimization   >  Optimization Cockpit

This document contains references to Agilent Technologies. Agilent's former Test and Measurement business has become Keysight Technologies. For more information, go to www.keysight.com.


Skip to end of metadata
Go to start of metadata

Optimization Cockpit

The Optimization Cockpit is a live, graphical view of an optimization job. You see the cockpit data (error graphs, goal plots, and variable values) change in real time as the optimization progresses. In addition, you can use the cockpit to control the optimization while the optimization is running. For example, during the course of an optimization, you can increase the range of an optimization variable, modify the limit line of a goal, tune the optimization variables, and change the algorithm from Random to Gradient.

Cockpit Panels

The cockpit has three main panels: Status, Variables, and Goals. It also has a control panel on the left-hand side.

Status panel

The Status panel displays the optimizer's status, type, elapsed time, and progress. It also has a button for changing the optimization algorithm settings.

Activities

  • Roll the pointer over the status icon to know the state of the optimizer
  • While the optimizer is paused or towards the end of an optimization, click Edit Algorithm to modify the optimization algorithm. See #Modifying the algorithm for more information.

Variables panel

The Variables panel displays the optimization variables as a row of sliders and as data in tabular format. It also has buttons for tuning and for editing the variable definitions.

  • The sliders represent the current value of each optimization variable relative to its minimum and maximum values.
  • Each row of the table shows important information for each optimization variable:
    • Name
    • Current value
    • Units (if any)
    • A history graph showing the values of the variable over the course of the optimization.

Activities

  • Click a row in the table to highlight its corresponding slider or vice-versa
  • Move the pointer over a slider to see information about that variable
  • Click Start Tuning to switch to tune mode. See #Tuning for more information.
  • While the optimizer is paused or towards the end of an optimization, click Edit Variables to modify the variable settings. See #Modifying variables for more information.

Goals panel

The Goals panel displays the current error, the error history graph, the goal contribution histogram, and the goals table. It also has a button for editing the goal definitions.

  • The Error history graph shows the overall error over the span of optimization.
  • The Goal contributions histogram is displayed when there are two or more goals. It represents each goal's contribution to the overall error.
  • The table contains a row for each goal. Each row represents the following information:
    • A plot of the goal's response and limit lines. The limit lines are red and the response is blue.

      The solid blue trace is the response for the current values of the optimization variables. The dashed blue trace is the response at the beginning of the optimization.
    • A history plot of the goal's error over the span of optimization.

      Sometimes you will notice the error for a goal increases. The optimizer is driving the overall error to zero, so the error for a particular goal can increase as long as the overall error decreases.

    • A Contributions histogram is displayed only when the goal has more than one limit line. The histogram shows each limit line's contribution to the goal's error.

Activities

  • Move the pointer over the Error history graph to see information about the error:
  • Move the pointer over a bar in the Goal contributions histogram to see information about that goal's contribution to the error:
  • Click a bar in the Goal Contributions histogram to highlight the corresponding goal in the table or click a goal in the table to highlight the corresponding bar in the histogram:
  • Move the pointer over a bar in the Contributions histogram to display the contribution percent and to highlight the corresponding limit line on the plot:
  • Move the pointer over a limit line to see information about that limit line:
  • Move the pointer over a data point in the trace to see the coordinates of the point:
  • While the optimizer is paused or towards the end of an optimization, click Edit Goals to modify the goal settings. See #Modifying goals for more information.

Stale plots

Goal plots go stale when the algorithm, goal, or variable settings change, but no simulation is performed.

This means that the simulation results (error value, the contribution graphs, and the goal plots) displayed in the Cockpit do not accurately reflect the current algorithm, goal, and variable settings. For example, click Edit Goals and modify one of the limit lines you will see the goals plot will go stale. Click Simulate to refresh the goals plot.

Control panel

This panel has several buttons on the left-hand side of the Cockpit that helps you to control the optimization while it is running.

Pause

Click Pause to pause a simulation while it is running. Once an optimization is paused, you can interact with the optimizer. It can take time for the optimizer to pause when the low-level simulation (for example, transient or harmonic balance) takes a while to run.
The Pause button changes to Continue when an optimization is paused or is completed.

Continue

Click Continue to continue the optimization. This button is available when the optimizer is paused or has finished.

If the optimizer has reached the maximum number of iterations and you click Continue, you can increase the maximum number of iterations.

Simulate

Click Simulate to re-evaluate the goals. This is useful after you have changed the settings (for example, modified a limit line on a goal) and you want to update the goals plots or see the new error value.
It is also useful in Simulate after pressing Simulate tuning mode. See Tuning modes for more information.

Data Display graphs do not get refreshed on clicking Simulate.

Update Design

Click Update Design to transfer variable values to the design.
This command can also transfer algorithm and goal data to the design. This is useful when you have used Edit Algorithm or Edit Goal to modify the algorithm or goal definitions. For more information on modifying the goal or algorithm settings, see #Controlling the Optimization.

Note: The options you specify while updating the design do not change from one optimization to other.

If the sweep panel is displayed (see Sweep Panel) when you click Update Design, the design is updated using the optimization specified by View.

Store

This option allows you to store the current optimization state. For more information, see Using Optimization states.

Recall

This option helps you to recall a state. For more information, see Using Optimization states.

Options

It allows you to view the Cockpit options. For more information, see #Scaling the plots.

Close

Click Close to close the Optimization Cockpit. If the optimization is still running, you have the option to stop it or to update the design.

Controlling the Optimization

Modifying the algorithm

Follow the steps below to modify the algorithm:

  1. Click Edit Algorithm.
  2. Change the optimization type or the maximum number of iterations.
  3. Click OK.
    Note: When you click OK, the Optimization component on the schematic is not updated.
  4. Click Update Design to transfer the new settings to the schematic.

Modifying goals

Follow the steps below to modify the goals definition:

  1. Click Edit Goals to view or modify the goal definitions.
  2. Goals are listed in the Goals list. Click an entry from the list to view its properties. You can change the goal's weight and also the definition of each limit line. From the cockpit, you cannot change a goal's name, expression, or analysis. Also, it doesn't allows you to add a new goal, delete an existing goal, or change the name of a limit line.

    Set Weight = 0 to disable a goal or a limit line.

  3. After changing the goals definition, Click OK.
    Note: When you click OK, the Goal components on the schematic are not changed.
  4. Click Update Design to transfer the new settings to the schematic.

Modifying variables

Follow the steps below to modify the variable settings:

  1. Click Edit Variables to view or modify the optimization variable settings.
  2. Select the check-box to enable that variable.
    The edit variable dialog box allows you to change the nominal value, format, and format's min and max values. It doesn't allows you to change a variable's name, add a new variable, or delete an existing variable.
  3. After modifying the variable settings, Click OK.
    Note: When you click OK, the variables on the schematic are not changed.
  4. Click Update Design to transfer the new settings to the schematic.

Note: Variables of +/- Delta or +/- Delta% Format are handled differently. For these variables, the minimum and maximum values for the variable are calculated as offsets from the nominal value. This calculation is performed at the beginning of the optimization and the minimum and maximum values do not change as the optimization progresses. In the above example, it seems as if Delay is defined as 3.99544 ns +/- 5%. This is not the case. At the beginning of the optimization, the nominal value for Delay was 4 ns, so the definition is actually 4 ns +/- 5%. When you edit any field for this variable and click OK or Apply, the variable's minimum and maximum are recalculated based on the value in the table; otherwise, the variable's minimum and maximum are not recalculated.

Using Optimization states

The cockpit has the ability to store and recall optimization states. This is useful when you explore the behavior of your design using the Optimization Cockpit. For example, before modifying some variable definitions or goal definitions, you can store the current optimization state and recall the saved state when required.
The optimization state contains the algorithm, goal, and variable settings.

Store

Follow the steps below to store an optimization state:

  1. Click Store
  2. Specify the name of a state in State name field
  3. Add a comment in the Comment field (optional)
  4. Click OK to store the current optimization state

The Optimization Cockpit automatically stores the initial state.

Recall

Follow the steps below to recall an optimization state:

  1. Click Recall
  2. Click a column heading to sort the states by that column
  3. Select a state
  4. Click OK or Apply to restore the selected state

When you recall a state, the cockpit's algorithm, goal, and variable settings are replaced with those of the recalled state. It is similar to manually reverting back to previous settings using Edit Algorithm, Edit Goals, and Edit Variables to restore the state. This means that the history graphs and goal graphs do not change to what they were when you saved the state.

When you recall a state the goal plots will go stale (see #Stale plots). Click Simulate to recalculate the error and update the goal graphs.

The automatically stored initial state has no Error value associated with it because the state is stored before any simulation is performed.

Note: Optimization states are discarded when the Optimization Cockpit is closed. If you want to use an optimization state in your next optimization, click Recall to recall the state and then click Update Design to update the schematic.

Scaling the plots

Click Options to view or modify the Cockpit options.

  • History graph X-axis options These options modify the X-axis scaling of all history graphs (both error and variable history graphs).
    • Display the last 10 iterations The graph displays only the 10 most recent iterations.
    • Display the last 25 iterations The graph displays only the 25 most recent iterations.
    • Display all iterations Scales the X-axis from 0 to Max. iterations.
  • History graph Y-axis options These options modify the Y-axis scaling of the error history graphs.
    • Auto-scale using visible data Consider only the iterations that are visible in the graph when scaling the Y-axis.
    • Auto-scale using all data If there are iterations that are not visible on the graph (for instance, the graph shows the last 10 iterations, but the current iteration is greater than 10) then consider the invisible iterations when scaling the Y-axis.
  • Goal plots These options modify the X-axis scaling of all the goal plots.
    • Fit to trace Set the X-axis scaling to show the entire trace.
    • Fit to limit lines Set the X-axis scaling to show all the limit lines.
    • Fit to trace and limit lines Set the X-axis scaling to show all the limit lines and the entire trace.

Note: All of these options are remembered from one optimization to the next.

Tuning

Click Start Tuning to start tuning. If the optimization is running, it will pause.

While tuning, the Start Tuning button is renamed as Stop Tuning.
To disable the tune mode:

  • Click Stop Tuning, or
  • Click Continue to resume the optimization.

You can modify the optimization variable values using the sliders or by entering values directly in the variables table.

Note: The minimum and maximum values of the variables are determined by the optimization variable definitions. The settings used by the standard tune mode are not used. To change the minimum and maximum values for a variable, click Edit Variables.

Data Display windows do not update while tuning.

Activities

  • Click a row in the variables table to highlight the corresponding slider. Click a slider to highlight the corresponding row.
  • On the keyboard, press Tab (or Enter) to move down in the variable table. Press Shift+Tab (or Shift+Enter) to move up.

Tuning menu

The Start Tuning (or Stop Tuning) button has a menu attached to it. Click the arrow to the right of the button to access the menu.

Reset values

Choose Reset Values to reset the optimization variable values to the values they had when you started tuning.

Use the Store and Recall buttons to store and recall intermediate tuning states. For more information, see #Using Optimization states.

Tuning modes

Tuning has three different modes. Use the Tuning menu to change modes.

  • Simulate while slider moves - In this mode, the simulator performs simulations while you are moving the slider. This is useful for faster simulations (less than 0.5s).
  • Simulate after each change - In this mode, the simulator waits to perform a simulation until you release the slider. This is useful when the simulations are not as quick (0.5s to 5s).
  • Simulate after pressing Simulate - In this mode, no simulations are performed until you click Simulate. This mode is useful when the simulation takes longer (greater than 5s) or you know in advance that you want to make several changes before simulating.

Note: This setting is remembered from one optimization to the next.

View-only mode

The Cockpit switches to view-only mode when either of the following is true:

  • The top-level simulation controller is not an optimization. For example, there is a ParamSweep sweeping an optimization.
  • There are two or more top-level simulation controllers. For example, there is a top-level harmonic balance controller and a top-level optimization.

In view-only mode, the cockpit still displays the real-time optimization data, but none of the interactive features is available.

Sweep Panel

The sweep panel appears when there is more than one optimization in the simulation job. This happens when an optimization is swept (using a ParamSweep component) or when there is more than one optimization in the simulation job.

Use the View drop-down list to select an optimization to be displayed by the cockpit.

Changing the view has no affect on the simulation, so you can view the results of the first point in the sweep while the rest of the sweep is running.
When you click Update Design, the design is updated using the optimization specified by View.

A red icon in the View list indicates that the optimization was terminated because of a simulation error. A green icon indicates that no simulation error occurred.

The green icon does not indicate that the error value for that optimization is zero. The error will be non-zero if the optimization was terminated for other reasons (for example, maximum iteration limit reached, gradient is zero).

Turning off the Optimization Cockpit

You can turn off the Optimization Cockpit by clearing the Enable Optimization Cockpit checkbox on the Parameters tab of the Optimization component.

Please help us improve
Please help us improve
Was this topic helpful? Yes No