Executions

---

When a model file is run in BQPhy, the instance is referred to as execution.

In the BQPhy platform, you can access the page by clicking the Projects tab on the left.

Loading the model

To run an execution of the model, we must load the model file onto the platform. This is done by creating a project based on the model

• On the Projects page, click on Create Project button on the top right
• From the pop-up menu, write the name of the project and select General in the Optimization types
• Select the model from the list of approved models

Running the execution

You can run the execution by clicking on the Execute button

• It opens a pop-up where you can name the execution instance (optional) and the server you are running it in. Click Next after finishing

• The next window opens where we add the necessary parameters

  • Number of design variables: The number of continuous/binary variables required for the optimization problem
  • Population Size: The size of the population i.e. the number of chromosomes / sample solutions considered in a generation
  • Theta: The parameter in QIEO that controls the speed and accuracy of the convergence.
  • Maximum Number of Generations: The maximum number of generations the problem must run if the convergence criteria have not been met.
  • Trials : The number of trials the optimization is to be run
  • Debs Approach: Whether the model file uses Debs Penalty or not.
  • Function Tolerance: One of the two convergence criteria. The tolerance limit for the fitness value. If the change in fitness value is below the tolerance value, the problem is said to have converged wrt fitness value
  • Design Variable Tolerance: One of the convergence criteria. If the design variables change less than the tolerance value, the problem is said to have converged wrt the design variables

  The problem is said to have converged if both the function tolerance and design variable tolerance are met.

  • Type of Optimization: Whether the design variables are continuous or binary values
  • Variable Bounds (for Continuous Variables): The upper and lower bounds for the continuous variables.
    • If the bounds are the same for all variables, tick the Global checkbox and provide the bound
    • If not, the bounds are to be provided as a list of floating point numbers, the length equaling the number of decision variables

• Click Submit Execution

To understand more on the parameters, you can refer the General Optimization page of the documentation.

Execution status and results page

The executions of a project can be viewed by clicking on the project. The details of the execution can be observed by clicking on the name of the execution.

• The program shows the following status while running

  • the time-stamps for execution submission, start and end
  • The real-time status of the execution (i.e. the trial and generation the problem is currently in)

• The results window

  • The best fitness values at each generation for different trials are shown in two formats
    • Interactive chart
    • Tabular view

• The results can also be downloaded for post-processing and other tasks. The platform provides the following files

  • the csv file containing the optimal design variables for the trial
  • the csv file containing the fitness value vs generation for different trials.