PDT Bounds

Predictive Design Technology's web interface requires your experimental campaign to fit certain bounds, as specified below. If your experimental campaign falls outside these bounds, experts at ProtoLife will be happy to discuss custom configurations for your problem.

Bounds on the parameters that you set on the Experimental Space Definition page:

  • Number of experimental parameters: minimum / maximum1 = 2 / 20
  • Number values per parameter: minimum / maximum = 2 / 20
  • Number of experiments in the experimental space: minimum = 100
  • Number of experiments in the experimental space (factorial): maximum = 2020
  • Number of experiments in the experimental space (mixture): maximum dynamically calculated based on how you set up the experimental space definition
  • Population size: minimum2 = 10 or = number of experimental parameters, whichever is largest
  • Population size: maximum = 2500 or = 1/3 of the total number of experiments in the experimental space
  • Number of replicates3: minimum / maximum = 0 / 10

Bounds on the (optional) initial experiments4 that you upload on the Initial Experiments page:

  • Number of missing response values: no missing responses allowed
  • Number of initial experiments belonging to the experimental space, not including replicates: maximum = 2500 or = 1/3 of the number of experiments in the experimental space, whichever is smallest
  • Number of initial experiments belonging to the experimental space, including replicates: maximum = 5000 or = 2/3 of the number of experiments in the experimental space, whichever is smallest
  • Number of initial experiments belonging to the expanded6 experimental space but not to the experimental space, not including replicates: maximum = 2500 or = 1/3 of the number of experiments in the experimental space, whichever is smallest
  • Number of initial experiments belonging to the expanded experimental space but not to the experimental space, including replicates: maximum = 5000 or = 2/3 of the number of experiments in the experimental space, whichever is smallest

Bounds on the PDT-design experiments and (optional) extra experiments4 that you upload on the Generation N Experiments (N = 1, 2, ...) page:

  • Number of missing response values for PDT-designed experiments: max = 50% of the population size5
  • Number of extra experiments belonging to the experimental space, not including replicates: maximum = 2500 or = 1/3 of the number of untried experiments in the experimental space, whichever is smallest
  • Number of extra experiments belonging to the experimental space, including replicates: maximum = 5000 or = 2/3 of the number of untried experiments in the experimental space, whichever is smallest
  • Number of extra experiments belonging to the expanded6 experimental space but not to the experimental space, not including replicates: maximum = 2500
  • Number of extra experiments belonging to the expanded experimental space but not to the experimental space, including replicates: maximum = 5000
  • Number of missing response values for extra experiments: no missing responses allowed

Bounds on the total number of experiments tried in an experimental campaign:

  • Number of initial + PDT-designed + extra experiments, not including replicates: maximum = 20000
  • Number of initial + PDT-designed + extra experiments, including replicates: maximum = 40000

  1. If the experimental space includes categorical parameters, the maximum number of experimental parameters will be dynamically calculated based on the number of values you specify for those parameters, and may be smaller than 20.

  2. A minimum amount of experimental results is necessary to build accurate predictive models from the beginning of an experimental campaign.

  3. PDT will average out replicate response measurements to estimate the "true" response of each experiment.

  4. In mixture experimental spaces, all initial and extra experiments must satisfy the mixture constraint.

  5. A PDT-designed experiment is considered to have 'missing response value' if no response measurement is available for that experiment, nor for any of its replicates.

  6. The expanded experimental space is the same as the experimental space, except it allows any values for the experimental parameters. In an expanded mixture space, however, the experimental parameter values must satisfy the constraint on the total number of units specified for the corresponding mixture space.