topic Re: Custom DOE does not use the whole range of discrete numeric values in Discussions

Custom DOE does not use the whole range of discrete numeric values

mvanderaa1 — Thu, 08 Jun 2023 21:14:21 GMT

I am trying to create a custom DOE with 1 continuous factor and 2 discrete numeric factors (3 levels).

One of the discrete numeric factors has values 1, 3 and 5. The generated table only uses the values 3 and 5 however. This does not seem to make sense to me from a experiment design POV. Even when changing it to a 4 level factors (0,1,3,5) only 3 and 5 are used. See JMP file for details.

Is this choice logical, or is this some kind of bug?

Perhaps of interest; I am trying to use a Disallowed combinations script.

Re: Custom DOE does not use the whole range of discrete numeric values

MRB3855 — Thu, 13 Apr 2023 08:56:51 GMT

Is it the Disallowed Combinations? If Nr_Foci=1, then (Pulse_energy*Nr_Foci) is always <60. Therefore, Nr_Foci is not an option.

Re: Custom DOE does not use the whole range of discrete numeric values

mvanderaa1 — Thu, 13 Apr 2023 08:57:05 GMT

I'm replying to my own post because I might have found already where this behavior comes from. I was having trouble getting the disallowed combinations to work by specifying the restriction as a product of the factors (i.e. X1*X2<60).

Apparently this script does not work (some runs do not fulfill this condition). After separating this condition to a combination of more simple restrictions the issue with was resolved.

Nr_Foci == 3 & Pulse_energy >= 20 | Nr_Foci == 5 & Pulse_energy >= 12

So in a way my issue is solved, but of course in other cases it's not as easy to dissect such a restriction into multiple restrictions. Kindly advice what to do with this issue.

Kindly advice what to do with this

Re: Custom DOE does not use the whole range of discrete numeric values

mvanderaa1 — Thu, 13 Apr 2023 12:41:36 GMT

wow ok you are completely right. I wrote a filter for allowed combinations instead of disallowed ones.

Re: Custom DOE does not use the whole range of discrete numeric values

Phil_Kay — Thu, 13 Apr 2023 13:03:12 GMT

This is quite common. I guess it is not very intuitive to define the disallowed combinations. It feels more natural to define allowed combinations, I suppose. You might want to add something to the JMP Wish List about this.