Different design evaluation iterations on online platform

[s.elan]

Hello everyone,

I am currently using the web-based Ngene software and have noticed a difference in the number of iterations that Ngene runs for for the web-based and installed software. I ran the syntax below online and Ngene stopped its search at 35566 whereas when my supervisor ran it on his installed software it kept going, past 150k. We were wondering if perhaps my syntax is not quite right. We would appreciate any advice. Thank you!

Kind regards,
Shalini

Code: Select all

design
;alts = alt1*, alt2*, neither
;rows = 18
;block = 2
;eff = (mnl, d)
;alg = mfederov
;model:
U(alt1) =   b1.dummy[0.0001|0.0002]       * att1[2, 3, 1] 
          + b2.dummy[-0.0001|-0.0002]     * att2[6, 12, 3]   
          + b3.dummy[0.0001|0.0002]       * att3[2, 3, 1]   
          + b4.dummy[-0.0001|-0.0002]     * att4[30, 60, 5]   
/
U(alt2) =   b1 * att1
          + b2 * att2
          + b3 * att3
          + b4 * att4
/
U(neither) = asc[0]
$

[Michiel Bliemer]

This is correct. The Ngene desktop version NEVER stops, you always have to stop it manually. In the online version this is clearly not possible as it could result in scripts that are indefinitely running in the cloud. Therefore, in the online version we implemented a default convergence criterion that stops the script automatically if Ngene was not able to find an improvement in the design over a large number of iterations. In most cases, letting it run longer than this will not lead to substantially more efficient designs. But if you really want, you can change the stopping criterion, for example you could use:

;alg = mfederov(stop = total(150000 iterations))

Note that the maximum run time for a script in the online version is 10 hours.

Michiel