Print

Print


Hi Andrew,

Thanks for point this out.  Looks like we returned the wrong variable in
that function—probably a side effect of our project-wide transition from
arrays to ArrayList last year.

It's fixed now if you do a new pull, with unit tests as proof.

Siggy

On Thu, Dec 6, 2018 at 12:10 AM Andrew Lensen <[log in to unmask]>
wrote:

> ec.multiobjective.MultiObjectiveFitness has a method
> getSortedParetoFront(inds) which "Returns the Pareto Front of the provided
> Individuals, sorted by objective 0, breaking ties with objective 1, and so
> on...". My stats output however is giving me the following example:
>
> Subpop 0 front: [1.7809114456176758 2.0] [1.7809114456176758 2.0]
> [0.7325587272644043 351.0] [0.7325587272644043 351.0] [1.6044001579284668
> 4.0] [1.6044001579284668 4.0] [1.4893463850021362 5.0] [1.3715533018112183
> 7.0] [1.2939209938049316 8.0] [1.4893463850021362 5.0] [1.4187469482421875
> 6.0] [1.2939209938049316 8.0] [1.2364723682403564 9.0] [1.2364723682403564
> 9.0] [1.1888344287872314 10.0] [1.4187469482421875 6.0] [1.3715533018112183
> 7.0] [1.1888344287872314 10.0] [1.1423251628875732 11.0]
> [1.1423251628875732 11.0] [1.102521300315857 12.0] [1.0965145826339722
> 13.0] [1.0592397451400757 14.0] [0.9846906065940857 18.0]
> [1.0151090621948242 17.0] [0.9283546209335327 22.0] [0.9567434787750244
> 21.0] [0.9840019941329956 19.0] [0.9656238555908203 20.0]
> [1.0592397451400757 14.0] [1.0422323942184448 15.0] [1.0422323942184448
> 15.0] [1.0258989334106445 16.0] [0.921392023563385 26.0]
> [0.9064592123031616 27.0] [1.0258989334106445 16.0] [1.0151090621948242
> 17.0] [0.8879514932632446 37.0]!
>   [0.8969906568527222 36.0] [0.9656238555908203 20.0] [0.9567434787750244
> 21.0] [0.798882782459259 159.0] [0.8798811435699463 39.0]
> [0.8712236285209656 42.0] [0.8870922923088074 38.0] [0.792111873626709
> 162.0] [0.8798811435699463 39.0] [0.7860276103019714 178.0]
> [0.749239444732666 258.0] [0.7664241194725037 236.0]... etc.
>
> Objective 0 is clearly not sorted in ascending (or descending order) --
> any ideas?
>


-- 

Doctoral Candidate, George Mason University
http://mason.gmu.edu/~escott8/