nl=-99.9999999999999
gen=-0.0 gen_max=0.0
batt=-99.9999999999999 batt_min=-99.9999999999999 batt_max=0.0
shed=-0.0 spill=-1.4210854715202e-14
Decision tree: dt=false dtcase=DT_spill
nl_neg=true nl_l_batt_min=false nl_le_batt_max=true nlmbatt_le_gen_max=true
Operating constraints: op=false
gen_ok=true batt_ok=false shed_ok=true spill_ok=false balance_ok=false (deltaP=1.4210854715202e-14)
----------
I guess this is due to rounding errors in floating-point numbers :( It has been pointed out in several discussions on minizinc with floating-points and constraint solvers.
BTW, I used the following annotation for search.
solve :: float_search([nl, gen, gen_max, batt, batt_min, batt_max, shed, spill],
1e-16, most_constrained, indomain_split) satisfy;
MiniZinc compiles to an intermediate representation called `FlatZinc`.
That is then compiled by other things to the final representation.
I don't know if they can output code but:
```
C:\Program Files\MiniZinc\bin\fz-cp-sat.exe --helfull
```
...outputs a lot of help so they may be able to.
Do understand though that the middle translation is rather gnarly - at least to me. I guess if I could understand and write the OR-TOOLs or GeCODE code directly, and as efficiently as MiniZinc's DSL I might use them (at the very least I'd avoid MiniZinc's sometimes long and memory intensive compile time)..
DSL
--
You received this message because you are subscribed to the Google Groups "MiniZinc" group.
To unsubscribe from this group and stop receiving emails from it, send an email to minizinc+u...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/minizinc/db3934bf-6418-4b2f-90db-95c28fbf2f44n%40googlegroups.com.