During my turn to deploy I would like to analysis the best unit and best place to deploy that unit from the previous round data which has to cause least damage to me, for this what is the syntax to create an action phase simulator during that deploy phase and to grab data from that simulation ?
There is no specific syntax for an action phase simulator. You will have to make your own. There have been many forum discussions about it, in particular about optimizing it. I am not going to list them, because there are several, but if you just use the search, you can find them easily.
The big question with action phase simulators (so far) seems to be optimization, with the best being Aeldrexan’s (I think), with 9 - 15ms per simulation. I have mine down to 50 - 100ms (on my laptop, apparently there is a big difference), so there is room for improvement.
If my experience is similar to others’, you can expect making an action phase simulator to be a huge time sink (my algos have really dropped in ELO, because I haven’t been updating them recently), but it is an investment. I believe that all (or at least the majority) of the top algos use turn simulation, so if you want to hit top ten, that is probably the way to go (unless you can successfully make a good neural network, or something like that)
Edit:
I have still been looking at optimization, and I ran into a post that I had forgotten about. Aeldrexan’s turn simulation takes only 2ms in a ranked match!
5 Likes
Aeldrexan’s turn simulation takes only 2ms in a ranked match!
Indeed, but I think that @arnby’s one is even faster.
But you should also consider that you probably don’t need your simulations to be 100% exact.
3 Likes
If my experience is similar to others’, you can expect making an action phase simulator to be a huge time sink
I agree. Most of my time working on my algos is spent debugging the simulator. I would recommend setting up tests early, so that you have it correct before you try to optimize it.
1 Like