Changes 2.43 - 2.43.1d

[EricL]

Bump.

2.43y Buddy Drop Download

[EricL]

2.43z Buddy Drop Download

[EricL]

2.43.1 Buddy Drop Download

Let the sex begin!!!!

hehe.  I crack myself up...

[EricL]

2.43.1a Buddy Drop Download

Addresses the turn cost issue.

[EricL]

2.43.1b Buddy Drop Download

A handful of performance improvements.  Otherwise identical to 2.43.1a.

[EricL]

2.43.1c Buddy Drop Download

Fixes crashing issue in 2.43.1b introduced by perf changes.

[EricL]

2.43.1d Buddy Drop Download

There's a crashing bug in here somewhere having to do with shot array compaction I think, but it doesn't happen often and I wanted to get this out before going on vacation.  

The big change here is to the crossover algorthim and general perfromance.  People experimenting with sexual reproduction should find this release much much faster.

[Testlund]

Congratulations! This drop increased performance in my sim from 10 to 18 cycles/sec!  I'm running with 1000 bots in a new evosim I just started yesterday. Just a few useless mutations so far, but hopefully I won't have to cripple my bots with too much costs just to keep performance.  

[bacillus]

I updated the wiki site, leaving only trefvelyour... trefvelmy... (because I don't know what they do) and dist (is it absolute or relative?)

[EricL]

I updated the wiki site, leaving only trefvelyour... trefvelmy... (because I don't know what they do) and dist (is it absolute or relative?)

Thanks for updating the Wiki!!!!!  I'll post separetly on the the trefvel's

dist gives the distance to a give set of coordinates from the bot's current position.

[bacillus]

Ok, thanks. I'll update.

[Trafalgar]

I noticed in the changelogs that xpos, ypos, and refxpos and refypos are relative to map size instead of in the same units as dist now.

Question: How do dx, sx, etc behave now? Are they in the same units as xpos/ypos, or the same units as dist?

Do we need to do dist(refxpos, 0) to convert refxpos, for instance, to normal units now? (And would that be slow?)

This affects the position and movement code in most of my bots, which generally have a fixed angle of facing that they maintain all the time, and thus have sx and dx on the same axes that xpos and ypos are on. For instance, they may set up to refypos - ypos, and sx to refxpos - xpos.

It also applies to the target-position-prediction on many of my bots, like basically all versions of Guardian.

[Numsgil]

The good news is that xpos, ypos, etc. etc. are only affected on large map sizes, so F1 results didn't change (at least from that).

Not sure how the dist or angle operators were affected.

[EricL]

dist, angle, xpos and ypos should all use the same units.

For field sizes above size 4, imagine the field is size 32000 X 32000, independent of it's actual dimensions.  Positions, angles and distances are calculated using those units.

dx and sx are still in the same units they have always been.  The amount of thrust they generate does not change as the field size changes.  Note comparing .sx, .dx, etc. to x,y coordinates is apples and oranges.   You cannot now nor could you before assume that thrusting a given amount will move you that amount, even on small fields.  The physics of the sim w.r.t. friction, fluid resistance, the bot's own mass, etc. make them just different things.  One is unit of distance, one is a unit of force.

[bacillus]

I still haven't found anything on the trefvel vars. Could you either post a link to where I can find this info, or fill me in here? Thanks in advance.