Stellar Forge Map Mechanics?

[Fanghur Rahl]

I’ve been trying to figure out the logic of the stellar forge for a while now. I’ve often found while exploring that the contents of each system often seem strongly correlated with that of neighbouring systems. For example, I go over a hundred systems without ever finding any earthlikes, or biologicals. And then I jump a hundred light years or so and start fining multiple systems in a row that have them, typically with very similar/sequential system names. I don’t think it’s just a coincidence either.

Can anyone explains exactly how this mechanic works?

 

[Sapyx]

The procedural generation of the galaxy is all powered by what we call "boxels". The galaxy is divided up into cubic sectors, 1280 LYs per side. Each sector is divided up into eight sets of "mass code" boxels. Boxel set h, the largest, is a single boxel that takes up the entire 1280 LY cube. Set g divides the sector into eight boxels, each of which are 640 LY per side. Set f has 4^3=64 boxels, 320 LY per side, and so on down to boxel set a, which has 2,097,152 boxels only 10 LY per side. Those code-letters are the last letter in the name eg. the "Phylurn ZJ-I d9-13" system, the mass code is "d".

Stars created within each boxel tend to have very similar properties. Mass is the primary indicator, which is why we call the letters "mass codes". For example, the boxel H set are always giant stars, black holes and so forth, while boxel A are always brown dwarfs with the occasional red dwarf.

Stars within each boxel also share the same basic properties in terms of average age (which in turn determines whether a large star will have exploded and tuned into a neutron star/black hole), metallicity (which determines the types and qualities of likely planets), and so forth.

For more details on the analysis of the boxel system, Jackie Silver's old thread is probably still the best. For a highly informative video on how the Stellar Forge algorithms create the ED galaxy, check out the hour-long Dr Ross' interview in Discovery Scanner Chapter 1.

However, your observation about biological signals on planets isn't really linked to mass codes. Biologicals tend to require certain preconditions to be met (which are often outlined in the Codex), and one of the preconditions for most biologicals is "must be near a nebula". So if you happen to be near a nebula, there will tend to be plenty of biologicals (mostly bark mounds). Other biologicals require certain star types (which is only coincidentally linked to mass codes), while brain trees require proximity to Guardian ruins.

For Earth-likes, the probability largely depends on the star type. According to my survey results, class A stars have about a 1.8% chance of finding an Earth-like, F and G stars have around a 1.6% chance, and K stars have about a 1.1% chance; every other common star class is far below 1%. So stick to A, F and G stars and you should maximize your chances of finding ELWs, but they'll still be (on average) an ELW in only one in every 60 stars. Note that's an average; you might go through hundreds of F class stars and never see an ELW, then see three in a row. Veteran ELW-hunters will have experienced both of those extremes.

 

[Valije]

For those who don't know, you can bookmark a single message within a thread. Just saying...

 

[marx]

I’ve been trying to figure out the logic of the stellar forge for a while now. I’ve often found while exploring that the contents of each system often seem strongly correlated with that of neighbouring systems. For example, I go over a hundred systems without ever finding any earthlikes, or biologicals. And then I jump a hundred light years or so and start fining multiple systems in a row that have them, typically with very similar/sequential system names. I don’t think it’s just a coincidence either.

Can anyone explains exactly how this mechanic works?

Sure, take a look at two of my guides: finding ELWs and finding NSPs, the latter of which often applies to biological signals too. (Finding bio signals is easier.)
In short, the same mass codes have similar characteristics, and the position in the galaxy also matters somewhat. For biologicals (and NSPs), the galactic region you're in is very important as well: not all biologicals can be found in every region.
You can see a compiled list of the Codex, what requirements are listed and what can be found in which region, plus what errors of it are known, in this thread.

At the end of the day though, there's no way of looking at a system from the galaxy map and knowing exactly what its contents will be. Chances for specific things can be improved, however.

However, your observation about biological signals on planets isn't really linked to mass codes. Biologicals tend to require certain preconditions to be met (which are often outlined in the Codex), and one of the preconditions for most biologicals is "must be near a nebula". So if you happen to be near a nebula, there will tend to be plenty of biologicals (mostly bark mounds). Other biologicals require certain star types (which is only coincidentally linked to mass codes), while brain trees require proximity to Guardian ruins.

Only bark mounds require nebulae, plus technically, Thargoid barnacles do as well. However, they are present around so few - and so close - nebulae that barnacles might as well not exist for deep galaxy explorers.

For Earth-likes, the probability largely depends on the star type. According to my survey results, class A stars have about a 1.8% chance of finding an Earth-like, F and G stars have around a 1.6% chance, and K stars have about a 1.1% chance; every other common star class is far below 1%. So stick to A, F and G stars and you should maximize your chances of finding ELWs, but they'll still be (on average) an ELW in only one in every 60 stars.

Hm, that's interesting. Looking at the EDSM data set, which is a sample size far larger than your survey, ELWs around A main stars are 1.59%, F are 1.53%... but G are 1.12% and K are only 0.66%. I wonder why the difference might be, when yours approached AF pretty well, but GK are curiously overrepresented in your survey.
So yeah, better to stick to A and F stars if you want maximum ELW chances.

 

[MattG]

Hm, that's interesting. Looking at the EDSM data set, which is a sample size far larger than your survey, ELWs around A main stars are 1.59%, F are 1.53%... but G are 1.12% and K are only 0.66%. I wonder why the difference might be, when yours approached AF pretty well, but GK are curiously overrepresented in your survey.
So yeah, better to stick to A and F stars if you want maximum ELW chances.

Does mass code make a difference for G/K? EDSM likely has more mass c than an ELW hunter or surveyor, bringing average down?

 

[marx]

Does mass code make a difference for G/K? EDSM likely has more mass c than an ELW hunter or surveyor, bringing average down?

Mm, I don't know. Take a look here: the overwhelming majority (94.5%) of K stars are in mass code C, and G stars are split almost evenly between codes C and D. In their case though, ELWs are more frequent around code D (G main star) than they are around code C (also G main star).
However, Sapyx's surveying by the galmap main star plot filter, not mass codes - no idea if the same 50/50 split applies there too, or not. Although if I'm reading the thread right, it's 800 systems for each star type so far, but I wonder whether 800 G main star systems would be small enough that the differences between the two mass codes would be this significant. Although given that we're talking about ELWs, it might well be.

 

[Sapyx]

Yep, my survey is by star class filter and whatever mass-codes I happen to hit in my efforts to take random-representative samples. For the 800 G-class stars surveyed, the overall c:d ratio was 340:458, with the 2 remaining stars in mass code e. I should point out that there were some severe differences in this ratio depending on the galactic region: the western arms and inner core surveys (Thraikoo and Wepooe sectors) had a c:d ratio of around 11:9, while the upper bulge and outer core surveys (Croomaa and Lychoitl) both had c:d ratios of around 5:15. Flicking between the EDAstro maps of G class mass code c and G class mass code d, I'm seeing some slight differences - the Outer Core is more pronounced on the D map, though the bulges are too.

Of the 13 ELWs I found around G class stars in those 800 systems, 4 were in class c, while 9 were in class d. That gives hit rates, broken down by class, of about 0.6% for class c and 2.0% for class d. Perhaps a small sample size, but a significant enough difference to confirm that for G class stars, mass code c are at least one-third as likely to generate an ELW. I suspect this is simply due to stellar mass: smaller stars are less likely to host an ELW, and a mass code c that's G class is likely to be down the small end of the G scale (G7 or thereabouts), making them not much bigger and brighter than a larger K class star. When you sort those 13 ELWs by star class, all but one of those four mass code c stars are at the bottom of the list.

As for why EDSM has far fewer G and K ELWs than my survey, it could simply be a combination of small sample size and "good luck" in finding a much higher than normal number of ELWs. Certainly with this small sample size, finding a single star system with a binary pair of ELWs can punch up the averages, and I found two of those in the survey, one of which was in a G class system.