Subsector metallicity and ELW ratios - analysis

[marx]

I've been looking into subsector / boxel metallicity and ELW / systems ratios for some time, but then carriers and such put that on hold. Now, with some more data since then, I redid the analysis and finished writing things up so that I can post them now.

First off, an intro to those who might not know what exactly I might be talking about. If you're familiar with boxels, metallicity and such, feel free to skip this part.

Spoiler

So, subsectors or boxels are the smaller parts of sectors, where systems share some similarities, such as the system mass, determined by the ZAMS mass of the main star, and curiously, also the Helium percentages in gas giants. The way the galaxy is structured is sector -> boxel -> system, for example: Synuefe AR-D d12-6 would be Synuefe sector, AR-D d12 boxel, 6 system. (The mass code, which is often referenced, is the last letter in the boxel, which can go from a to h. In this case, it's d.)
For further reading, see the Sector Naming article on the DISC wiki.
As for boxel metallicity, we measure the Helium levels in gas giants then.

On to the methodology: I used the EDSM data dump from 2020/09/30, and processed metallicity and ELW data from @Orvidius 's EDAstro. I broke down two sectors into boxels, counted the ELWs and total systems in each boxel, then compared the ELW / Sys ratios of boxels to their Helium levels (min / max / average / diff). To get rid of outlier ELW / Sys ratios from boxels where there's 1 ELW and 1 system visited, I set two cut-off levels: charting only boxels with over 100 systems discovered, and only those with over 500.

As it turns out, there aren't that many sectors which have enough systems spread over enough boxels visited that it's worth analysing them. In the end, I chose two sectors: Eol Prou, for obvious reasons, and the nearby Eoch Flyuae sector. (Sectors around the bubble might also have been good, but they are carved up by all the spherical sector overrides, and it would be a pain to stitch all those together - especially when a sphere overlaps more than one sector.)

The results were somewhat surprising: contrary to what I expected, it turns out that more metallicity as in higher Helium levels lead to lower ELW ratios. This isn't true the other way around: just because a boxel has less Helium doesn't mean it'll have more Earth-likes, only that it might have more. Or it might not. Helium levels alone won't tell you that, but they can be handy to judge where not to stay.

Take a look at these charts (the rest can be found on the sheets I'll link below):
Average He% in Eol Prou, subsectors with over 100 systems, and with over 500 systems:

Or the same two in Eoch Flyuae instead:

The other charts (min / max / diff) are quite similar too.

It's worth noting how the majority of boxels tend to fall below the 0.02 ELW / Systems ratio, which would be the often-cited "1 in 50 systems". But when it comes to boxels where we had over 500 systems visited, in Eol Prou, only five of them had a better ELW / Sys ratio than that. So in boxels with, say, 150-250 systems, you might see the luck of the draw produce quite high ratios, but as you explore more, it'll even out.

So, the summary: boxels with higher Helium levels tend to have fewer ELWs per systems, but that doesn't mean boxels with lower Helium levels are going to have more ELWs per systems. What it does mean that if you want those, you'll likely want to spend more time hanging around in boxels with as low Helium levels as you can go. If you're lucky, you might find one with higher than usual chances for ELWs.


I'd be quite curious about doing the same somewhere inside the galactic core, but we could use more data there. Especially data outside the suppression corridor, and unfortunately, Sagittarius A* and Explorer's Anchorage are smack in the middle of that. There is a cap on maximum Helium levels inside the core, which is why no non-bugged HRGGs have been found there (see the HRGG map on EDastro): I'd be curious to see if this has any effect or not. However, such a survey would require a large coordinated effort - let's see, to at least match Eoch Flyuae, we'd need 30 boxels with at least 500 systems, and 300 boxels with at least 100. So that's a minimum of 30,000 systems.
Well, in any case, I'll revisit this later, and see what other sectors might be worth looking at.

Also, this is just an anecdote, but to me, while exploring boxels with higher Helium levels, it seemed like they have more gas giants, and less planets. I might look into this later too, but in the meantime, I would be interested if others have observed the same. If (if) it were true, then that might explain why such boxels look to have fewer ELWs - but I am not an astrophysicist, and I have no idea how the Forge models planet formation.


Here's all the data and the other charts then:
Eol Prou at https://docs.google.com/spreadsheets/d/1ZNvjJnPcsdHyJRrpDpM9XWA_nN8zwH58DS__bVeurao/edit?usp=sharing
and
Eoch Flyuae at https://docs.google.com/spreadsheets/d/1FCnAUKavGh2OZAq1y290lc_8XdDkRjBVhlec0R1d3tA/edit?usp=sharing

Hope this was useful to someone. Thanks for reading!

 

[Sapyx]

The results were somewhat surprising: contrary to what I expected, it turns out that more metallicity as in higher Helium levels lead to lower ELW ratios.

Actually, this would be exactly what I would expect, based on our current scientific understanding. Because there's a missing link in your correlations that can explain this: age.

"High metallicity" sectors are almost always comprised mostly of younger stars, because the metals that make them "high-metallicity" are derived from the supernova explosions of earlier generations of stars. It's impossible - or at least super-improbable - for a star out there in the real world to be both old and high-metallicity. In an attempt to be realistic, the Stellar Forge would be programmed with this bias as well. And the older a star system is, the more likely that complex life will form on its planets.As we all know, there are statistical freaks that arise from the Forge algorithms, like Earth-likes forming around young B-class stars. But the general rule of thumb would be "the older, the better".

Thus, low-metallicity star systems are likelier to have been around for long enough for complex life - e.g. Earth-like worlds - to become more probable.

This bias towards older stars having more life will also explain the observation that Earth-likes are more probable the further away from the disc you get.

 

[MattG]

Also, this is just an anecdote, but to me, while exploring boxels with higher Helium levels, it seemed like they have more gas giants, and less planets. I might look into this later too, but in the meantime, I would be interested if others have observed the same. If (if) it were true, then that might explain why such boxels look to have fewer ELWs - but I am not an astrophysicist, and I have no idea how the Forge models planet formation.

I mentioned to you on discord that the HR boxels seemed to have more bodies on average than other boxels - I think it's because of the increased number of Gas Giants, and their associated moons.

A while back, I did a full survey of a mass d boxel not too far from Colonia. I also did a survey of a much smaller boxel that had HRGGs in, and am mid-survey of another. But these are the results (excludes water giants, but they should probably be included):

Number of Systems	Number of Gas Giants	Min He%	Max HE%	GG:Sys Ratio
1972	2289	25.855827	26.605988	1.160750507
36	92	31.737577	34.958164	2.555555556
155	437	30.833794	33.819412	2.819354839

The last one is the WIP I mentioned - it's around 1000 systems total but it's already hitting a high average ratio.

(Great write-up BTW )

 

[marx]

This bias towards older stars having more life will also explain the observation that Earth-likes are more probable the further away from the disc you get.

I'll reply to the rest later, but there's something here that I wanted to ask about first. Could you give me a source for that observation? At least, I assume you meant it for Elite, but if you didn't, then never mind. I don't recall seeing any such for Elite though, and checking back on sector distribution data I have from last year, with 37 million systems, doesn't look to support this. But that was then, and I wonder if there was anything more on this since.

Hm, I think I could also do an approximate height (from the disc) of the boxels listed here, see if anything jumps out there. Will look into that later.

 

[Sapyx]

I'll reply to the rest later, but there's something here that I wanted to ask about first. Could you give me a source for that observation?

I recall seeing someone posting it here in the exploration forum some time ago, and it gave a small but noticeable pattern of increasing probability the further away from the disc you got. Can't recall who posted it but it must have been one of our stats-gatherers. @Orvidius ?

 

[Eahlstan]

Hm. The ratio between body types doesn't seem to change in a odd way here:

Extremely sparse heights are odd by nature (too small sample size).

 

[marx]

Yeah, but the problem with that graph is that it shows the numbers of Earth-likes in total, not the numbers of Earth-likes per systems at that height. So it's not the whole picture.

There was also my analysis on ELW distribution in the galaxy, see here, but that one's based on sectors - so there are 1280 lys of height lumped together there. There seems to be no such effect there either, but it's hard to tell - and there's a lot less data the farther from the galactic plane you go anyway.

I've chopped together an estimate of boxel positions for the data in the sheets here, and I don't see any such effects on those either, but I still want to check if I did the position calculations correctly or not. Even that would only be for two sectors though, but at least it would be on a better scale and a decent sample size.

But yeah, if there was analysis done on that, I'd be curious to see it, because elsewhere, I don't really see such an effect.

 

[Eahlstan]

Yeah, but the problem with that graph is that it shows the numbers of Earth-likes in total, not the numbers of Earth-likes per systems at that height. So it's not the whole picture.

Yeah, I'm aware of the limitations of this graph, but I still would expect a 'spike' in the ELW line if the chance of finding ELWs would be different.

 

[marx]

First off, I've added estimated boxel coordinates to the sheets, to see if there's anything about more ELWs / Systems farther from the galactic plane. When looking at the charts, be mindful that both Eol Prou and Eoch Flyuae are below the plane, so the higher the in-boxel Y coord, the closer the boxel is to the plane - in essence, it's reversed.
With that in mind, it doesn't look like chances are better farther from it. It seems like the decline is shallow, and in most cases, it doesn't really matter. What is curious though is that when you compare Eol Prou to Eoch Flyuae, you can tell quite well where Colonia is. (I've also included X and Z, but didn't make charts for those.) It makes perfect sense that many boxels with large enough numbers of systems uploaded would be near the inhabited area, of course.


@Sapyx : sorry for being late with my promised reply.
Thanks for the tips, checking the ages would be a good idea, and something to do in the future then.
There is one thing though: I've always been wary of saying that since something is like this in real life, we assume it'll be the same in Elite as well. That turned out to be false in the past before. Maybe it is, maybe it isn't: we need to check and verify whether the Forge's galaxy follows our Milky Way there or not.

As for ELWs and complex life, based on what the Forge generates, I find it entirely unlikely that it would check for the planet's / system's age before determining whether it can be an ELW. See: ELWs in systems that are two million years old. As far as the Forge is concerned, if the atmosphere is breathable and the gravity not too low, then it's an ELW. It would be quite fun to see some primordial Earth-likes though which aren't covered in green flora visible from space, but well, the reason all ELWs have that is sadly only because all ELW surfaces use the same template.

 

[Sapyx]

There is one thing though: I've always been wary of saying that since something is like this in real life, we assume it'll be the same in Elite as well. That turned out to be false in the past before. Maybe it is, maybe it isn't: we need to check and verify whether the Forge's galaxy follows our Milky Way there or not.

Absolutely agree there. I don't have good statistical data, only a "feel", and a few pieces of supporting anecdotal evidence:

• There was that post I recalled, where someone broke down ELW probability based on distance from the galactic plane, and found a statistically significant increasing probability. I finally found the graph I was recalling; the third graph in RUKUS' post, here in this old thread. There could well be other explanations for that graph, like "people visit more KML stars when they're closer to the plane".
• My survey seems to show that A-class stars were more likely to host terraformables than F-class stars, but less likely to host life-bearing planets; your own stats perhaps show why this is the case: there's a sharp drop-off in the likelihood of ELWs for the hotter A-class stars. I suspect that age is a factor there; since A-class stars generally have shorter lives than F-class stars, the bigger A-class stars will statistically be the youngest, and those stars that are still in existence (i.e. haven't flared out and turned into red giants or white dwarfs) simply haven't been around long enough to get enough "dice rolls" for life to become likely on their Goldilocks planets.

I don't know, but I strongly suspect that when the Stellar Forge synthesizes a star system's history it uses staged dice-rolls, based on how old a star system is. Maybe one dice-roll per million years. So if a star system is calculated to be 160 million years old, then it gets 160 dice-rolls. 160 rolls to find out if there's a "rogue planet" encounter, 160 dice rolls to see if that moon migrates too close to its planet and gets smashed into a ring system, 160 dice-rolls to see if this Goldilocks planet has life develop on it or not, that sort of thing. And if a planet gets simple life spawned on it at dice roll #120, then it only gets 40 rolls worth of biosphere evolution.

Maybe it's simpler, a simple "life? yes/no" result, with a biosphere that's then given random stats, and if the stats happen to hit the ELW window then it's an ELW.

 

[marx]

It has been a while, but I looked into the ages of systems - specifically, their main stars - and saw how that works across all systems and only ELW-bearing systems. As usual, thanks for @Orvidius for all the EDAstro stuff! I asked him to add the ages to all sheets too, because I originally looked at Eol Prou only, and wasn't quite satisfied with how that looked.
So, the point was: I built histograms of primary star age of AFGKM and NS, both of total stars and ELW-bearing systems only, to see how the two distributions compare.

The results: Almost everywhere, the two fit superbly. That means that system age doesn't increase ELW chances.
There are some small exceptions to this:

1. The youngest A class stars have far less ELWs - these are probably all the Vz stars:
and

2. G stars younger than around 3.5 billion years have somewhat lower chances for ELWs:
and

3. In a certain interval, younger K stars have considerably less ELWs - this one seems more interesting:
and

And those are pretty much all. Looks like @Sapyx's theory of system age increasing ELW odds was proven wrong. In real life, it should, but in Elite, it doesn't.

There appear to be some interesting things to the age distributions though. Notice how with GKM, there appear to be a shift around the "halfway point" (roughly 6.75 billion years). With G stars, I wonder if this comes from mass code C instead: I'll have to look into separating them into the two mass codes later on.
Also, with these and neutron stars, there are significantly many at the far end of the age scale. Most likely some different generation, it might be worth looking into too. (And then, of course, there's the obvious spike in neutron stars too.)
But more research into these should probably go into another thread, as this thread is about ELWs.


For those interested, here are all the histograms I made:

Spoiler

Class A: total: , ELWs:
Class F: total: , ELWs:
Class G: total: , ELWs:
Class K: total: , ELWs:
Class M: total: , ELWs:
Neutron Stars: total and ELWs (couldn't upload more than ten files here)

 

[Sapyx]

Excellent work, Marx. I would only disagree on one issue:

2. G stars older than around 3.5 billion years have somewhat lower chances for ELWs:
and

This looks to me, rather, that G stars - rather like K stars - younger than 3.5 billion are slightly less probable for finding ELWs. If you compare the relative heights of the first and second peaks, the first peak for ELWs is significantly smaller.

So it seems that if there is any "young stars are less lkely to have ELWs" effect, then it's quite small, and only seems to apply to younger A, G and K stars. There is certainly no gradual increase in probability with age, as I had hypothesized.

 

[marx]

Oh yeah, I mixed the two up. Corrected it now, thanks!