PSPACE

Comp-Sci has an old tradeoff between the time one needs to run an algo, and the space required to hold relevant data for it. Database engineers, and here the term "engineer" may indeed apply, are always pondering how best to index their tables. Last year Ryan Williams researched (famous) polynomial P and PSPACE, and failing at that (like we all do) instead succeeded at time[f(n)] and memoryuse[f(n)] whether P or not. It got to the arxiv February 2025; our man did a seminar last October, and now some autodubber did a youtube so I guess it's serious.

Like experimental physicists don't really grok HEP theory, I'm not pretending I understand more than the basics of what Ryan Williams has done here. All I understand is that he might be able to cut down on RAM usage for any given algo as hasn't already been so "optimised". With the understanding that we're not saving on time.

RAM matters. Prices keep rising, which is trickling into game boxes and further giving the impression of inflation, for the average man[-child]. The engineers buying RAM are doing it for data centers, themselves becoming a political problem for their needs in energy, thermodynamic release thereof (by water, mayhap), and land-area. "Reckless data centers!" has united slopulists from Brandon Herrera to Mary Peltola, neither of whom this blog is endorsing. And the antigrowth crowd including various lawyers and other scum might be keeping compute from launching to NEO space as well.

Data centers solve more the ROM problem than RAM. Those Taylor-series algos require storing (small) arrays, as does the geode rising rather larger. If the numbers can't change, they can be offshored to some shared server. Deterministic random-number results also count as ROM-storable, although in graduation from dev to unit-testing we should be using natural noise (quantum noise seems most-scalable; here's a paper).

RAM would be used where the random-number seed expands dimensionally, to the limitless possibilities of the human imagination. We don't distribute this memory.

I'm considering here how (deterministic) random numbers can give rise to a generated maze for some game. Or to cellular-automated caverns. Fractal landscapes.

I found with some of these algos (including the geode as solution to quintics, since the geode was not in 2025 stored in any offsite server) that storing an array that is too large will, itself, cost time. This is because of the Swap File pattern in just-about any machine with an internal hard drive. If you don't tell the computer otherwise, it will use the drive as an extender to RAM. That is unpleasant for the drive and it wastes your time.

This puts us back to the good ol' days of the TRS-80 when you saved your code on tape and all RAM was hard-limited to, what, 32KB (K for kibi-, these days). Modern bithackers experience those days with stackalloc, after limiting that memory themselves to maybe just one kibi.

The Catholic state

A few days ago Paul Thompson poasted at Anxious Bench a tantrum: on the topic of race, hierarchy-construction, and Christianity. I find difficult to describe this article in any other sense. Little argument was made; we do get some definitions, but only at the end at which point it simply isn't earned. Meanwhile a Christian, Eric Aaron Castro, has offered the classic Habsburg view of hierachy-construction. This one, also a bit screedy, does offer its own definitions up front.

"This is the time of the blog that we juxtapose", as Small Dead Animals used to say. I don't believe that the two articles were aware of each other.

To link the two, race has existed since the old folk of Neander Valley. Families are good; extended families inevitable. But. The Catholic argues by Divine writ - preceding Paul - that race ceases to exist in heaven, which is glimpsed at the Altar of Christ. Likewise the Catholic argues from Queen Isabella's decision against Columbus that race does not exist in her court, either. The Admiral of her fleet enjoys the same rank as the cacique of an Haitian village, and likewise the meanest sailor and the dirtiest villager under each.

"Nationalism" from this perspective is an abomination. Nahuatl (say) can be the voice of sermons and hymns; but the celebrant priest speaks (more-usually, sings) Latin to God, be he a Tlaxcaltec or a Castilian. We can argue over whether we should be using New Testament Greek or Jesus' dialect of Aramaic, or Imperial Persian, rather than Latin. Doesn't matter. It's not like the Basques are any of the above either. Latin is nothing more or less than some link to a shared transnational culture. Gotta choose something.

We are not here to argue for or against the Catholic conception of a transnational and dyothelete Church and State in harmony. This blog has handled that ad, as they say, nauseam. We are here to argue for whether, from a Western Christian perspective, Thompson offers a rational counter. I don't think he is.

Thompson is speaking the cant of a rather different Jew than that of the son of Mary when he speaks of "oppression". He slips into critical-race cant when devolving to "discrimination", "racializing", "European colonialism and imperialism", "inequities". Thompson should know what Castro knows about what Empire meant; but Thompson elects not to cite those sources. It is all resentment, for Thompson. Likely affected, for his Anxious Bench base.

Thompson may want a dictatorship of the racialized proletariat. Christians must oppose this... and oppose the Patheos website.

The eclipse of Lu

In our 709 BC, the core of what is now called "China" was under the Zhou regime... sort of. Many duchies were autonomous and dynastic, rather like what Pourshariati says of Sasanian Iran. We've all heard of Song, where the old Shang survived. Another such was Lu. It seems Lu, at least, had shifted from the old Shang practice of oracle-bones and were doing astrology instead. On 17 July (Julian), their observatory at Khoɡ-Bu' saw an eclipse. By modern pinyin it's Qūfù.

To make a correction here, this is hardly the first eclipse in astrologic history. The Mesopotamia observed one in 763 BC, heralding (or underlining) the failed Tiglath-Pileser III in Assyria, albeit invisible to Chinese. We'll get to that. For another interjection of personal frustration, ScienceDaily saw fit to tell us only now, almost a year after the paper. Weak!

We know Lu's eclipse from two later records: the "Spring and Autumn" annals, in a time when the Zhou were pretty-much being ignored; and Jīng Fáng's still-later (78-37 BC) account. I think the latter ended up in "Book of Han". For Jing, all these events were (literal) ancient history, like the various Jewish Antiquities books were for the Romans. The former is taken more seriously given that Confucius himself will be born in Lu, and seems to have had a hand in preserving their annal. But the Hans' Shu adds detail of two ears around the lunar-disc. A Japanese-led quintet, including a Meng Jin presumed Chinese, believes the latter was corona.

Although everyone has accepted the date (-708\07\17) and the place (old Lu... somewhere) of this eclipse, discrepancies persisted about the exact time and place of the Qufu observatory. It seems we've pinned down that capital, correcting other estimates that it was eight kilometers away (later dukes may have moved house). On the assumption the ducal astrologers were close to the palace, the paper must adjust ΔT to run its totality over the new site. This further adjusts ΔT for the later eclipses over Lu until ~548 BC.

For the sun himself, all this validates a maximum around 709 BC, perhaps on par with 18 October 1848. This particular eclipse, of course, was invisible to the astrologers of Babylon; but as I've noted, they had others. Coronae might not have been observable for what eclipses they had 808-717 BC which is tagged the "Neo-Assyrian Minimum". Which mostly it was... for said Assyrians, between Shamshi-Adad V and Pulu that second "Tiglath-Pileser III".

Gravettian migrations in the Upper Palaeolithic

Gelabert et al. have a new paper, on the evolution-or-replacement of the Aurignacian culture in Iberia by Gravettian. They've used sedimentary DNA, which also gets laid down in layers, easing our ability to set dates. The dates for the new remains, themselves late in Gravettian-even-Magdalenian ages, point to an event in the millennium 31500-30500 BC.

Among the finds is a female, or at least her maternal line, with mito-DNA U8a. This is presently rare outside northern Iberia, which is Basque-country today. U8a is elsewhere known in famous Bacho Kiro also deemed Aurignacian. Otherwise there is continuity.

This suggests the local Aurignacians, called here the "Fournol cluster" exemplified by "Goyet Q116-1" in Belgium, responded to an incursion of the central Europeans. The latter are the "Věstonice" descended from Bacho Kiro, but not itself well-documented. Together they made, or brought, the Gravettian in/to Aurignacian Iberia.

The Věstonice by including their women was indeed a migration. I don't know if we can call it a forced invasion; populations may have been small enough, the newcomers culturally-compatible enough, and resources plentiful-enough that a merger was possible. The paper believes Věstonice itself had been a merger of Bacho Kiro and some western population. So a back-migration and assimilation seems a better bet than if the incomers had not brought their women; or if the locals had simply vanished.

The Isturitz seems also to have the widest breeding-pool, inferred to 265. Their kids should have been reasonably healthy.

Which brings us to the south Italian sample, a sad one. This is miscarried "Ostuni1b", so called because he is, or would have been, the son of Ostuni1 previously-published: mitochondrion M (like me; no longer common in Euroland). These two samples allow for narrower dates in the middle 25kBC millennium; and gets us about half Ostun1b's father, who had yDNA in the C family (also no longer common here). This is, further, flagged as about the second-least inbred Gravettian: population size inferred to 160 [breeders].

Ousiometrics

Human language - left on its own - has a positive bias, sometimes called the "Pollyanna principle". A fair example is the indo-european *arktos, which we Germanics call a "bruin". Last May, Vermont has clarified that this bias is safety.

Nobody should be surprised that animal thought should gear toward survival, balancing "can I eat this" against "will it eat me" (per Mimic). That our communication should be aimed specifically toward warning, suggests the Prairie Dog.

NAMD

Here is another mathematic constant for you, dear readers: Turrini-Zinzi-Belinchon's normalized angular momentum deficit (2020). This takes a solar-system for which we may account for every planet in it: semimajors and masses (as fractions of the stellar constant GM) - only (for now). We calculate that ideal's whole angular momentum. Then we consider the eccentricities and inclinations and run the calculations again. The eccentricity and inclination, which we'll remember from Kozai von Zeipel oscillate, turn out to lower the momentum of the total. Therefore the deficit reflects how messy the system.

This works for systems where we own those statistics, of course.

Mostly they'll be transits such that multiple planets here will not be mutually inclined. Those will be tight orbits. Einstein starts wrangling Newton for Mercury's orbit. Further-in, tides should be raised on the star, which where eccentric Kepler teaches to undercut the planet's momentum from ideal i.e. increase the deficit. If the planets be resonant, their mutual planetary (and, internally, tidal) interactions are non-Keplerian; making "eccentricity" into a moving target. Rather, we see from the Jovian system that a resonant system should lose energy: here through Io, literally disintegrating over time and radiating heat.

On the high end, sometimes - for systems closer to ours, or very young - we can image the planets directly. Or, the dynamics have been studied to stick some constraints upon eccentricity and inclination both. Which together lower a ceiling upon momentum.

Their map marks our system's NAMD to two thousandths: 0.002, or 2 x 10^-3. "Thanks" to Velikovsky, others have done enough dynamics to declare this as longterm stable, although we'll not live to see what Gliese 710 shall do unto us.

Turrini, Zinzi and more - Polychroni and Danielski were media-available - have posted a followup from their model. Evan Gough (h/t toughsf) summarises: 0.002 is a mean and median for sub-G stars with their outermost planet ("Pout") rolling around in less than a Venereal year. Our system's longest year, for such as matter, is Neptune's at 1.648 x 10² years.

It is time we must consider that stellar mass. "Pout" year, NAMD, and this mass all correlate. Pout and starmass should indeed correlate - with age. Those smaller systems (like GJ 710) brush against larger (like us) with the outermost former planets being less-bound. But.

I mentioned that our system has a Neptune. Such giants are sometimes visible for us in younger systems, still emitting infrared maybe even still red. High-mass systems are all younger. Up to now I could imagine little reason, oh, a Neptune-mass at a Saturn-like decades-long orbit cannot exist around a midage K. We're simply unlikely to find them soon (for GJ 710 I suppose we'll find out). So we could well be missing available data for Gs and Ks.

Danielski predicts that sliding some older K systems now Pout ~0.5 over to Pout ~5 won't budge their models: selection effects acted as the tip of the iceberg, revealing the existence of much deeper structural diversities between these systems. To that, I see one smaller star with a Jovianlike Pout - ~8 years - and its NAMD is high. One suspects it already did get tugged at by a flyby but retained that planet in a wonky orbit. So I may need to recalibrate my imagination.

Polychroni pipes in for the higher-masses: The greater abundance of material from which planets form around their host stars causes them to be born too large, preventing the system from remaining stable.

For skiffy worldbuilders, this suggests constraints on K "goldilocks" stars' systems. Where they have an ice giant or a super earth outside the HZ, it either formed wonky or else got wonked by "now" (4.5 Gy after first carbon 'stroids). Such has already scattered viable mass in the HZ into uninhabitability. I get the impression this happens a lot. So your constructed K system shouldn't have long-period superearths-on-up. All this, perhaps, could have saved poor old van de Kamp some time.

Small reactors have problems

This blog was jumping with my joy when I first heard we were to approve some modular fission-reactors, in the runup to Trump 1. Six years later, Prof. Sabine Hossenfelder, (now) sponsored by the IEEE, is delivering a report-card. For a third of the advertised gains.

She would be going short. - if she were allowed to invest, which journalists shouldn't be (I know, stop laughing).

So far the record shows that the main investors today, as they were in the 1950s, aren't plebs. In the past six years, when a nuclear company has sought non-government investments, the monies aren't enough to float payroll. If Amazon or other data-facilities be ordering these, they're subsidised. The 1990s called and are calling it corporate-welfare.

Her source is Marisol Garrouste, Brendan Kochunas, Max Vanatta & Michael T. Craig from August. As well-known, economies of scale apply to nuclear physics. Small reactors produce less energy per the waste generated; this isn't like stacking rocket-engines for additional thrust. And every reactor requires its own slate of safety and environmental reviews.

We might dodge some environmental concern by using reactors in the middle of Nevada, or the vast outbacks of Canada or of Australia, or isolated mining valleys. Such could bypass some of that since they sometimes figure that revenue and labour trump already-remote and -barren environments. "First-Nation" sovereign lands with their own laws might get in on this one.

Given the above, Garrouste's quartet actually consider/s SMR a good bet if they are only delivering high heat, directly. So they could be used in endothermic smelting near those mines. They also might be used in breaking hydrogen off of water, but that too is a government project dependent upon "green energy goals". Where they don't much help, at their small scale, is in electricity generation, even to supply the remote miners' towns.

To check in on SpaceNukes: they're selling to a prototype rocket and to a prototype waterbearer. It seems however that most of their income is coming out of DARPA.