Live CEOing Ep 674: Language Design in Wolfram Language [HistoricalCountries]

Okay. Hello, everyone. We are doing a design review about historical countries for Wolfram language. All right. I'm supposed to point this to a different place. Oh, wow, look at this. It's all changed. Where do I go there? I haven't used this since it changed its. Okay, so I need somebody to. Ooh, how do I set this?

There's probably a dollar variable, isn't there?

How do I set this?

Guys, I think you need to write there only current.

In here. Yes, yes, like that.

That's right. Look.

Okay, all right. And I need somebody to remind me at the end of this to switch that back, please. Okay, and so if I do something like this, is it going to work? Can we tell? Give me a test case so I can tell where I am. Give me a test thing that only works on current. Please. Told me to do something on current.

So run one of the queries in the historical country notebook.

Well, I'm seeing it. It's right here. Tell me a queries.

Try running the first one.

All right.

Okay.

Okay.

Does that prove it's the right place?

Yes.

Okay, great. Okay, let's get into action here. So give us a big overview of what we're seeing here.

In the last meeting, we discussed that given that the historical entities, like historical countries, probably had associated with each many shapes or polygons, we needed a way in order to distinguish the different kinds of uses the users could have of these polygons in the case that a user needed a specific polygon or something like that.

Okay, so. Okay, hold on. We've got. Okay, so we have geovariant, and we have various kinds of different representations of the thing with this entity, maximum expansion and so on. We're going to look at these. Okay, then what is. Could you either find a quieter place or put yourself on mute?

Thanks.

One of the things we need to decide is which of the GEO variants we're going to use for your functions.

What is GEO function?

You know, geo. Geo. Geo entities. Geo with.

Okay, there's nothing called geofunction. I. Nothing stuck by called geofunction. Okay, okay. All right. So hold on. So the first thing here is we have this idea of given a named historical country.

Which we.

Which is defined by some kind of region in space time. We have a variety of things that we can ask about that region in space time. Is that correct?

Yes.

Okay. And so there are now some of those. So this is describing a differently sculpted region in space time here. And I'm. Well, actually, this is a slice, this particular thing. So. So you Got this.

It's not slice, it's. It's a projection. Right?

Yes, you're right, you're right. Right. It's a spatial projection. So geovariant is defining spatial projections of the space, time, volume. Other things like a duration property, for example, or duration function would tell us things about the time direction. Am I making sense?

Sorry, no, I think the geovariant, I mean, this represents the maximum extent sort of stretched out as like a prism through time. Right. As opposed to like the dated polygon, which also represents a sort of 4D object, but one that you know is not just projected, if you get what I mean.

I thought this meant, you know, is there a piece of territory that ever under Roman control.

Maximum expansion is a slice. The one that I think you're referring is the one called GeoGroup.

Right.

So what I don't mean is. Right, it's. It's. Okay, it's.

It.

It comes from a slice of a particular moment in time. But the way to think about it in terms of like the geometry is that it's like you're taking that slice and you're stretching it out into like a prism that goes to infinite past and infinite future.

I see. I should say. Yeah, I get it.

Right.

So if you were to do dated of maximum expansion, even if it was a thousand years before the Gallic empire existed, you would still get its maximum expansion region. Is that correct?

Yes.

Okay, so, okay. By the way, I don't understand why duration doesn't work. It should imagine you've got this thing stretched out in space time duration is simply the time from the, you know, the maximum time from the place where there was a beginning to an end in that space time blob. Does that make sense? Does it happen. Sorry?

Does it happen that we have an object that exists intermittently?

That's a good question. As something that came and went.

Exactly. Because then duration is not well defined. Or at least it's not well defined.

What about countries that were conquered and then got back their independence?

Well, those show up as different historical countries. Right now.

Is that correct, guys? Can you confirm that?

Yes, in most cases.

Okay, let me give you an example. Let's take Austria.

Yes.

In World War II, Austria disappeared, but then it came back again. Is that the same Austria or is that a different Austria?

A different one.

Okay, fine. And that's what we've always followed. So every. So one of our axiomatic principles is there is a time. You know, there's a time constancy to anything. There is it never happens, what Jose said never happens of it disappearing and then coming back again as the same entity. Is that correct? We can presumably confirm this. We should look at the database and confirm that. Right.

We don't have anything that has multiple start dates and end dates. Right. I mean, that's the simple case. Cesar or anybody.

Yes.

So I think for our purposes, that is how we have been defining historical countries, that they don't come into existence over and over again.

Okay. It seems a little unobvious because I can imagine, you know, conqueror X wanders into some place, takes over for six months and then is expelled and the country goes back to the way it was before. Am I making sense?

Yes.

I think in general there are a lot of entities. Like if you look at the historical country list, there's a lot of, you know, we looked at, I think France last time and there's like, you know, 25 Frances because there's, you know, French kingdom from this period to this period kind of thing.

Right. But I mean, I'm just thinking practical history. Right. There are certainly plenty of cases where, you know, before the French Revolution. After the French Revolution. Before. I mean, let's take the. Well, no, I mean before. Well, that's one where it still was France. It wasn't like there was a conquering entity from outside.

What I mean to say is that we have a very low bar for breaking things up into two entities. And so if it was taken over for six months, I think already it's very normal for that to be broken up into two entities.

Although we still want it to be the case that we can say France forever, so to speak. All the things that were knitted together in France. But presumably coming to that later in this conversation.

Right. I remember last time lots of discussions of Poland, Lithuania and things like that when we got to that question.

Yes. And we don't have Gosha here, so we won't hear about Poland this time, unfortunately.

The thing is. Sorry, you have the definition of geo entities, for instance, of military conflicts inside the Roman Republic.

Why is that entity military conflict? I don't understand that.

It doesn't need to have the. It could. It works with the string. It's just.

Okay.

Something I have.

Okay, okay, fine, fine, no problem.

Okay.

Yeah. The thing is, what should be the answer to that query? Using the variants, we could return different answers.

I understand that, I understand the question. I'm still getting up to speed on what we have here. Is it all possible for you to find a quieter location? Cesar.

I would assume that the Geo entity would use the full space time thing. I mean, dated geobarity.

That's what I would assume, right? That it would basically, if it's in the space time volume, then it is that rather than a prism or anything like this. So.

The default, that's the thing that right now is called geogroup is what you're expecting.

Is that what it is?

Wait, no, I thought geogroup was. Was sort of all. It was sort of every region that was ever part of the Roman or the Gaelic Empire, whatever. But I think the idea here would be, you know, if. If there was a battle that happened in what was part of the Roman Empire, but wasn't at this particular moment part of the Roman Empire, that would not be counted. That makes sense. It's okay, right?

Is it contained in the space time volume of the Roman Republic?

Right, when the battle happened, would it have been said that it happened in the Roman Republic?

Yeah. Now is that. Do we have any representation of the space time volume? Can our geometry stuff represent that? Anybody know?

Last time we. Someone made a little demo thing a bit. It wasn't totally trivial though.

Okay. I mean, arguably it's a thing, but it's more of a larger scale system issue of being able to represent a thing in space time. Who made that demo? What was it? I'm sorry, it's been a while since I saw it. I remember seeing the demo, but I.

Don'T remember Francisco made it, only he has the code still.

Francisco, can you remind us what that was? I will look for the code here, but can you remind us what the demo was? The Demo was a 3D graphic of the two dimensional region on the earth as a function of time. Correct? Right.

So I stuck it. The slices of the polygons on top of each other sorted by Tank.

You have this in one of your notebooks from an earlier meeting.

Okay, fine.

But I remember that this actually brings up one of the broader issues of these objects in general, which is that the interpolation is the hard part. I remember that was the. I think if I remember right, in the meeting we had sort of a basic stacked thing and then later I think he made like a one with more interpolation, but it was a little bit jankier. But that's how it has to be.

Right, but the problem is, right, the question is how does this temporal. How does the time interpolation work? Because we've got, you know, there.

Right. I think the answer is. I think it's got to be. I don't think we can do Anything smarter than just sort of a step function, if you know what I mean, where it's just right, there's no interpolation.

It's like order zero.

Yeah.

And I think it has to be some interpolation. I mean it's just like. It's just like list step plot assumes a fixed value.

Right.

So if in. We assume that if the territory expanded, for example in 200 B.C. that in 199 on December 31st of. Let me not do B.C.

Because it's all do A.D. i'm going.

To do A.D. 199 A.D. okay. December 31st, 199 A.D. the territory was its smaller size. And then the data point we have is for 200 AD and therefore we will assume that on January 1, 200 AD the territory jump to the new form, but that it's. It's a hold and then jump, you know, list step plot type thing. Is that right? Is that the best we can do? I mean, it's a separate issue that might be an interesting, perhaps even crowdsourced curation effort to try and do better than that.

So I see two issues here. One is that these things tend to change abruptly in battles or something like that. So the stepwise approach may not be so bad. And the second is that to do this in generally, generally we need a mapping from boundary points to boundary points because there may be topological change islands that appear suddenly, things like that. So it's quite difficult mathematically to do general interpolation here.

Right, yeah.

Yeah. Which seems, which is why it seems like the, you know, it's like interpolation order zero, that the piecewise way is the best way to do it.

Yeah, I agree.

Right. But I mean, an interesting question is, which seems not absolutely undoable, for example, with a crowdsourced curation approach to say what battle was it that caused this proboscis of territory to change? In other words, here's a question, Here's a question. If we tried to align military conflict battles with changes of territory, have we tried doing that? What will we find?

It is a project tying up each change in historical country with either military conflict or historical event is a project that we have to the future.

Okay, have we looked at it at all? Do we know if it's doable? Do we know if typical changes of territory are associated with one of those two things?

Yes, but the alignment could be.

I would doubt that we have enough historical events to say something useful about that. Military conflict probably is a little. The data is a little more dense and we could see more of a.

Connection, but also a lot of. I mean, just historically, I mean, it depends on the era and region that you're talking about. But, like, if you're looking at something, you know, the 30 years war or something like that, the territorial changes will be from some treaty that sort of was precipitated by military conflict. But it's not clear whether you want to then say, you know. Right. Which. Which caused it. The battle of whatever or the treaty of whatever.

Fair enough. But at least if we know the treaty was signed this day, then we know that's the time when everything changes.

Oh, you're trying to increase the resolution of the.

I'm trying to find out where the.

Right. I think right now. I mean, for reference, right now, I think everything is at the resolution of a year, or like, pretty much everything is at a resolution of a year.

Right.

Everything changes on January 1st. Right.

Yeah, go ahead.

There is like, for example, in many countries, the independence is declared first and then the battles happen.

Right.

So.

Yes.

Right. Or there's also lots of sort of right. Post facto things of, like, such territory was conquered and then there was a treaty that sort of confirmed the conquests. So did the borders move when the treaty was signed or moved when. When control was seized?

I mean, look, in any war. I mean, as we can see in Ukraine, for example, territory is moving around every day and it's not even clear. You know, there are regions where it's not clear who's in charge and it's untrackable. So it seems like the only way, you know, I don't even know. For some fraction of, you know, fairly modern history, wars are usually ended by. Is it true? I mean, wars are usually ended by some kind of treaty which defines the motion of territory on a particular day. Is that a fair assessment? I have no idea.

I think that's, you know, that's been the goal of the last few hundred years.

I mean, like, I don't know, the Korean War, for example. Do you know if that end. I don't know. I mean, I don't know that famously.

Didn'T end in a treaty. There's just a ceasefire.

It didn't end.

Right? Yeah, right.

I mean. Yeah, right. Okay, so that's an example where. Okay, so like that one, where that moves to be, you know. Yeah, yeah. Okay.

So I think it's something like the Treaty of Versailles. There'd be endless pages talking about, you know, this territory goes with these people, this territory goes to these people and so on.

Right. But basically, the point is, what is the thing that we're representing? The space, time, region, what is it? Right, so what it presumably is. What is it? I mean, because, you know, it certainly is not tracking every army's progress during some campaign. But. So what is our definition of when it changes? I mean, it's obviously messy, but do we have an operational definition? Go ahead.

Well, I mean, someone who actually works in this should probably answer. But I was just gonna say, I mean, the problem is very analogous to just, you know, it's like asking where. Where country borders are because they're contested borders. And the answer is that it's kind of messy and fuzzy and I'm not sure there's really any, you know, any good answer.

Yeah, Right. Well, so then that's the best we're going to have probably that it's that, you know, there are things which are uncontroversial, where something was that way for 100 years and wasn't controversial. And everybody. That was the common course of business was that. That everybody decided that was something in, you know, Spain or something like this. And then there'll be sort of disputed places where. I wish we had a way. I mean, like, for example, you know, obviously with quantities, we have this notion of around. We don't have any such similar thing for maps. Well, anyway, we have to do what we can do. We've got certain data. Maybe we can improve that somewhat. Although we are probably, as we split those hairs, we're probably going to run into all kinds of definitional problems. Is that. What do people think about that? In other words, if we try and tighten this up.

I mean, I think last time we sort of concluded that, if I remember right, we sort of just concluded that basically there's nothing you can do that's that precise. It's all going to be kind of approximate. And so, you know, this is fine.

Yep. Okay, fair enough.

Everybody needs to know that coming in.

The door that it's going to be approximate. And we should signal that in the documentation.

Right.

And we should just make some statement. You can. Alan, you. I, I would encourage you to write some basic statement.

I'm jotting that down.

Right.

You know, historical countries are necessarily blah, blah, blah, and you can even have a little micro essay. Maybe we even should do some post somewhere that we can refer to an essay about what some of the issues are, making it clear. It's, it's, you know, to make it precise is hopeless. Okay, let's get, let's get into what we really want to Talk about here. Okay, so I just want to get some scope of what we're talking about. So is it the case that the main thing we're talking about is these various geovariants and what should be the default for which geovarian should be the default for what is that. Is that correct? Is that the primary thing we're talking about or is there something different?

That's one of the things, but probably the main one right now.

Okay, all right. Okay, fine.

So our proposal is to make the maximum expansion the default geovariant. Except in the cases where you stated. We'll get to that. Which is a different case with, you know, with existing countries right now where the, where the default is. Well, the only polygon on the last polygon. But what we've seen is that when people ask for historical countries, they usually get the first thing. The first the polygon they get is the maximum expansion. So that seems.

Okay, so one point is maximum expansions won't tile the Earth. They'll overlap. Right. We all understand that. Right.

So, you know, but even, even, even then, I mean, we don't have, you know, at one point, all of the historical countries didn't cover the Earth either.

So in a different way, I mean, there were. There were little city states and then there was desert in between.

Well, there were a lot of, you know, empty spaces or unexplored.

Exactly.

But what I'm saying is the maximum expansion is going to overlap many times.

Oh, sure.

Maybe. Here's a question though. So I mean, it seems like, Right. If you just say, you know, Gallic Empire of Polygon. I agree, you know, maximum expansion is the obvious, the obvious thing, but I think if someone does something like the Geo entity, you know, of Roman Empire comma, you know, Battle of Adrianople or something. Are you saying that. That you think should also use maximum.

No, I. No, I think we all agree that should use the space time region.

Okay, fine, fine.

Right.

When things, when you need dated things, that's a different story.

That's not dated. That's overall dates, it's asking.

Well, no, because the Battle of Adrianople is dated. I mean, I think maybe what you're saying is that if in that case the badger Battle of Adrianople is dated and so in their case you want to use the space time diagram. But if I said is Mount Vesuvius in a Geo entity of Roman Empire comma, volcanoes or something, then you'd say that then you should use the maximum extent. But really it seems like those are equivalent because it's just that Mount Vesuvius goes to the infinite past and the infinite future in the space time.

Yes, exactly.

Diagram. Right, Right.

So it either is there or it isn't there. But if it's a point in time, if it's a point in space time, like a battle, then it either is in the region or it isn't in the region, in the space time region, or it isn't in the space time region. But you're right. I mean, to ask the question, you know, was, was this piece of territory, Sicily ever in this empire, whichever empire this is, the Roman Empire or whatever, this piece of territory, Geo Group?

Not. Not even a Maximum extent, actually, but the Geo Group.

What does Geogroup mean?

Yeah, that's another thing we wanted to discuss. I'm not sure if the name Geo Group is appropriate one. Maybe we could use Union instead.

Well, let's. I don't even know what it is.

I think we talked about this a bit last time with names. A basic Geogroup is all of the different shapes together. So it's the, you know, if Rome at one point controlled this territory and at another disjoint point controlled some other piece of territory, the Geo Group would include both.

What's the difference in that? And Maximum Expansion.

Well, it may have never controlled both things at once. The Maximum Expansion is just the largest it ever was.

Is it the largest polygon in a time series? The other. You can see it as the union of all the polygons.

Oh, wait a minute. Then I'm not sure I agree with myself about what one should do. Basically you're claiming that the default polygon for the Gallic Empire is the point at which the Gallic Empire was at its greatest success, the moment of Maximum expansion.

Which is very typical. Like if you go to Wikipedia and you look up a map, if you look up any map of the Roman Empire, they always say Roman Empire, greatest extent. And it's. It's like a particular moment that they all show.

Why don't we call it Greatest Extent rather than Maximum Expansion?

Because your name usually uses Maximum Expansion.

Sorry, what uses Maximum Expansion?

The name you usually see in all these places that refer to the largest polygon is Maximum Expansion.

All right. Okay. Okay. So that is a point in time that is wherever it was, you know, 322 BC or something, it was at the height.

Right.

GeoGroup is incomprehensible as a name for the thing that you.

I agree. That's why we were suggesting perhaps Union would be a better name. Right now. Geogroup just gives you all the polygons available. And a better name would be all perhaps. Yeah, but then we don't need all the polygons available. We need all the different polygons available so that, you know, we can work with that. So maybe union is a.

Okay, well, hold on. Maybe what we could say, because the point is geovariant is a thing that's used for many purposes. So.

Indeed this is a separate discussion that we could have here, which is we have used the ovariant because the entity might be interpreted in several ways. But we could discuss whether we need a separate set of functions that are capable of handling all these various time computations. So at the moment we are trying to put it inside the geographic only functions. Yeah, be a dated Angiovanian, but it's not clear that's a good decision.

Okay, so hold on. Whichever. I would say it should be maximum historical expansion or something. You see what I'm saying to make.

It clear, because it's historical countries and we don't say historical polygon, we just call it polygon.

No, I. I understand that, but what I'm saying is this is a geovariant, which is a shared wrapper with other things.

But what would maximal expansion mean other than over time?

Well, okay, my point here is that we could then go historical union, historical reason.

It isn't even necessarily his.

Well.

Right, so maybe here's a question then. What would union mean in a geovariant other than over time? Like it seems to me that the only time when you're going to have a geovariant there first to multiple polygons, you know, where you're going to want the maximal one or the union of all of them or whatever is going to be over time.

I don't know about that. You could have all these islands or something. Include the islands or don't.

That's a different thing, right?

It is.

It is a geovariant, but it's. It's all areas which is essentially the union one versus maybe you could say maximal region or something.

Yeah, we have different geovariance for that also.

Okay, so what are the geovariance that we have right now? And then Jose is making the point that maybe we did a different space time geovariant from these geovariance.

Well, actually, I think. I think the more useful thing, which I think maybe is what Jose was saying would be to have actual functions. Because right now the point is if you ask for geovariant or if you ask for the dated polygon GEO variant, which is the time series, there's no way to take that time series and turn it into the maximal extent or turn it into the union one without writing that code yourself. I think the point would be that we could have those functions.

Right? Right. Yeah. As I say, your variant is just a sort of selector, right? You select a certain polygon or list of polygons for something.

So.

But if you want to do operations over dated things, then you're actually computing some something over that certain.

What operations can we do things right now, other than inclusion, which I think we can only do very junior versions of, is this point within the space time region. By the way, notice that all these end in area.

Yeah, that. That shouldn't be the case, but.

Well, I'm not sure it shouldn't be the case.

I. I think I would have stayed with all default map.

No, it's. No, I don't think so. I. I think this is good because I think that, that the problem is geovarint might be used for something completely different. You see what I'm saying? Like center, for instance.

What is center?

I don't know, it says.

I think it represents the point in the middle or something.

Yeah, I would guess.

Yes.

Oh.

So I mean that.

Yeah, but at the end your variant will be used to choose a point polygon or a point or whatever to work with. Right?

Yes. Okay.

Sorry, not just a polygon. Here we are discussing whether to have the cylinder in space time.

Oh yeah, yeah, yes. Except that most functions don't deal with space time. The only one that does is geo entities.

Maybe, maybe we should have separate functions.

Which I think.

Well, I wish that we had a better way. I mean this space time. Look, you know, we've got all of the stuff for spatial statistics and so on. We don't seem to have a representative from that team here, which is a shame.

But Gosia is on call in case she is needed.

Why don't we ask Gosha to join us?

Okay.

Because you know, in spatial statistics they are doing well, let's wait for Gosha to show up. But I mean, that's spatial regions and they have ways to describe spatial regions. They don't have ways to describe space time regions to my knowledge. I just don't think we should build a big piece of machinery for doing space time. Okay, let me ask you this question, Jose. In the Astro world, are we going to need space time regions there?

Not now. The next step in Astro is curves, orbits. But I haven't thought about Regions.

Okay. And we've also want that for geo.

Right?

We want curves.

Yes, the GEO trajectories.

Yeah.

Okay, so we've got two levels of this coming. We, we've got, you know, we've had points, we have static regions. Now we're about to have lines through time, you know, points through time, world lines. And then we will have, you know, eventually we're going to be at string theory so to speak or something or D brain theory or something like this. You know, we're going to have these things sweeping through time, but we're some distance away from that right now. Is that a fair assessment here?

Yeah, I think so.

So I think what we got to do right now is things that do not pretend that we're doing serious work with spacetime regions. Therefore we're going to have to stuff this into geovariance and things like that.

Right. Another point might be that I think a lot of the operations will, I mean if we get sort of efficient like geopolygon operations then a lot of these operations on space time regions will hopefully be relatively simple. I mean it'll be something like you do like a. I forget if it's time series fold, but you either get, you get the values out of the time series and compute the union of all of them that gets you basically the GEO group region. Or if you want the maximal extent you just say, you know, take the time series, get the values and then say maximal by area. Right, Stuff like that. Or geo area.

Right, but now, okay, but so okay, you're raising an important point here which is that we do not have a property that gives us the time series and we should have that.

That was dated polygon.

Is it? Oh yeah, there it is. These are incomprehensible geovariance.

Yeah, I know, I know.

Right, that's part of the discussion.

Yes, we're not sure about this geo orientated polygon because you can use just dated for that. Dated all. And then you get all the, all the time series of polygons.

Okay, let's understand what are the geovariance we're talking about.

And then there's another thing that if you use that data polygon then you'll get polygons plus dates which is a different output from what you get from other kind of geovariance where you only get polygons.

Yeah, but apparently center will only give us a single point. Right?

Well, but it's at least according.

But anyway it's a shape. But, but anyway it's a shape. The idea is that your variant will be a, a geographic object anyway.

Yes, well, but I would argue that there still has to be some geovariant because presumably the point is that like date, if you use dated comma.

All.

Right, like basically what you'd be saying is that the default geovariant, like. Okay, sorry, the default geovariant needs a name, right? Presumably the default one is the whole space time object and you can get these other ones by using other geovariants. Well, you need to have some way of naming the default one.

I don't think they're claiming that is the default because for many purposes it's not the default. I mean, if you want to just say Gaelic Empire Polygon, the effective default is this maximum expansion geovariant. I think.

Yes, right, yes. Okay, sure.

Let's go through this again for a second. Okay, point number one, what we're doing is not the ultimate solution. The ultimate solution is to genuinely have, you know, space time regions and things like that. Hey, we have Gosha here. In spatial statistics, do you have a representation of a GEO region and how does that work? And we're trying to understand what would be a conceivable representation of this space time region in our system. Do we have, I mean, we don't have space time.

Right. So in spatial statistics right now all we do, it's spatial, we don't have a connection with time.

How many dimensions can the spatial statistics be in?

So theoretically in how many the geometry supports in woofie talking about the Cartesian space, but practically up to three because we're dealing with the realistic situation mostly. And for the geo, we still consider things kind of on a map in that way. So it's two dimensional.

Okay. And the region stuff is arbitrary dimensional, isn't it?

Regions? Yes, but the question is, you know, what can you do on the region? So in theory you can specify your data, you know, with whatever, with arbitrary dimension, but then not, you know, all the functionality can be specified yet to deal with that.

Fair enough. Okay, all right, we're going to have to scuff this into GEO variants for now. And so the only question is how to name them. And I think we should go with as explicit names as we can get because I think that will prevent, you know, so we've got three things we have to name, Am I correct?

So wait, so we're going to have a different output for the geovariance in data polygons?

Yes, as we, as we do here.

Why, why, why don't just use data that.

Possibly because it seems like they're.

Two different things, right? Because data tells you, Data tells you. Given a space, you know, a space time object, what slice or what region do you want? Right, like what slice or what section do you want? As opposed to the geovariant which is telling you which space time object are we even talking about?

So the way I see it is that with dated you have a, you have a time series and then you choose and start an end date and you get the, the, the values for a property for that. For that timeline, right?

Within that section, yeah, within that section, yes.

And otherwise with geovarian for me again, it's like some sort of selector. You select which polygon or polygons you're going to work with without taking on account data or whatever.

Okay, so. Yes, okay, so look, I kind of agree with you. So maybe data polygon goes away and it becomes data to blah comma. All because we do have notion of Gallic empire. Okay, so what we're saying is Gallic Empire is really the space time thing that we can say dated of Gallic empire comma, you know, 450 AD or something like this. Right. And that, that is then in a geo thing, you know, we might say geo area of that. We might say geo whatever it is, geographics of that. And that would work. That is a particular time slice of the space time region that is the Gallic empire. Do we agree? And then it is consistent with our use of dated that this would be a time series.

Right.

Okay. All right, so then the question is if we say sort of the naked Gallic Empire polygon, then what we are saying is we default that to be. We default that by. We assume that to be the geovariant of this with maximum expansion. Okay, now we've got one more geovariant.

And, and to be clear, so data. So geovariant of that comm. Maximum expansion is sort of the same thing as writing dated of Gaelic empire comma, you know, and then whatever the date is for its maximum expansion, like those represent the same objects.

That is correct.

Right, right.

And.

Except that, except that with dated you'll get a polygon plus a date.

I don't think you do it except with dated comma, all whatever.

I think, I think.

And with your variant you only get the polygon.

I don't think you get a date with dated here, do you? I mean it may be wrong, but you don't get it.

Well, no, here's maybe the question, I think, I think we May even talk about this last time. And this relates to the thinking of dated as a slice or a, you know, whatever. If you say geo entities of dated of Gaelic empire comma all or comma, you know, 450, and then you ask for military conflicts. The question is, is it going to interpret that as, as the, the Polygon at 450 projected into all time or is it going to interpret that as just a slice?

You know, I think that it's the polygon projection at all time. If you say geo entities of this comma, battles or something. I think if you wanted that for battles on that particular date, I would have thought you would put that dated thing somewhere else. Am I wrong? Where would you put it though?

I think yes, if you put that with military conflict. Yeah, you should get all the military conflicts in that particular polygon. I'm not sure if it.

Well, but does this work then? If you want the one. Okay, so you're defining with the dated thing here, you're defining, this is the territory that is a naked polygon, which is the polygon for the gallic empire in 450 AD end of story. And so then you ask what volcanoes are in it. You know, what battles at all time, including battles in World War.

Yes.

Without intersecting the, the dates.

Yes.

So we have to go separate way to, to, to make the intersection as well.

Well, that's this potentially.

So I guess the question is, I mean, because again, I think we talked about this last time, the question is what does data do to a space time object? Right? I mean the Gaelic Empire entity we've said represents one of these space time objects. So are we saying that dated, you know, if you give it a particular date, it takes, it finds that date and it projects into all time. What happens if you ask for an interval? Right. Does it just take that interval but then on either end it will project it into all time or what?

Sorry, inevitable. For what?

Like if you give it a dated, if you said dated of gaelic empire comma 450 to 500.

Where in the geo entities thing?

Anywhere. I mean, the point is that data should represent a consistent object no matter what function you give it to.

Yes, I understand. So I claim. Well, what does that do normally? Now does that you're thinking that's a space time chunk, right?

And if that's a space time section, then dated comma 450 should be just a slice as opposed to a slice projected until time.

For example, if you request the polygon for an interval, we are returning a Time series. But if you request only one date, we return only one Polygon.

So this would then just return a time series.

Which represents the space time section.

Well, it would return a very weird time series. This would return a time series of battles that occurred at any time in the gallic empire of 450 AD and the gall. And then the Gallic empire of 500 AD. So let's say the Gallic Empire expanded and that included the position of the Battle of Waterloo in 500, but not in 450. Right.

For Geo entities, we are not yet sure what this should return. Right now I think is returning with the maximum expansion or am I.

No, I know, but that's when we have dated there. Okay. The default should be. In my opinion, GEO Entities returns the forall time, maximum expansion. In other words, if I say geo.

Right, which is right now, the GEO group or union that we suggested.

No, it's not. No, it's not. No, it's not.

Yes, it is. I mean, I think it's like.

No, because I'm saying for all time. That's right. We say military conflict here. Or what do we call it anyway? Military.

Military conflict, yes.

Why didn't it come up on the autocomplete? Was it missing in the GEO entities? Autocomplete? That's a project.

Yeah, we'll check.

Okay.

All right.

Anyway, so this here, I'm saying that what this should mean for consistency is in the region that was the maximum expansion of the Gallic Empire, what military conflicts occurred in that region at any time. Do we agree that's what.

Yes, Wait, no, we said the opposite earlier. Earlier we said that the geo entities would check for military conflicts that happened, you know, in the borders of the Gaelic empire when it existed.

Right, you're right. You're right. That's what we said for this case.

Right. I think it's useful to think of it outside of GEO entities because I think we need to know what, you know, what does the entity represent? What is the dated entity and what do the geovariant entities represent? Before we try and think about what the most useful thing is for, you know, GEO entities. Because GEO entities can have its own kind of defaults and so on. But yeah.

Sandra is asking a question here. Given a geoposition, what country does it belong to? I mean, for current countries, that is simply geo. What is it called? Is it GEO nearest or is it what? What is it?

Geo identify.

Oh yeah, okay, right, fine. So geo identify, country and then geoposition, you know, 50, 60 or something. Why didn't that work?

Identify.

Thank you. Okay, great. So that works fine. So now is it the case that we can say, do I identify historical country? Okay, what do we think that should do?

Well, shouldn't we specify a date?

Yeah. By default, the only thing we do is give a time series.

A time series of entities.

As of right now. I don't know why that isn't working. You identify of historical country and a position. Now returns a list of all the historical countries that have contained that position. And if you use a dated wrapper with one of the domains, it returns dated objects of. Maybe try with the entity wrapper in historical country.

Just like this.

Yes. That's weird.

I mean, this is a position in Europe, isn't it? Or am I getting the coordinates the wrong way around?

Try using here, just so that you know that you're not getting something in the ocean.

Yeah, I understand, but I can. I can just. Do I predict this is France? Okay, so it's a thing that should have a historical country identification. So. Okay, so there's a bug here, but what do we want it to do? We want this to return a time series. Is that a true statement or not? Or a list? No, no, no. We want. I see. What we want is a list. And how would we get.

And only if it's dated, we will get a list of dated entities.

Okay, so what would. How would we say that? We'd say dated, comma, all. Where would we put that? Around this? Or what? Is that what you mean? Okay. Would it work if I put the data around the geoposition?

It should.

Cesar, do you have some examples for this in the notebook?

Yes.

Okay. All right, but let's. Let's go back again to. All right, we're saying that this means. This thing here means space time, you know, space time containment, so to speak. Whereas Gallic Empire of Polygon. Defaults to a geovariant of maximum expansion. Is that correct? Well, so presumably if I say this. If I say dated of Gallic Empire. I mean, this would presumably just work. I don't know whether it works now. 450 AD Polygon. Does that work?

Well, no, because the entities. The date is outside of the scope of the entity. Yeah.

Gallic Empire didn't exist, you know.

Okay, when did it exist?

Is that a crisis of the third century thing? Try like, list 250. Because it doesn't have to be in a list as the date. Because isn't that an absolute time if you just give it as an integer?

No, not undated. Okay. When did the freaking Gaelic empire exist?

Try minus 50.

Oh, could you try just taking out and saying start date.

270.

Like it existed for 10 years or something?

Yes, it's during the 10th, the 3rd century crisis.

Okay, okay. What was the crisis of the third century? Or do I not want to ask?

It was the Roman Empire broke into several pieces for like a few decades.

Okay, fine. All right, so this worked basically. Now if we say this comma, what happens if we say this comma? 290, for example. Oh, for God's sake. What should that have worked?

Are you sure that that's not being interpreted objects? Yeah, yeah, because otherwise that's. That's the year 270 and the month of 280.

I see. Well, that didn't work either. But it probably should have done, right? Hello? Is there a reason why that didn't work? Let's assume that's going to work. Okay, can we do that? Can we assume that's going to work? Okay, so now, okay, so the dated thing is at that moment, except for all, in which case it's the sequence or these things, in which case it's the sequence. And then we're saying without dated in things other than GEO identity, GEO entities, if it is a property, it's, it's by default the geovariant maximum expansion and then the property. Do we agree?

Well, I would prefer if the entity represents always the same thing. So I would say unless you add something temporal, it represents something geographical. You have to opt in to get dated stuff.

But then this wouldn't work.

Right, so that would represent say all the military conflicts in the combined.

I don't think it's the right thing. Again, the thing, the preamble that I was giving is we're not going to be able to get this perfect right now because we don't have a representation of these space time regions yet. So it's going to be a little bit of a hack.

So wait, yes, so in this case you're going to get all the military conflicts in the maximum expansion of the Gallic Empire, right?

All the ones for all time in the maximum.

No, no, no, no, because we are right now using the default as the maximum expansion. Then you just, you have to use a Geo variant, GeoGroup or.

No, wait a minute. Maximum expansion does not have a date associated with it. It is simply a region of the Earth.

Right. Projected. But in this space time interpretation it's just that region projected into all time.

Right.

So I think we, we are not going to be able to make it perfectly consistent right now without having a representation of space time regions. And I think in the future that. Yeah, I mean, arguably, instead of calling it historical country, in the future we might have some. Well, it's either a property of the. I don't know, something that represents space time, region, but we're not going to get there for several years. So therefore we need a solution right now. And I think the solution right now that takes GEO entities by default without dated to have a different meaning for historical country from what it means in the pure property with polygon. I mean, we can just decide for every property what is for every property. Because for example, if we say Gaelic empire start date, it's absolutely meaningless to say the Gallic Empire is supposed to be, you know, maximum expansion Gallic empire and then the start date. That's meaningless. So really what we're saying is this thing here is for that property Polygon. I mean, one way to do it is to just define.

I didn't understand what you said. Why is it Gallic empire maximum expansion started where. No, no, no.

I'm saying the following thing. Okay. If we say that that Gallic Empire by default is geovariant of Gallic empire comma, maximum expansion, the start of that is meaningless.

Right, but the, the. The maximum span. The, the geovarian maximum expansion is only used in the, in the geofunction context.

Not only. So I, I know, but, but this is why people are asking in the case of. Okay, fine, but I'm. I'm saying if you say that. Okay, basically the point is in the definition of polygon for a historical country, it is really the maximum expansion polygon.

Right?

We don't have to say that.

Yes.

And although possibly we should just have that property in. And we should just say by default polygon means maximum expansion polygon, but we should actually introduce the property maximum expansion polygon into. Into historical country data. Okay, let's imagine that we did that. We have that as a property. Then another property is, you know, polygon time series. Okay, we all know what that means. And then the other one is the, you know, union polygon, basically. Is that correct?

Well, having different properties will be much clearer, definitely. At least from the point of view of.

Yeah, but.

But I believe that's what geovarian does, right? It chooses one or more polygons to work with.

Right, but. But the problem is that we need them also in your variant, because how do you specify that when searching like in GEO entities? That's why we introduce it also center as A possibility. Because what if we want the geo distance?

Okay, so. Okay, so listen, for geographics, we have these properties, okay, which you can explicitly specify. Geographics would default if you don't specify anything to the maximum expansion polygon and probably the Polygon property on its own, because we use Polygon for lots of different things. So we should support a Polygon property. Defaults to maximum expansion polygon.

Can I ask one other question about maximum expansion?

Do we.

Do we have or have we ever had talked about putting a property into historical country that we were saying up above? The maximum expansion has no temporal information.

But we should have a maximum.

That's the one. That's what I was looking for.

Yeah, yeah, we should have.

How do you.

How do you know otherwise what you're really looking at?

Yep. And presumably we have that data. Okay, so let's. Let's go through. So this, I think, is a reasonable approach. Now. Now we have to ask the question. Okay, so with dated, we know what we're getting there. What happens if you say dated of Gaelic empire comma, and then polygon property of that. What are you sending here, Francisco?

So we have that information of the date as an annotation.

Oh, yeah, fine. But I think we should promote it then. Let's also promote it as a maximum expansion date because it's a useful thing to know, you know, when was this?

Sure, sure.

It can be promoted to a property, right?

Fine. Okay. But now in a case like this, when we have a dated wrapper, I'm trying to understand, can we ask for the polygon property of this dated wrapper? Yes, we can. So the only issue is what does polygon mean? Okay, maybe we just shouldn't support this polygon. Okay, Polygon with no dated wrapper would be the maximum expansion polygon. If it has a dated wrapper, it's the polygon for that date. Do we agree? Yes. No.

Sorry, can you say that again?

If it has a dated wrapper, it is the polygon for that date. It has no dated wrapper. It defaults to be the maximum expansion polygon. That is the polygon for the maximum expansion date.

Yes, exactly.

Okay. All right. Okay. Now geo. Geo identify. Okay. What does it do?

So if you date it, should the output of dating always include a date?

No.

Besides the polygon?

No. No, only when it's a range.

Why the different behavior?

Because that's what we've done before. Okay. Is it perfect? No. Is it. Is it consistent? Yes.

Right.

I think the idea is just that. The default, if you give data to specific date, it gives you a specific object. If you give it a range, it gives you a time series. And the default, you know, the default for Gaelic empire is interpreted as dated of Gaelic Empire comma 270 or whatever the year when it was biggest.

Right.

A possibility that we had explored is that if the dated is in the object like a Gallic empire, then we don't return a dated. But if we have functions in which the domain is wrapped in dated, then we could return the objects wrapped in.

Dated because it provides more information with other things. I don't know if that's consistent, if it is. Great.

Well, I mean related to that though, because you were just asking about geo identifying. So arguably what geo. Because okay, geo identify could do two things. The most consistent thing it would do is if you just say geoidentify of historical country comma here is it would return all the historical countries whose maximum extents included that point. Although honestly that's kind of a useless computation.

Yeah, it should be the.

The other thing it could do should.

Be the union polygon in that case.

Well, maybe, but, but, but we can't be that inconsistent. Another thing it could do related to I think what what someone was saying earlier maybe was supposed to be doing is it could. It could interpret the domain of sort of historical countries to be, you know, all of the dated historical countries. So you imagine it's the set of dated of Gaelic Empire comma 270 and dated of Gaelic Empire comma 271 and so on. And it would return all of this sort of, you know, dated of entity comma interval objects for all the countries that ever contained that point. If that makes sense.

Where does it think I am? It thinks I'm in Illinois. That's why it was. That was before the Louisiana Purchase or something. What am I. I'm totally confused. Just explain this history to me for a second. I mean, I think this is. I'm sorry, this is. So this is now returning. Sorry, Christopher, I spaced on what you were saying. Could you. Is what you were saying.

I was basically saying it should do exactly that. Yes. Or that it should assume. I mean that maybe we could sort of interpret for the geo identify. Could interpret the domain of historical countries as sort of the domain of all of the dated historical country objects. If that makes sense. Because it just seems like otherwise the default behavior which would be finding the maximum extents just seems kind of useless.

Yeah, that's useless. I mean this is useful, but a bit hard to predict.

Well, but the idea is that we can return things that are not trivial.

Right.

So I think this is the most useful answer in the sense that it's telling you where and when.

Right.

So that's exactly what you will want to know about the historical context.

Okay, so what about geo within Q?

Yo, within Q is a computation between entities, right? Not between domains.

Okay, so if I said you within Q of Gallic empire, Comma, how come I got Gallic empire? How come I got that Gallic empire? Okay, Paris, okay. Paris is defined for all time. What the heck? What the heck? Well, okay, let's do. Give me another. Give me another place.

I don't know, I mean, is still within Q. Doing a date intersection. Maybe that's a problem? Cesar.

I'm not sure.

No, I think so. I think that it's doing. It's doing a date intersection and that's why you're getting. Not getting anything, probably.

Well, I don't understand this. I mean this particular thing here. If I say geo list plot, for example of this. Well, I got my answer. Did include London. I didn't know that. It's a lot bigger than I remember learning was. Anyway, who knows? That's presumably its maximum extent, Correct?

That's right. Yeah.

You're working with the boy. That's a weird case. Look at this thing here. You know what I think is going on there? I think that's a landfill that filled. In other words, this empire went to the ocean. But the ocean as it was then.

I'm not sure that whoever drew that polygon was thinking that.

That's right.

I think it's only. Yeah. A matter of granularity of a polygon.

Yes.

Yeah, it's more a low resolution polygon for the historical ones.

Okay. Okay. In any case, what would you say? What was it?

I think we think you should work simply.

Okay, so that's just about. Okay.

Okay for this case.

Yeah.

Okay. So do we think we've nailed what we want here? Are we happy with what we've got? I think. I think it's going to have to be worked through and I think we need to write down with some precision what. Okay, so again, geo entities here will be things in the space time region of the Roman Republic. If you want it to be within the.

See for example, in that case what you have is the intersection of dates also.

Yeah, exactly. That's what I'm saying. The space time region. Yeah, but whereas if I.

Should that be the case or should we include another wrapper so that it does that. So that everything is.

I Think we should do that? I think we should do this. And I think what we should do is were you to say, okay, so this is yet another case is. Oh, this is bizarre. Okay, so I don't even understand this. I think this is wrong. Oh no, this, this is. Oh, for goodness sake. Yeah, this is, this is really confused.

This is a somewhat weird query. It shows every country in the in 1500s that shares territory or serves borders with the.

Yeah, okay, I understand.

Roman Republican.

Yes, right. So here what I would say is if this says geo variant of maximum expansion, then that is a for all time polygon. So what I'm saying, as opposed to the. I mean maybe what we need to do is say geo variant of blah of thing comma, you know, space, time, region. Maybe that's what we should do. We should have that. And that is the effect that that is the default. Just as we say maximum expansion. Polygon is the default. If you just say polygon for one of these historical entities. So similarly inside GEO entities, this GEO variant is the default. And I don't really mind calling it space time region, which is a bit space age so to speak, because it is a more complicated concept and it's one that eventually we will cover in a serious way. Comments from people. So in other words, the default for GEO entities is space time region. The default for GEO list plot, for example, is maximum expansion polygon or maximum expansion.

So space and region would have a polygon plus a date, right?

No, it's, it's, it is, it's. No, it's not going to pull out the dates, it's going to just be which military conflicts which are space time points, right?

No, no, but you're using here geovariant.

No, I'm saying it's a geovariant that includes dates, but it doesn't return. In the case of geo identified GEO entities.

Let'S say Roman Empire, what would you.

No, the question is does it return dated of battle X comma date or does it just return battle X?

No, wait, wait, wait, wait again, Geovarian will go on to the historian countries very safe. For example, it's a selector again of the different polio. So what would you give with your variant Roman Emperor spacetime region.

I'm saying this is geovariant. Okay, what's it now we say military conflict here.

Wait, wait, wait, wait, wait. Yeah, but on the one before the only Jovarian thing, space and region, what will be your output there? A polygon.

There's nothing output there. It's Just geovaring. It just sits there. Doesn't do anything.

Right.

Okay. Okay.

Right. So in this case here, it would simply return a list of those military conflicts which lie in the space time region.

Yep.

Okay. As opposed to the following, maybe dated of this comma, all which would return some kind of time series. I don't know if that makes any sense. That may not make any sense.

So the reason why they were objects.

Like the ones below dated entity and the date.

But I think that should be controlled from the domain, not from the first argument.

You mean from the military conflict part?

Exactly. So the date that's in the next output are there because the second argument in that input has stated.

Yeah, fair enough. Okay. How do we make progress on this? Okay, I think we. Yeah, go ahead.

Right.

This is wrong. This is. This is wrong. This should be a time series.

Yes. And while developing the geodistance function, the argument was made that geodistance always returns quantity. And then we thought of making another function like geodistance time series that would return no time series.

Listen, guys, listen, listen, listen. We've got points in space. We've got points in space time. We're about to have curves, world lines and space time. We haven't got the full space time geometry that we need to represent all the things we're dealing with. Therefore, we are going to have to hack it. You see what I'm saying? So there's no way we're introducing for something like this some kind of, you know, dated geodistance. Not a chance. That's for the future. That's part of trajectories. Right in. In in Jose's orbit calculations. We're going to be able to.

One of proposals was having a geovarian dated polygon. The name may be confused. Confusing, but which gave you the time series of all the polygons. Right.

Anybody who is watching this, we. We don't know why this hung up, but we're back. Okay, go ahead. Say that again.

Okay, so 100 process above if you can you scroll all the way up maybe?

Yep. Yeah.

And then you have there geo entities geovari and Roman Republic Dated polygon. Right. And that will give you a time series of all the polygons?

No, we just made new names for this. Right, right, right.

So but what I'm saying is that we're just changing the name Dated polygon for space and regions.

Polygon time series.

No, that's a property we want to add. But that's not the name of a geovariant.

I don't really like having a geovariant that returns the, the, the time series. I think that should be a dated wrapper.

The question is what will be the polygon property of the ovarian space time region?

Right.

Because I was wondering that should be the time series because that's the closest representation we have for the spacetime region, right?

Yeah.

Well, it could be an error. It could be. It could be.

I think it's fine to have two representations of the same thing.

Right.

Because maybe at some point we can improve one of them and leave the other as the time series only.

Right?

Yes. Okay, so where are we leaving this? I mean, okay, first point is, have we made useful progress here? And do you guys think you can. I mean, you're going to have to go through all these cases because we haven't gone through all the cases here. But do you think.

I just want to make clear. So we're, we're, we're adding properties.

Yes.

Okay. And keeping the geovarians as well.

Yes.

Maximum expansion, union, et cetera. And then we're going to.

No, no, no, we're renaming them. We're renaming them.

What were the names?

Well, I just told you the names right here.

But I thought those names were. Yeah, but those are names for the properties, not properties.

Okay, so fine, so the geovariance, the corresponding geovariance would be times.

Well, I don't know.

Geovariance. The corresponding properties would be probably space, time, region. And maybe we got a better name for that time series. That might be a geovariant name. Okay, okay, okay. The polygon that. The union case is. Something like historical union or something.

Well, union is pretty clear.

I don't think it's clear. I don't think it's obvious what that means across, you know, what are union.

Can you get of polygons?

Well, many different possible unions. If there's, if there's a. We got to make it clear that it's a time union. It's a union through time.

Okay, what about maximum expansion, combined expansion and evolving expansion, some say.

Okay, so this is the term expansion. I mean, the, the existing geovariance. I'll use the word area for that.

Yes. But because this is going to be about time dependence, we could change to some.

Well, right, but maybe that becomes maximum area.

Yep.

And then this becomes combined area. Okay.

Or unified area.

I don't think unified. I mean, that's not.

Yeah, yeah, that's maybe too political.

Yeah, yeah, yeah.

Right, right, right. And time series.

The only downside of area. But maybe this applies to the existing ones as well. That it sounds a bit like it's going to return a number.

Yeah, I don't think we worry about that. We've already got that for the existing ones. But at least. At least this makes it clear. You're getting a single polygon. Single polygon. And that one we're getting a time series. That seems useful. Okay. Okay.

Right. Union area will be actually a multi polygon, but that's fine. Anyway, so it's a one shape, right?

Yeah.

Okay. What else do we have to figure out here?

No, I think we have enough to. To implement that and see where.

Let's try this, right?

Yeah, I think so.

Okay. Okay, good stuff. Great. This is very nice to see coming together. Thanks a lot. Hey, practical thing for Christopher. Let's go to the meeting. That was supposed to be 45. Okay, see you there. Thanks, everybody. See you later.

Thanks. Bye.