Aurora 4x

Other Games => Pulsar 4x => Topic started by: SwordLord10 on April 22, 2016, 08:51:24 AM

Title: Newtonian Pulsar
Post by: SwordLord10 on April 22, 2016, 08:51:24 AM
This may just be me, but I was quite excited to hear about Newtonian Aurora, and it has been on hold for a while, which disappoints me. I saw that this game was toying with Newtonian ships, and I think that this game would benefit from making the Newtonian version now, which would draw support from new players(easier to learn), and from the niche of aurora players who want the most realistic experience(me)
Title: Re: Newtonian Pulsar
Post by: Hamof on April 22, 2016, 09:04:06 AM
Quote from: SwordLord10 link=topic=8583. msg90107#msg90107 date=1461333084
This may just be me, but I was quite excited to hear about Newtonian Aurora, and it has been on hold for a while, which disappoints me.  I saw that this game was toying with Newtonian ships, and I think that this game would benefit from making the Newtonian version now, which would draw support from new players(easier to learn), and from the niche of aurora players who want the most realistic experience(me)

This sounds like something for later, when this game isn't still in early-development.
Title: Re: Newtonian Pulsar
Post by: SwordLord10 on April 22, 2016, 10:00:23 AM
This sounds like something for later, when this game isn't still in early-development.
I know that is probably how it will be done, but I can dream
Title: Re: Newtonian Pulsar
Post by: Nathan_ on April 23, 2016, 02:35:20 PM
Newtonian has been shelved for a long time, and will almost certainly never be done. As far as being easier to learn, I very much doubt that, the current model of using brute engine force to get to some place interesting is vastly simpler than having to do math for every ship. If you're interested though, there are several games like IFH, KSP, and what have you that do make an effort to have realistic physics for ships.
Title: Re: Newtonian Pulsar
Post by: MarcAFK on April 23, 2016, 05:50:16 PM
Meanwhile whenever I mod non Newtonian engines into KSP everything just explodes from g forces. Need to find some way to mod out those pesky laws of physics.
Title: Re: Newtonian Pulsar
Post by: exdeathbr on April 26, 2016, 12:59:55 PM
If someone want to know the rules that newtonian aurora would have, check here.
http://aurora2.pentarch.org/index.php?topic=4329.0
Title: Re: Newtonian Pulsar
Post by: se5a on April 26, 2016, 05:16:58 PM
I would love to do newtonian pulsar, unfortunately is *really hard* to do right, and right now we have to triage.
Maybe after we've got some shooty things and something that could be considered a 'game', I'll look at it again.
Title: Re: Newtonian Pulsar
Post by: alex_brunius on April 27, 2016, 10:09:54 AM
I guess I am the only one that want to see a hybrid game where both Newtonian and trans-Newtonian propulsion is available?

That would give a whole new meaning to "conventional start"...  :) 

( and could be balanced so things that are not as time/speed critical is often much cheaper/efficient to transport the old fashion way during all the lower tech levels of the game )
Title: Re: Newtonian Pulsar
Post by: sublight on April 28, 2016, 03:04:09 PM
I've continued to poke around with Newtonian models ever since system incompatibilities forced me out of active Pulsar development.

Tracking Newtonian motion is easy enough. The tricky part is supporting 'Plot Interception' commands. The only two ways I can find to work that out are to either use brute force iterations or else to simply the problem by ignoring gravity.
Title: Re: Newtonian Pulsar
Post by: se5a on April 29, 2016, 05:14:29 AM
...ever since system incompatibilities forced me out of active Pulsar development.

What system incompatibilities? openGL? the version in the crossplatform branch does not currently use openGL, instead it uses eto.forms version of canvas. similar to how aurora does it.
Title: Re: Newtonian Pulsar
Post by: sublight on May 02, 2016, 07:59:23 AM
What system incompatibilities? openGL? the version in the crossplatform branch does not currently use openGL, instead it uses eto.forms version of canvas. similar to how aurora does it.

Basically. I also discovered a few undocumented WinForm/OSX incompatibilities, but OpenGL was the real deal breaker. In theory my hardware supported the minimum version used by Pulsar but there were problems in practice. For a while I tried to independently develop a GTK/Cairo user interface but I never did get the graphics side of that project to work properly.

Glad to hear about crossplatform progress. It sounds like I may need to try jumping back in.  :)
Title: Re: Newtonian Pulsar
Post by: se5a on May 04, 2016, 08:05:30 PM
Yeah have  look at at the cross platform branch which uses Eto.Forms.
We need someone who's osx to try see if they can get that working.

Title: Re: Newtonian Pulsar
Post by: se5a on May 05, 2016, 04:13:40 AM
...The tricky part is supporting 'Plot Interception' commands. The only two ways I can find to work that out are to either use brute force iterations or else to simply the problem by ignoring gravity.

have any of the KSP mods solved that? Mechjeb maybe? it's been a while since I've looked.
Title: Re: Newtonian Pulsar
Post by: sublight on May 05, 2016, 02:30:15 PM
have any of the KSP mods solved that? Mechjeb maybe? it's been a while since I've looked.

Mechjeb doesn't really solve it (the user guide suggests it is limited to hofman transfers unless the user supplies depart/arrival constraints), but numerical method interception solutions are known.

The usual method is to calculate P1(t1), P2(t2), and then iterate through intersecting orbital arcs for a deltaT match... and to then repeat this calculation a couple hundred/thousand times with different t1/t2 values to build a porkchop plot you can search for an optimal dV/dT solution.

I saw one speed comparison paper (link (http://digitalcommons.calpoly.edu/cgi/viewcontent.cgi?article=1080&context=aerosp)) that suggested such plots could be generated in a fraction of a second in C++ whereas my own implementations usually take a couple minutes to run. However even a quarter second of computation when multiplied by multiple ships will trigger a heat death any time a ship changes direction and every missile or ship on an interception course tries to recalculate.
Title: Re: Newtonian Pulsar
Post by: se5a on May 05, 2016, 02:54:48 PM
most the time it'd be a single fleet on intercept, and a handful of groups of missiles, you'd only do the calculation once per fleet and once per missile group.
Title: Re: Newtonian Pulsar
Post by: exdeathbr on May 06, 2016, 02:17:11 PM
Since we are talking about newtonian stuff, is 3d space planned on the future or its out of question.
Title: Re: Newtonian Pulsar
Post by: Hamof on May 06, 2016, 02:31:32 PM
Do you want your computer to burst into flames? This is going to be resource intensive enough with only 2 dimensions to worry about.
Title: Re: Newtonian Pulsar
Post by: exdeathbr on May 06, 2016, 02:34:50 PM
Do you want your computer to burst into flames? This is going to be resource intensive enough with only 2 dimensions to worry about.

to me 3d space is a important thing of space games,
games without 3d spaces feels like a "futuristic airplanes" game in my opinion
Title: Re: Newtonian Pulsar
Post by: Hamof on May 06, 2016, 02:43:55 PM
Quote from: exdeathbr link=topic=8583. msg90663#msg90663 date=1462563290
to me 3d space is a important thing of space games,
games without 3d spaces feels like a "futuristic airplanes" game in my opinion

You do realize that the instant you have so much as two possible "levels" for things that exist at the computer has to do twice as much work? Games that incorporate 3d space usually have much less stuff in that space (Homeworld for example. ), Pulsar is going to involve a lot of very complex equations all the time, and if you add in a third dimension it will require much more processing power.  (I don't think any of us would be satisfied if they added in 3d space just to give us a depth of 2 meters. )
Title: Re: Newtonian Pulsar
Post by: Thanatos on May 06, 2016, 07:17:52 PM
I believe it is actually a quadratic increase, but do not quote me on that.

As far as resources, it's not actually that bad, unless you are trying to model physics. Orbital motion is not really part of the problem, if you use simplified solutions, but the nth body problem is a thing. However, defining a dot of some sort to orbit around a celestial body at a given distance, is not very resource intensive, 2D or 3D, neither is trying to figure this out based on a speed of the celestial body.

However, in a stable system like this, if you added another large object, and had to recalculate all the gravity effects (Which you shouldn't do)... now that's gonna be a problem.

Even Navier-Stokes needs a simplification, and it is mostly a efficient 2D solution. Navier-Stokes is the movement of gas, by the way.

TL;DR 3D space is not that bad so long as you don't use celestial mechanics, and instead use simplifications like 'orbit distance' or 'body velocity'.
Title: Re: Newtonian Pulsar
Post by: se5a on May 07, 2016, 01:36:49 AM
I'll quote myself again shall I?

Quote
I'd like a 3d system map. but definitely not in this iteration.
Newton is not getting any more of a lookin in this iteration either.

both of these are hard maths.
both of these are hard UI.

I'd also like to point out that most star systems are flat. So adding the third dimension wouldn't add as much as you'd think, other than a bit of pretty, and hard work.
Title: Re: Newtonian Pulsar
Post by: se5a on May 07, 2016, 07:07:53 PM
actually, I'll clarify that a bit more, since it keeps coming up.

Many of the time consuming calculations can be threaded, the way we've set up the ecslib should make that easy to do safely.  (the threading, not the calculations).
I'm less concerned about the processing load as I am about the programming load.

To do either would be extremely time consuming, for someone with both the mathematical background AND with the user interface experience to do in a way that wouldn't be a buggy, kludgy, annoying to play mess.
not only does it have to be not buggy, kludgy, and playable from the users perspective, it has to be clear, concise, and expandable from the coders perspective, so that somone else who comes along and wants to add a new/planned *something* feature, doesn't throw up their hands in disgust because they can't figure out what's going on. 

currently none of the regular coders have *either* the math background or the user experience background, let alone *both* to want to even attempt this.

If someone came along with both those things who wanted to attempt it, they'd still need to work with us to get the current code to a place where doing these things even makes sense.
we've still got a handful of medium sized problems to work through before we even get to the place where we can do non-newtonian in 2d (in a non buggy, fun, playable, clear, concise and expandable, we currently do have a buggy, not yet playable somewhat messy 2d implementation of colonies for example).
We need to solve those problems first. we need something simple, small, playable, and fun.
All the indi games that I've come across that were successful grew from something small, simple, and fun, then iterated on it.
Title: Re: Newtonian Pulsar
Post by: DPO on June 28, 2016, 06:00:03 AM
Found a great thread discussing the formulas involved in 3D newtonian plot interception.

hxxp: boards. straightdope. com/sdmb/showthread. php?t=16114

It comes down to differential equations, which I can't say I understand as well as I would like.
Title: Re: Newtonian Pulsar
Post by: iceball3 on June 28, 2016, 02:43:42 PM
The problem with 3D precision is that, a lot of the time, trying to view this 3d depth is going to result in a loss of precision and certainty that we originally had. Defining points in 3d space is going to be a lot more of a pain than 2d space as well, as well as making AI that can handle that and UI that can consistently meaningfully tell you all the distance information you'd need at a glance will be a lot harder in 3d space than in 2d space.
Title: Re: Newtonian Pulsar
Post by: Anubis_Prime on August 16, 2016, 10:10:49 AM
Quote from: iceball3 link=topic=8583.   msg93185#msg93185 date=1467143022
The problem with 3D precision is that, a lot of the time, trying to view this 3d depth is going to result in a loss of precision and certainty that we originally had.    Defining points in 3d space is going to be a lot more of a pain than 2d space as well, as well as making AI that can handle that and UI that can consistently meaningfully tell you all the distance information you'd need at a glance will be a lot harder in 3d space than in 2d space.   

I agree, unless we were using something like Unity it would be pretty difficult to get to a playable level of precision.    Not that it couldn't be do-able but at this stage of the project you should be more focused on the mechanics and completeness of the game and work on aesthetics later.    Or even provide ways to have the user change the look of it, much like Dwarf fortress and its multiple tilesets have done.   
Title: Re: Newtonian Pulsar
Post by: TaliesinSkye on September 02, 2016, 08:38:10 AM
What about just getting a pretty good intercept estimate rather than worrying about an optimal one?  It sounds like stopping the process early could save a great deal of computation.