Suggestion: tracking of events and "invisible" mode

[osezno_]

Dan,

If we want to make a system to analyse something, you really need to stare at the screen all the time. I think it'd be great if we could somehow implement some features (some simpler than others):
1) decouple "time step" from visualization speed (I already suggested this in another post). Perhaps you could time-step 2min and visualize (render the system) every 10x steps, so that simulation is sped up without losing stability
2) Taken that to the extreme, you could run in "invisible mode", so that no images are rendered at all (but the rest of features are on, see below), one could toggle visibility on and off at any time

This is only useful if the CPU effort in rendering is significant with respect to calculating a step (which I guess is the case).

3) Ideally, one could "record" the whole system evolution. You run it (in visible or invisible mode), and it is "stored" somehow. Then you could watch it again at any speed, watching the same event from different points of view, etc. I don't know if the stored info would be manageable or not

This way, one could set-up a complex system, with many bodies and a short timestep, run it overnight, and analyse it the morning after.

4) It'd be great if there were an "event log" (perhaps there is) with info such as "object X has collided with object Y", with a timestamp (perhaps also positional information, velocity, etc)
5) I'd love to be able to plot some variables. E.g. to plot a chart of velocity of object X and Y, or distance between two objects, etc. Considering the dynamics, some beautiful curves could pop-up (à la Lissajous), especially of you could freely define x and y axes (such as acceleration vs velocity diagrams). If you could render 3-dimensional charts (play x,y,z in a grid-cube) I'm sure it could be pretty interesting. Of course, you should be able to plot several variables at once (different colors). Take into consideration that one could set-up a system, and plot the variables in invisible mode (without watching the system at all).

I think these sort of features would be useful if we are just curious about long-term stability of a system, etc and would greatly help less powerful computers. It could lead to a more "scientific" use. Also for teching purposes (for example, one could plot how total energy is constant in a system, plotting kinetic and potential energy, and alumni would see how one transforms into the other).

Additionally, regarding the "habitable zone" discussion (which I find pretty interesting), one potential feature could be to simulate bodies' temperature depending on the incoming radiation (light) from stars (I think you were already going to implement "light" as a feature) and the Stephan-Boltzman equation (you would need to input the albedo of the body). Of course, this wouldn't take into consideration greenhouse effect (perhaps there is some way), but it could be a nice first step. This variable (sort of "surface temperature") could also be tracked and plotted, etc.

I think most of us are intrigued at the possibility of habitable systems, and I guess it's a cool workstream to deepen on.

My two cents

In any case, congratulations for the program.

Regards

Osezno_

[atomic7732]

4) It'd be great if there were an "event log" (perhaps there is) with info such as "object X has collided with object Y", with a timestamp (perhaps also positional information, velocity, etc)

I like that.
It would show:
Collisions
Created objects
Edits of Info
interactions (e.g. Object 5 has changed orbits by interaction with Object 2)

Another would be showing Orbital elements(of every object) when you go in menu (you can't run the sim while you view it)

[FGFG]

...
5) I'd love to be able to plot some variables. E.g. to plot a chart of velocity of object X and Y, or distance between two objects, etc. Considering the dynamics, some beautiful curves could pop-up (à la Lissajous), especially of you could freely define x and y axes (such as acceleration vs velocity diagrams). If you could render 3-dimensional charts (play x,y,z in a grid-cube) I'm sure it could be pretty interesting. Of course, you should be able to plot several variables at once (different colors). Take into consideration that one could set-up a system, and plot the variables in invisible mode (without watching the system at all).

I think these sort of features would be useful if we are just curious about long-term stability of a system, etc and would greatly help less powerful computers. It could lead to a more "scientific" use. Also for teching purposes (for example, one could plot how total energy is constant in a system, plotting kinetic and potential energy, and alumni would see how one transforms into the other).
...

Great idea.

I like also the first and second points 

[Thot]

...
Additionally, regarding the "habitable zone" discussion (which I find pretty interesting), one potential feature could be to simulate bodies' temperature depending on the incoming radiation (light) from stars (I think you were already going to implement "light" as a feature) and the Stephan-Boltzman equation (you would need to input the albedo of the body). Of course, this wouldn't take into consideration greenhouse effect (perhaps there is some way), but it could be a nice first step. This variable (sort of "surface temperature") could also be tracked and plotted, etc.

I think most of us are intrigued at the possibility of habitable systems, and I guess it's a cool workstream to deepen on.
...

Hey I am glad to se that there really some people are interested in this idea. I thougt alot about this in the last days and especially about the basic equation for the habitable zone. (wich was actually L_sun divided by L_star)
Ther is just one problem: This equation just gives out the exact middle of the habitable zone and so we need another equation, wich shows clearly in wich exactly range about the start the temperatures are between 0°C and 100°C, so water can be liquid and life hoe w know it can form.

However your ides are looking very well and would be very useful, especially if you are designing solar systems from the scratch. I hope Dan will soon give a statement to this thread.

Greetz Thot

[Dan Dixon]

Thot:

Those are all really good ideas. I wish I was in a position to either clone myself or hire people to help me implement all of them.

Decouple "time step" from visualization speed
1 - You can kind-of do this now. If the program loses focus (switch to another window or minimize the window) it will just calculate in the background and only update the screen occasionally.

Invisible Mode
2 – Yes the CPS effort is definitely (almost always) a limiting factor since everything with mass is effecting everything else. I expect to add better support for to handle this in the future.

Recording
3 – This is a good idea, but a low priority. I’ll have to think about how this would work logistically.

Event Log
4-  This wouldn’t be too difficult to add as I already have a working log system in place already. It wouldn’t be too difficult to create an additional file that stores system events, but I do wonder how much it might be used.

Plot Data
5 - That would be awesome and you sold me with your wording of “beautiful curves”. This would be fairly time consuming (or at least it would be in order to make it user friendly and something wouldn’t require another program - Excel). Simply outputting numbers to a file that could be opened in Excel might not be too complicated, but I doubt that feature would be used very much.

The biggest problem is that I really need to stop adding features and just polish/fix the ones that already exist.   And I’m still planning on a major UI overhaul (which should make everyone happy, but will take time away from adding new features).

I’ve recorded all of these ideas on my master list. However if there’s one that you must have ASAP, I’m open to rational arguments and/or financial incentives.

[FGFG]

Event Log
4-  This wouldn’t be too difficult to add as I already have a working log system in place already. It wouldn’t be too difficult to create an additional file that stores system events, but I do wonder how much it might be used.

I needed such a feature some times... Like when you explode a planet of a system in 100 pieces or more and you want to know if they will hit another planet... It's hard to follow all those objects

[osezno_]

Thot:

Dan,

Errr, it was me...

The whole philosophy here was to be able to make "experiments" and to "dissect them". This implies trying to avoid the program from being necessarily "live" (but, rather, "playback"). I think may of us are intrigued by stability of solar systems. But to test system stability we need short time steps and long time spans... and the program currently obligues you to be present. If you create a system and leave it overnight, the morning after you find out your system has collapsed but you have no idea of why or how or when. If you could "replay", analize "events" and "plot" some variables, you could more easily understand what is going on.

Those are all really good ideas. I wish I was in a position to either clone myself or hire people to help me implement all of them.

Cloning you would be more efficient, but it might take some time yet... 

Decouple "time step" from visualization speed
1 - You can kind-of do this now. If the program loses focus (switch to another window or minimize the window) it will just calculate in the background and only update the screen occasionally.

Good idea. I will test how much it accelerates the whole thing.

Invisible Mode
2 – Yes the CPS effort is definitely (almost always) a limiting factor since everything with mass is effecting everything else. I expect to add better support for to handle this in the future.

I don't understand this answer. What does CPS stand for? My point was that invisible mode would be useful if rendering the image takes a significant portion of CPU time (vs calculating the matrices of the n-body problem). And it would only be practical if, while invisible, you could still visualize some data (eg. time, or the data sheet that appears when you click on an object)

Recording
3 – This is a good idea, but a low priority. I’ll have to think about how this would work logistically.

My concern is it might take and overwhelming amount of data for large systems and long time spans. I wpnder if you could simplify this if you only stored one step in many (say one step in 10), so that, when actually replaying, the system would interpolate the missing points. (Do you know what I mean?)

Event Log
4-  This wouldn’t be too difficult to add as I already have a working log system in place already. It wouldn’t be too difficult to create an additional file that stores system events, but I do wonder how much it might be used.

The idea would be to define what "events" are recorded in the log. For example: collisions, etc.

Plot Data
5 - That would be awesome and you sold me with your wording of “beautiful curves”. This would be fairly time consuming (or at least it would be in order to make it user friendly and something wouldn’t require another program - Excel). Simply outputting numbers to a file that could be opened in Excel might not be too complicated, but I doubt that feature would be used very much.

I'm glad I pressed the right key with Lissajous  I guess plotting of some variable could lead to amusing results. Check LHC@home project (BOINC), in which they test orbit stability for particles in the ring. If you could just add a feature "export to excel" and the program could make up an excel (or csv) type file with a number of colums you could predefine to be analysed later in Excel or Matlab it would be great.

The biggest problem is that I really need to stop adding features and just polish/fix the ones that already exist.   And I’m still planning on a major UI overhaul (which should make everyone happy, but will take time away from adding new features).

The clasical trade-off :-)

I guess this a program that will ultimately be used by aficionados that enjoy astronomy stuff and, therefore, who are more "resilient" to user interfaces. I find it fine for me. I don't miss anything regarding UI in the current program, but I would love if there were a "composer" that would allow you to set-up systems visually in a "microsoft-paint"-like interface but typing in numerical values. You would just "draw" the system, and add properties, velocity, adjust xyz position and velocity data numerically, etc. Some intermediate point between the "xml editing for machos" and the innaccuracy of setting-up the systems using the program interface.

I’ve recorded all of these ideas on my master list. However if there’s one that you must have ASAP, I’m open to rational arguments and/or financial incentives.

 

Regards and thanks for the wonderful program.

Augusto