Release Neutron Highway long range route planner

Spansh

S
CmdrMacBolan said:
I am a little baffled as well. I just went to document the error I was seeing, by plotting my current route again using FCR, but instead of the original problem, the calculated fuel is now off by 7~10 per jump, right off the bat. But don't take my word for it, plot the route and see.

Origin: Spoihaae YP-V d3-628
Destination : Sol
Hold : 5601 (1050 Services, 4092 Tritium, and 458 Misc items)
Depot : 916
Spoiler - Actual Fuel used for first 10 jumps was: 84, 84, 83, 83, 83, 83, 82, 82, 82, 82

My Eq calculates: 84, 84, 83, 83, 83, 83, 82, 82, 82, 82
Your Eq calculated (*): 84, 83, 83, 83, 83, 82, 82, 82, 82, 82
* requires your verification. In Excel I used: =ROUND(5 + (Distance * (Capacity+Depot+25000)) / 200000, 0)

Fleet Carrier Router
1) If you do as the page instructs, and only include 'Capacity' (5601), we get: 91, 91, 91, 90, 90, 90, 89, 89, 89, 89
2) If we include Depot as the Eq demands (6517), we get: 94, 93, 93, 93, 92, 92, 92, 92, 91, 92

I got 0 wrong, you were off by 1 twice, and the attempts using FCR got 0 correct, and averaged +7 and +10 respectively.

I still don't rule out user error though. I've done dumber things. But here is an easily documented example, and I can't have dorked all four of them up (I hope).

Lastly, I appreciate you entertaining me on this issue, and I look forward to your feedback once you've run your own numbers. If this turns out to be user error, I'll make a donation to the server, starting at $50 and going up from there based on how stupid my mistake was.

Cheers.
Click to expand...

My router seems to have gotten your actual fuel usage spot on for the first 10 jumps. The only thing I can think of was you copied a number down wrongly somewhere when putting it in the router.
 

Attachments

  • Screenshot 2022-10-23 at 19.30.40.png
    Screenshot 2022-10-23 at 19.30.40.png
    840.1 KB · Views: 99

CmdrMacBolan

C
Neilski said:
OK, this is not what I get when I run the FC router. I instead get numbers which match the list you referred to as my equation.
See pic below. Did you have different numbers in the left-hand panel? I guess you must have done, but I can't see why they should have differed.
Further, as @gazelle suggests above, I had believed that our two equations were identical. This means that either we're wrong about that or the two Excel implementations you created differ slightly by accident, and this is creating the off-by-1 on two of the jumps.

Meanwhile, to get an excess of 7 units per jump (making the first jump use 91) I need to boost the cargo by around 2800, which isn't making any sense either.
Probably best to stop here and let you tell me how my pic below differs from yours (y)

View attachment 328838
Click to expand...
This screen shot was enough to identify the issue. I've never seen the inputs for 'Tritium in Tank' or 'Tritium in Market'. I never unchecked the 'Determine Tritium Requirements' option. It's my guess that that option adds the total estimated fuel requirements to the 'Capacity' value, which would increase the total weight, and hence fuel needed.

So who do I make that check out to? I'm thinking that 'failing to test the one and only option' rates about a 7 on the stupid scale, so I'll double my original 'bet'.
 

CmdrMacBolan

C
Spansh said:

My router seems to have gotten your actual fuel usage spot on for the first 10 jumps. The only thing I can think of was you copied a number down wrongly somewhere when putting it in the router.
Click to expand...
Confirmed. After seeing @Neilski's screenshot (#999) I knew immediately where the problem was, and it wasn't in your calculations. Once I followed his example and unchecked 'Determine Tritium Requirements', I was able to input all of the numbers, and your FCR did indeed produce the same results. (Kinda falls under the RTFM umbrella).

Sorry for causing a ruckus over nothing.
 

CmdrMacBolan

C
gazelle said:
It's the same formula:
Code: 
round(5 + (Distance * (Capacity+Depot+25000)) / 200000)
--> P == 5, Capacity+Depot == TC
P + D * (TC + 25000) / 200000
--> 1 / 200000 == 0.000005 == TU
P + D * (TC + 25000) * TU
P + D * (TC * TU + 25000 * TU)
--> 25000 * TU == 0.125 == CO
P + D * (TC * TU + CO)
The base formula would be:
View attachment 328829
You can bring the 1/8 inside the parenthesis and you get
fuelscost = round(5 + distance * (1/8 + totalcargo / (8*25000)))
now, 1/8 = 0.125 and 8*25000 = 200000 and 1/200000 is 5e-6
and we will get your formula
fuelscost = round(5 + distance * (0.125 + totalcargo * 5e-6))
Click to expand...
Reading your post, my basic mentality was like...

No way, that can't possibly be doing the same thing.
Oh, look at that!
No way, that part actually fits.
But that's...
Didn't see that coming.
Well F me.

Thank you for dumbing that down enough for even me to understand. (And that's no small achievement).
 

CmdrMacBolan

C
Neilski said:
This means that either we're wrong about that or the two Excel implementations you created differ slightly by accident, and this is creating the off-by-1 on two of the jumps.
Click to expand...
I neglected to address this part in my original reply.

I agree with your conclusion. @gazelle has already illustrated how the two equations are equivalent (ergo MUST produce the same results), so the only possible explanation is that I am feeding it an incorrect value somewhere along the line.
 

CmdrMacBolan

C
Neilski said:
Further, as @gazelle suggests above, I had believed that our two equations were identical. This means that either we're wrong about that or the two Excel implementations you created differ slightly by accident, and this is creating the off-by-1 on two of the jumps.
Click to expand...
I neglected to address this part in my original reply.

I agree with your conclusion. @gazelle has already illustrated how the two equations are equivalent (ergo MUST produce the same results), so the only possible explanation is that I am feeding it an incorrect value somewhere along the line.

Edit: Found the error. I was reading the wrong row for the cargo values, which resulted in one jump worth of fuel being excluded from the calculation, falsely reducing the calculated fuel-needed by ~0.21T. (84T * 5E-6 * 499.743 = .20989206). With that corrected, both Eqs are now generating identical values.
 

Parduz

Parduz
I'm trying to use the Galaxy Plotter to plot a route from Zandu to "Star One" (Byaa Thoi GC-D d12-0) using this T10, and i have some difficulties, i don't get if it isn't working or if i'm not understanding how it works.
  1. "Use Supercharge" is checked
  2. "Use FSD injections" is not checked 'cause the plotter will use it ALWAYS instead of keeping it as "last resource" when no stars are reachable with a "normal" jump.
  3. all other options unchecked
  4. 120 secs of allowed time (the maximum)
"guided" algorythm gives me 19 jumps, for a total of less than 1000Ly of travel
"optimistic" returns 64 jumps, for a bit less of 4000Ly of travel
"pessimistic" finally gives 870 jumps, for about 58250Ly. I don't know how much far the last star is from the destination, as the plotter don't returns this info when it cannot complete the route.
So now i open another router page, set the starting system as Star One and the destination as the last found star from the previous page, and by enabling the FSD injection it can plot a route of ~ 28 jumps, all with injection.

Now, a bunch of question/feature request:
  1. Am i doing something wrong? Why that poor results in the first two cases?
  2. If the router could use the FSD injection only as last resource it could have succeed in plotting a complete route with the minimal injections possible: there's a reason that it uses it ALWAYS instead?
  3. could we get the destination system in the last row also when the route is only partial (and see the destination also in the galaxy map)? it could be useful to have an idea about how short the plotted route is.
Thanks :)
 

Spansh

S
Parduz said:
I'm trying to use the Galaxy Plotter to plot a route from Zandu to "Star One" (Byaa Thoi GC-D d12-0) using this T10, and i have some difficulties, i don't get if it isn't working or if i'm not understanding how it works.
  1. "Use Supercharge" is checked
  2. "Use FSD injections" is not checked 'cause the plotter will use it ALWAYS instead of keeping it as "last resource" when no stars are reachable with a "normal" jump.
  3. all other options unchecked
  4. 120 secs of allowed time (the maximum)
"guided" algorythm gives me 19 jumps, for a total of less than 1000Ly of travel
"optimistic" returns 64 jumps, for a bit less of 4000Ly of travel
"pessimistic" finally gives 870 jumps, for about 58250Ly. I don't know how much far the last star is from the destination, as the plotter don't returns this info when it cannot complete the route.
So now i open another router page, set the starting system as Star One and the destination as the last found star from the previous page, and by enabling the FSD injection it can plot a route of ~ 28 jumps, all with injection.

Now, a bunch of question/feature request:
  1. Am i doing something wrong? Why that poor results in the first two cases?
  2. If the router could use the FSD injection only as last resource it could have succeed in plotting a complete route with the minimal injections possible: there's a reason that it uses it ALWAYS instead?
  3. could we get the destination system in the last row also when the route is only partial (and see the destination also in the galaxy map)? it could be useful to have an idea about how short the plotted route is.
Thanks :)
Click to expand...
I'll look into it, but as I always say, the best way of getting help is to give me the link/URL to the route you tried to generate, it makes things massively simpler for me.
 

Spansh

S
Parduz said:
Sorry, i didn't thought that it was important.

Here, from Zandu to Star One: https://spansh.co.uk/exact-plotter/results/B879C7F6-5F75-11ED-81BD-68BA94EB4526
(if you change algorythm, you'll get a very short result)
(also, compute time was set to 120, that variable is not set by the link)

And this from Star One to the last system in the previous route: https://spansh.co.uk/exact-plotter/results/EE81A782-5F76-11ED-9959-CBBC94EB4526

Thanks
Click to expand...
So there is a slight bug which is causing the very short route which Ive fixed. However the failure to find a route is somewhat expected, you actually cannot make the route without using Jumponium. You will note that even with Jumponium it has to take a somewhat roundabout route to get around the gap.

With regards to having the plotter only use Jumponium when it absolutely needs to, that's a much more difficult thing to determine for a number of reasons.

1. When should the plotter determine to use Jumponium, are you willing to travel an extra 5000LY to avoid using a single jumponium?
2. I'd have to actually try to generate the full route, then fail to find a suitable route before trying it again from the clost point with Jumponium, potentially many times, massively increasing ther search space for finding a route.

It's something I can potentially try to develop a new algorithm for but that's not a trivial exercise. For now you may just have to generate multiple routes by hand
 

Parduz

Parduz
Thanks for your reply.

About pt. 1 and 2, i see your concerns, but (i clearly don't know how you build a route, so i may be saying a stupid thing) i'd just be happy if, when you fail to find the next system, then just try to search again using FSD injection, and if it succeed, continue from there without it, just as if you've found that last system in the "normal" way. (perhaps adding a column in the report about FSD Injection being used?).
The result should be somewhat with a "minimum use" of injections (within the range limits your algorythm already have), at the price of being half (at least) as fast in findind the systems reachable only with injections.

In my mind it sounds feasible :D dunno about in reality :)

thanks again
 

Spansh

S
Parduz said:
Thanks for your reply.

About pt. 1 and 2, i see your concerns, but (i clearly don't know how you build a route, so i may be saying a stupid thing) i'd just be happy if, when you fail to find the next system, then just try to search again using FSD injection, and if it succeed, continue from there without it, just as if you've found that last system in the "normal" way. (perhaps adding a column in the report about FSD Injection being used?).
The result should be somewhat with a "minimum use" of injections (within the range limits your algorythm already have), at the price of being half (at least) as fast in findind the systems reachable only with injections.

In my mind it sounds feasible :D dunno about in reality :)

thanks again
Click to expand...
It's not that the system 'fails to find the next system'. When generating a route the system considers hundreds of thousands of systems (sometimes each hop), for a long route this can add up to billions of checks. So the system doesn't run out of things to check, it simply decides it has spent too much time on this route and times out. If I removed the timeout the system would continue to check systems, and would probably finish a couple of days later (after failing to find a route exhausting every known system in the galaxy), then it would have to start again from the closest system it got to with a jumponium jump, that's just not feasible.

However as I said, there might be something I can do with a different algorithm but it's not a trivial exercise to do that. I have a couple of ideas how it can work, but I have limited time for new development at the moment.
 

forbiddenlake

forbiddenlake
If you find out a way to automatically get that information out of the game, let us know.

Allegiance: Thargoid is a clue (after the search index regenerates) but I cannot find a way to get the Thargoid war state.
 

Spansh

S
I've increased the payouts on biology landmarks, and also bumped the minimum on the Expressway to Exomastery. It might be worthwhile doing some as it's pretty profitable right now
 
You must log in or register to reply here.
Back
Top Bottom