Supply path question

[Mickey3D]

One could always send a PM with a link

I'll do it this evening when I'll be back home if he hasn't replied until then.

[ohms_law]

That almost never works unless you take Morgantown and keep the Federals from flipping the rails into Harper's Ferry. Supplying a force across the mountains is foolish, because winter will almost always cut you off on those long, weak roads. Even as the Union, I tend to clear my way down, build a depot in every town, and then improve the regions, advancing out no sooner than mid-'63.

Challenge... accepted!

[ATTACH]31165[/ATTACH]

[Mickey3D]

I'll do it this evening when I'll be back home if he hasn't replied until then.

Message sent.


@Pocus: the .hst file is attached here below :
[ATTACH]31186[/ATTACH]

[Mickey3D]

Pocus will have a look but due to the release of EAW 1.01 patch this might take some times, certainly at least 1 week...

[Captain_Orso]

Thanks Mickey

[Pocus]

Hi,

I have looked into it. Did you replay several times the same turn or different turns?

Because, here is what I saw. The actual supply algorithm is prohibited from entering (by the 2 CS fleets) the 2 rivers, I can confirm that. At no point, there is a US supply point entering these 2 rivers.

Now, there is also the possibility to transfer supply "over the river" with the links between Chowan, VA (287) or Currituck, NC (293) toward Henderson, NC (302). This can be prevented if the link is cut, and it is cut if the success roll coming from

bloValueBySU_LinkCut = 23 // Perc chance for each combat ship (cumulative) that a transition link over water is cut between 2 land regions.
bloMaxPerc_LinkCut = 90 // max chance to have the link cut.

Normally, CS should be at 90% blockade in both river but there is 3 links in total, meaning there is overall a 90%^3 % chance per turn that supply is blocked, translated to 27% per turn supply is not blocked.

So if you played several turns and you see there is some resupply, I admit it is weird although not impossible. If you replayed the same turn, that's normal to get the same result.

Also there might be the possibility to go around from Washington, NC then using coastal transport in Pamlico Bay and then to the north up to Norfolk.


I'll probably not spend more time on that. I saw nothing obviously buggy and there are others things to attend. I admit it can be frustrating in this case to get this oddity though.

[Mickey3D]

I have looked into it

Thanks a lot as I know you are very busy.

Did you replay several times the same turn or different turns?

I replayed different turns and the test was made several times starting from the same situation.

Because, here is what I saw. The actual supply algorithm is prohibited from entering (by the 2 CS fleets) the 2 rivers, I can confirm that. At no point, there is a US supply point entering these 2 rivers.

Now, there is also the possibility to transfer supply "over the river" with the links between Chowan, VA (287) or Currituck, NC (293) toward Henderson, NC (302). This can be prevented if the link is cut, and it is cut if the success roll coming from

bloValueBySU_LinkCut = 23 // Perc chance for each combat ship (cumulative) that a transition link over water is cut between 2 land regions.
bloMaxPerc_LinkCut = 90 // max chance to have the link cut.

Normally, CS should be at 90% blockade in both river but there is 3 links in total, meaning there is overall a 90%^3 % chance per turn that supply is blocked, translated to 27% per turn supply is not blocked.

Not sure to understand how you get 27% ? I get 72.9 % to block the supply on the 3 links.

So if you played several turns and you see there is some resupply, I admit it is weird although not impossible.

It's what happened and the test was made several times.

Also there might be the possibility to go around from Washington, NC then using coastal transport in Pamlico Bay and then to the north up to Norfolk.

Right, could be a possibility. But there is no structure in Washington(NC) so the supply must reach Halifax and/or Bertie in one supply phase. As Washington region is a swamp, it's a very long path all the more in winter.

[Captain_Orso]

Hi,

I have looked into it. Did you replay several times the same turn or different turns?

I loaded the scenario which Mickey3D provided and ran 4-5 turns. <= This I did twice.

Because, here is what I saw. The actual supply algorithm is prohibited from entering (by the 2 CS fleets) the 2 rivers, I can confirm that. At no point, there is a US supply point entering these 2 rivers.

So, the riverine transport pool cannot move THRU the Chowan River region nor the Albemarle Sound region because of enemy navel elements. This is as I expected, but one question remains about this:

- How many enemy naval combat elements are required to block friendly supply from using the riverine transport pool through such a river region?

Now, there is also the possibility to transfer supply "over the river" with the links between Chowan, VA (287) or Currituck, NC (293) toward Henderson, NC (302). This can be prevented if the link is cut, and it is cut if the success roll coming from

bloValueBySU_LinkCut = 23 // Perc chance for each combat ship (cumulative) that a transition link over water is cut between 2 land regions.
bloMaxPerc_LinkCut = 90 // max chance to have the link cut.

Normally, CS should be at 90% blockade in both river but there is 3 links in total,

I assume you mean
Chowan, VA <-> Henderson, NC
Currituck, NC <-> Henderson, NC
Currituck, NC <-> Bertie, NC and then up to Henderson, NC overland.

meaning there is overall a 90%^3 % chance per turn that supply is blocked, translated to 27% per turn supply is not blocked.

So if you played several turns and you see there is some resupply, I admit it is weird although not impossible. If you replayed the same turn, that's normal to get the same result.

So supply using the jump-links to CROSS a river region use the standard Blocking-a-River-Crossing algorithm and thus there is always at least a 10% chance of supply getting ACROSS a river region and this is without using the riverine transport pool.

Actually I will assume that the possibility of supply thus crossing the Chowan River is even greater, because there are not only 3 links which can be used, but also 3 phases in which they can be moved across the river; or is blockage rolled for once per link for all three phases?

Also there might be the possibility to go around from Washington, NC then using coastal transport in Pamlico Bay and then to the north up to Norfolk.

I think you mean the other way around , coming from Norfolk, down the coast thru Cape Henry, Currituck Sound, Pamlico Sound, Pamlico Bay and landing on Washington NC.

I thought of that too, but Forts Morgan and Clark have Pamlico Sound and Pamlico Bay under their guns and thus supplies cannot cross them.

Actually if that were possible, the shorter route would be from Currituck, NC, through Oregon Inlet, Pamlico Sound, Pamlico Bay and then into Washington, NC. But Pamlico Sound being under the guns of both forts still blocks Pamlico Sound from using the riverine transport pool and thus is a no-go.

I'll probably not spend more time on that. I saw nothing obviously buggy and there are others things to attend. I admit it can be frustrating in this case to get this oddity though.

The most irritating thing is that supplies are obviously crossing, but the supply-debugging messages say nothing about it. During the first turn's first 2 distribution phases I see messages of supplies arriving in Henderson, NC (stockade) [there's actually a level 2 depot there too] and going directly to some of the stacks with supply trains. But after that I see not one single message of supplies arriving in Henderson nor any of the stacks there, but the stacks are either maintaining their levels or gaining when they should be using at least 1/4 of their maximum total supply capacity per turn.

[pgr]

I have looked into it.

Yay, what a nice guy!

[Mickey3D]

Actually I will assume that the possibility of supply thus crossing the Chowan River is even greater, because there are not only 3 links which can be used, but also 3 phases in which they can be moved across the river; or is blockage rolled for once per link for all three phases?

Good point, I was not thinking the supply check was made during the 3 supply phases. But even like this the chance to block the supply is (0.9^3)^3 = 38.7%. I might have been unlucky but my tests (see Excel file attached to post #32) showed the Union forces were receiving supply each turn.

[Pocus]

The supply algorithm can only enter a region with 25% more MC and devoid of ANY enemy military unit. So the answer to your question Cpt is ONE ship is sufficient.

To cut a link, you have the formula above, capped at 90%. The roll that will make a link disappear for one turn is done at the begining of hosting, so there won't be 3 chances though.

Mystery yes...

[Captain_Orso]

The "fog" is slowly lifting

Many thanks

I've been busting my head trying to figure out why, every time I start over again running the first turn and checking for supply distribution that I get the exact same results. And then I realized that it might well have to do with the die-roll "randomizer. Is there a simple way to reset this?

[Captain_Orso]

The supply algorithm can only enter a region with 25% more MC and devoid of ANY enemy military unit. So the answer to your question Cpt is ONE ship is sufficient.

To clarify, the entire rule must read:

[TABLE="class: outer_border, width: 600, align: left"]
[TR]
[TD]
>=1 uncontested[SUP]1)[/SUP], enemy combat element (not in Passive Posture) will block friendly supply from entering a region. If the region in question is a land region, the phasing faction must also have >25% MC for supply to enter the region.
[INDENT][SUP]1)[/SUP]To contest friendly supply being blocked, the friendly phasing side must also have >=1 combat element (not in Passive Posture) in the region.[/INDENT]

If the phasing faction controls an harbor in the region and at least one exit-point of that harbor is not blocked as per the above rule, supply may enter the region through the harbor.
[/TD]
[/TR]
[/TABLE]

I know that if both factions have forces in a region that supply still at moves into that region, and probably also through it.

Am I missing something?

To cut a link, you have the formula above, capped at 90%. The roll that will make a link disappear for one turn is done at the beginning of hosting, so there won't be 3 chances though.

Mystery yes...

[Pocus]

People must understand that supply can flow either through the river or 'over' the river, i.e riverine transport or moving through a ford/bridge. This is what we discussed before with the 'moving into river', 'moving through the link'.

[Captain_Orso]

Yes, this was very clear to me, about the through a river region (per the riverine transport pool) or OVER a river region (per the jump-link).

From my understanding, the rules for blocking the use of the jump-link OVER a river region are the same for moving units as supply.

In other words, the friendly side cannot negate blocking the jump-link by simply putting his own naval combat elements in defensive posture into the river region. The player must actually drive the enemy naval combat units out of the river region.

But the rule for -supply entering- a region I do not think can work the same. Otherwise, when friendly and enemy forces are facing off at each other in a region--happens a lot when large forces battle and end with "Stalemate"--neither side would receive supplies at all, and both would starve after several turns.

This is what I am trying to confirm.

[Mickey3D]

People must understand that supply can flow either through the river or 'over' the river, i.e riverine transport or moving through a ford/bridge. This is what we discussed before with the 'moving into river', 'moving through the link'.

This difference between riverine transport and land transport over a river was always clear to me. But given the comment here :

[INDENT]

To cut a link, you have the formula above, capped at 90%. The roll that will make a link disappear for one turn is done at the begining of hosting, so there won't be 3 chances though

[/INDENT]

and knowing that the path through Pamlico Bay was cut by Forts Morgan and Clark, I can't understand why the three links were never cut by the CSA fleet on the Chowan river. If my statistic skill is correct, the chance to cut it was nearly 73% per turn

[Mickey3D]

But the rule for -supply entering- a region I do not think can work the same. Otherwise, when friendly and enemy forces are facing off at each other in a region--happens a lot when large forces battle and end with "Stalemate"--neither side would receive supplies at all, and both would starve after several turns.

My understanding is that if you have at least 25% MC in the region, you will receive supply despite the presence of the ennemy.

But it often happens that you attack a force and can't drive it out of the region and at the same time you are still in the region at the end of the turn (your retreat time is extending over the end of the turn) with no MC or very low MC. In this case you won't be supplied.

[Captain_Orso]

You have to think of all situations.

In a region in which you have 100% MC. The enemy moves a single Partisan/Raider/Ranger/Cav in Defensive/Retreat/Avoid Combat with Cohesion level 20/60 and because the enemy Partisan is there, no supplies can enter that region, which includes moving through it.

My question is, if I also have a cav in Offensive or Defensive in that region, does that negate the partisan supplies from enter that region?

Or is my supply blocked no matter what, even if I have a corps with 5000 Power in the region?

Remember, supply distribution is before movement. So if your enemy moves a Partisan/Raider/Ranger/Cav into the region through which you need to move supplies; if at the end of movement, if the rule is absolute--it makes no difference how many friendly units you have in the region--you will get no supply into nor thru that region.

[Mickey3D]

In a region in which you have 100% MC. The enemy moves a single Partisan/Raider/Ranger/Cav in Defensive/Retreat/Avoid Combat with Cohesion level 20/60 and because the enemy Partisan is there, no supplies can enter that region, which includes moving through it.

If nothing has changed since AACW then supply will pass in/through the region as long as you have at least 25% MC, independently of the number and type of ennemy units.

[Captain_Orso]

You see, that stands in direct contradiction to what Pocus wrote above: "The supply algorithm can only enter a region with 25% more MC and devoid of ANY enemy military unit".

[Pocus]

Sorry, I meant 'uncontested' military units.

As for the probabilities Mickey3D, yes, 73% a turn... So I admit there is still something odd.

[Captain_Orso]

Sorry, I meant 'uncontested' military units.

Ahhhh--- That feels good now.

This difference between riverine transport and land transport over a river was always clear to me. But given the comment here :

and knowing that the path through Pamlico Bay was cut by Forts Morgan and Clark, I can't understand why the three links were never cut by the CSA fleet on the Chowan river. If my statistic skill is correct, the chance to cut it was nearly 73% per turn

As for the probabilities Mickey3D, yes, 73% a turn... So I admit there is still something odd.

Back to testing I guess.

I wonder if one issue with testing with Mickey3D's save is that during the first turn supply handling, the die-roles are always the same. So you can start over as much as you want, but the results are always the same.

If there's anything I can do to test this better...

[Mickey3D]

I wonder if one issue with testing with Mickey3D's save is that during the first turn supply handling, the die-roles are always the same. So you can start over as much as you want, but the results are always the same.

That would be strange to say the least : the game would store in the .hst a list of random generated numbers in advance ?

[Pocus]

No, but random numbers in computers are often (but not always) pseudo-randoms, and for AGE, it is necessary that the same entry data give the same result, or I could not reproduce bugs using your saves.

[Mickey3D]

No, but random numbers in computers are often (but not always) pseudo-randoms, and for AGE, it is necessary that the same entry data give the same result, or I could not reproduce bugs using your saves.

Ok, so we now understand why we always have the same result. But still, this is really bad luck to miss the 73% chance to block supply through several turns. As asked by Captain Orso, is there a way to change the random number seed ?

[Captain_Orso]

I think currently the main issue is that:

1. The Supply Debugging output is showing some supply moving into Henderson, NC during the first turn only, but not afterwards, and

2. The supply status of the stacks in Henderson, NC does not correspond to the amount arriving (after the first turn, none) less the supply consumed.


I can only think of two things which might cause this discrepancy:

a. Supply is arriving in Henderson, NC which the debugging output is not showing, or

b. Supply is not being consumed.

[Pocus]

You can't change the random numbers except by doing a different turn

[Mickey3D]

You can't change the random numbers except by doing a different turn

So, if I understand you well, it is normal that the first turn is always the same but the following turns can differ between two tests (i.e. playing several turns in a row and then starting again from the initial position) ?

In this case it is really strange the supply is never blocked.

[Captain_Orso]

The engine has a loooong string of random numbers, which it uses for generating its dr (die roles).

So, if you do the same thing every time, you will always get the exact same results, because all the dr made by the engine--which are generated from the random string--for each probability-check encountered by the engine, will be identical during each iteration of running a turn.

-

It is not yet certain that supply is actually reaching Henderson, NC while the supply debugging log is not reporting any reaching there. But it seems that either that is occurring, or the stacks in Henderson are not using supplies.

[ohms_law]

It doesn't really have "a loooong string of random numbers", but I suppose that's a decent enough explanation if it helps people to understand something of what's going on.
For a more accurate description, you could start with: https://en.wikipedia.org/wiki/Pseudorandom_number_generator

The important thing, for this discussion, is that the RNG will give the same output with the same input. What Pocus is saying is that in order to get different input, you need to run another turn.
He,y the good thing about the design being the way it is, is that there's no possibility of "cheating" with PBEM games. Re-running a turn will always provide (very nearly) the same result (assuming that nobody changes the orders for either side, of course).