Matrix Games Forums

Forums  Register  Login  Photo Gallery  Member List  Search  Calendars  FAQ 

My Profile  Inbox  Address Book  My Subscription  My Forums  Log Out

RE: Accelerating Japanese airframes

 
View related threads: (in this forum | in all forums)

Logged in as: Guest
Users viewing this topic: none
  Printable Version
All Forums >> [New Releases from Matrix Games] >> War in the Pacific: Admiral's Edition >> RE: Accelerating Japanese airframes Page: <<   < prev  1 2 [3]
Login
Message << Older Topic   Newer Topic >>
RE: Accelerating Japanese airframes - 9/17/2011 11:33:31 PM   
Numdydar

 

Posts: 3211
Joined: 2/13/2004
Status: offline

quote:

ORIGINAL: darbymcd

out of curiousity, when you say focusing on, how much focus? how many factories are you talking about to produce a one month acceleration every 2 weeks?


One in particular is spread over 4 cities with 1 factory each that are fully repaired and shown as follows
7(0)
55(0)
3(0)
1(0)

So if my understanding is correct this is giving me the following points toward acceleration
7(0) = 1
55(0) = 3
3(0) = 1
1(0) = 1

for a total of 6 points/day. Which means that I will get an acceleration every 16-17 days which matches up with what is accutually occuring in my game.

(in reply to darbycmcd)
Post #: 61
RE: Accelerating Japanese airframes - 9/17/2011 11:41:29 PM   
darbycmcd

 

Posts: 394
Joined: 12/6/2005
Status: offline
hmm, that is very interesting, thank you.

doesnt' this strike you as far too quick? i mean you could get the same results with 2 more factories (replace the 55 with 3x1) so for just 6 factories you could get that kind of acceleration?

i am going to post this in the tech section as well, because i think there is some problem here.


< Message edited by darbymcd -- 9/17/2011 11:47:49 PM >

(in reply to Numdydar)
Post #: 62
RE: Accelerating Japanese airframes - 9/18/2011 1:32:18 AM   
Archangel85

 

Posts: 26
Joined: 3/15/2010
Status: offline
Concerning the switching of models, I always assumed that it went along the same lines as switching production factories, i.e. the damage of a factory when switching from one aircraft to another is determined by the difference between the two planes' durability. Aircraft subtypes of the same general type usually (always?) have the same durability (see comparison between Ki-44 and Ki-44 IIa above), meaning that they do not suffer production damage when production is switched because of this. Has anyone tested if this also applies to research, even outside the normal upgrade path?

(in reply to darbycmcd)
Post #: 63
RE: Accelerating Japanese airframes - 9/18/2011 1:35:50 AM   
michaelm75au


Posts: 13500
Joined: 5/5/2001
From: Melbourne, Australia
Status: offline
Please refer to this post in this thread that shows the correct R&D sequence.
http://www.matrixgames.com/forums/tm.asp?m=2908046&mpage=1&key=&#

Generating results based on beta p8 to q3 is meaningless, as I have reported that these versions have a bug.


_____________________________

Michael

(in reply to Archangel85)
Post #: 64
RE: Accelerating Japanese airframes - 9/18/2011 2:14:06 AM   
michaelm75au


Posts: 13500
Joined: 5/5/2001
From: Melbourne, Australia
Status: offline
quote:

ORIGINAL: Numdydar

Sorry to be dense but it has been a while since I last did any programing

Random(30)/30 means a random number between 1-30 is generated and then that result is divided by 30. Do I have that right? So 1/30 times you are able to get a 'bonus' of an extra plane produced?

If that is true, 29/30 times this will have 0 as a result as you only have a 1/30 chance of getting a 1. Unless you always round up. If you round up, then you always get 1 as a result.

So a 2(0) production factory whould have 2 points produced + a 1/30 chance of a +1. Then the Production routine would modify this to see if an actual plane was produced or not.

A 2(0) R&D factory would still be 2 points produced + a 1/30 chance of a +1 as for production factories
Then another random number based on the most likey outcome of the above which would be 2, would produce either a 1 or 2 as there were two devices. If the 1/30 'bonus' point was generated then this numbe could be 1,2, or 3. Correct so far?

As all these results would be 0 when divided by 10, (2/10= 0 rounded down) no R&D points would be generated.
However because the R&D factory was fully repaired, it gets a 'bonus' of 1 point

So ANY R&D factory that has less than 9 factories fully repaired could never exceed getting 1 point towards acceleration. If you had 9 then you would have a 1/30 chance of getting 2 points. ([9 factories + 1/30 chance of +1]/10 =1 + 1 'bonus' for fully repaired =2)

To complete the picture, 2 cities with 1 2(0) R&D factory each would generate 2 points/day. So every 50 days this plane would accelerate.

While 2 cities with 1 30(0) factory each would most likely produce 6 points/day. This would allow an acceleration every 16-17 days.

And if you can do it, 3 cities with 3 30(0) factories each could produce 27 points/day (3 cities * 3 factories * 3 points generated = 27). Evey 4 days or so you would accelerate.

Of course getting the R&D factories repaired is the key to all of this Once that is done and you have factories spread out in different cities, things can really speed up

Thanks for ALLyour help Michael. Hopefully I have it right now. Finally, right


Using the correctly restored initial formula for R&D, which is the same formula used for daily production.
active devices + random(30)/30

However I can see that the way it is written in not mathematically correct. it should be written as "(active devices + random(30))/30". And it just uses integer arithmetic which is truncating fractions.

A 30(0) factory would produce on a day "(30 + random(30))/30" planes - a number from (30+1)/30 to (30+30)/30; ie 1 to 2 with '1' more common.
(a) If a production factory, it would go off and build the number of planes from above.
(b) If a R&D factory with no damaged devices and the number above was >0, it would follow my original formula:
(i) divide by 10 -> 1 or 2 divided by 10 would be '0'
(ii) add 1 as there are no damaged devices -> above (0) + 1 would be '1'
If the factory was 2(0), the calculation for the initial number per day would be '0' for most days of the month.

This is why I say that 30(0) is a good size factory as it ensures some R&D per day.

To gain more from (b)(i) would require that the daily production be 10+ which would equate to a 300+ factory.

< Message edited by michaelm -- 9/18/2011 2:55:53 AM >


_____________________________

Michael

(in reply to Numdydar)
Post #: 65
RE: Accelerating Japanese airframes - 9/18/2011 2:56:01 AM   
kmitahj

 

Posts: 100
Joined: 4/25/2011
Status: offline
Hmm, looks like there are interesting (and somewhat scared)changes to R&D process... For the record I'll try to describe below how it looked like in the old, classic WitP. Not that it was better but it may be useful to compare. In fact I thought it was working the same in the old early versions of AE - could be wrong though. Besides in (early) AE there were two AC-Research paths - one "hard" which looked to me (again could be wrong) like in old WitP, and the other "easy" one. I assumed - maybe incorrectly - that the "hard", classic one was for human players whereas the easy path was implemented for AI to use?

Anyway in classic WitP:
Every day procedure loops through all ac-factories (actually it loops through all bases and all factories, manpower, oil, and resource centers in the given base at once but I specify below only steps related to ac-factories):

A. Number of planes build that day in the given ac-facory is calculated: PlanesBuild = (NumberOfReadyElements + Random1(30)) div 30.
This equation is the same for both production and R&D aircraft factories.
NumberOfReadyElements - number of undamaged elements/devices in the given ac-factory
Random1(N) - function returns random integer value between 1 and N.
div - integer division. That is important point - it means that result of division is zero as long as numerator is less then 30, is 1 when numerator is in (30,59) range, 2 for (60,89) range etc.

B. When for given factory PlanesBuild=0 processing is done for that ac-factory and code skips to next one.
Note that this is serious condition discriminating factories in regard to their (undamaged) size. Factory of size (undamaged) of 1 has only 1/15 chance of passing the test. Factory of size 14 has exactly 50% chance of passing this test. Finally factory with 29 or more undamaged elements has 100% chance of passing the test. This is where my "evangelisation" of 30+ size factories as an R&D sweetspot was coming from.

C. R&D factory processing is also skipped when:
- ac-factory has any damaged element/device.
- Index of the aircraft produced in the factory is above 75. Ac-index is the number which can be easily seen in WitP Database Editor. Looking in the editor it is easy to see that 75 is highest index designed for japanese aircraft. Numbers above are reserved for allied plane models which are not supposed to be reserched.
- Last (and most trivial) R&D path is skipped if given aircraft model is already in production.

D. Actual research pts generated in the ac-factory are calculated: ResearchPts = 1 + Random1(PlanesBuild) div 10
PlanesBuild - is the number calculated in step "A".
Note that if factory reached that step it generates at least 1 r&d point. However to reach step "D" factory must pass tests in step "B" and "C". And again test "B" depending on the factory size is discriminating for small size factories.
Note also that to have any chance to generate given day more then 1 r&d-point PlanesBuild must be equal at least 10. But looking at step "A" we see that to get 10 planes build a day factory must be a size of 270 at least. Rather rare and certainly costly case...

E. Research Points are capped to no more then three a day: ResearchPts = min(3, ResearchPts).
This step is hardly ever used as factory of size (at least) 870 would be needed to generate 4 r/d-pts in step "D".

F. Global per aircraft variable AircraftResearchPts is adjusted by value from step "E": AircraftResearchPts += ResearchPts.
AircraftResearchPts is hidden variable not seen on any of user interface panels. People often mixed it with the total number of undamaged elements of all ac-factories producing/researching given airplane model which IS displayed for BOTH production and research airplanes.

G. If AircraftResearchPts >= 100 then
AircraftAvailDate -= MONTH;
AircraftResearchPts -= 100;

H. Loop to next factory.


EDIT: took me long time to gather the post (looking into old notes and all) and in the mean time Michael post confirmed (as far as I understand) that aircraft-R&D works in AE eseentailly the same way as in classic WitP and made my post rather redundant.



< Message edited by kmitahj -- 9/18/2011 3:23:35 AM >

(in reply to michaelm75au)
Post #: 66
RE: Accelerating Japanese airframes - 9/18/2011 3:14:58 AM   
Numdydar

 

Posts: 3211
Joined: 2/13/2004
Status: offline

quote:

ORIGINAL: michaelm

quote:

ORIGINAL: Numdydar

Sorry to be dense but it has been a while since I last did any programing

Random(30)/30 means a random number between 1-30 is generated and then that result is divided by 30. Do I have that right? So 1/30 times you are able to get a 'bonus' of an extra plane produced?

If that is true, 29/30 times this will have 0 as a result as you only have a 1/30 chance of getting a 1. Unless you always round up. If you round up, then you always get 1 as a result.

So a 2(0) production factory whould have 2 points produced + a 1/30 chance of a +1. Then the Production routine would modify this to see if an actual plane was produced or not.

A 2(0) R&D factory would still be 2 points produced + a 1/30 chance of a +1 as for production factories
Then another random number based on the most likey outcome of the above which would be 2, would produce either a 1 or 2 as there were two devices. If the 1/30 'bonus' point was generated then this numbe could be 1,2, or 3. Correct so far?

As all these results would be 0 when divided by 10, (2/10= 0 rounded down) no R&D points would be generated.
However because the R&D factory was fully repaired, it gets a 'bonus' of 1 point

So ANY R&D factory that has less than 9 factories fully repaired could never exceed getting 1 point towards acceleration. If you had 9 then you would have a 1/30 chance of getting 2 points. ([9 factories + 1/30 chance of +1]/10 =1 + 1 'bonus' for fully repaired =2)

To complete the picture, 2 cities with 1 2(0) R&D factory each would generate 2 points/day. So every 50 days this plane would accelerate.

While 2 cities with 1 30(0) factory each would most likely produce 6 points/day. This would allow an acceleration every 16-17 days.

And if you can do it, 3 cities with 3 30(0) factories each could produce 27 points/day (3 cities * 3 factories * 3 points generated = 27). Evey 4 days or so you would accelerate.

Of course getting the R&D factories repaired is the key to all of this Once that is done and you have factories spread out in different cities, things can really speed up

Thanks for ALLyour help Michael. Hopefully I have it right now. Finally, right


Using the correctly restored initial formula for R&D, which is the same formula used for daily production.
active devices + random(30)/30

However I can see that the way it is written in not mathematically correct. it should be written as "(active devices + random(30))/30". And it just uses integer arithmetic which is truncating fractions.

A 30(0) factory would produce on a day "(30 + random(30))/30" planes - a number from (30+1)/30 to (30+30)/30; ie 1 to 2 with '1' more common.
(a) If a production factory, it would go off and build the number of planes from above.
(b) If a R&D factory with no damaged devices and the number above was >0, it would follow my original formula:
(i) divide by 10 -> 1 or 2 divided by 10 would be '0'
(ii) add 1 as there are no damaged devices -> above (0) + 1 would be '1'
If the factory was 2(0), the calculation for the initial number per day would be '0' for most days of the month.

This is why I say that 30(0) is a good size factory as it ensures some R&D per day.

To gain more from (b)(i) would require that the daily production be 10+ which would equate to a 300+ factory.


So the ONLY way you will ever get 3 points towards acceleration is to

Have a fully repaired 30(0) R&D factory (or larger)
Win the lucky 1 out of 30 times so you will have 60/30 - 2 points
Always add 1 due to fully repaired

quote:

If the factory was 2(0), the calculation for the initial number per day would be '0' for most days of the month.


But because if was fully repaired it would ALWAYS produce 1 point because of the below correct?

quote:

(ii) add 1 as there are no damaged devices -> above (0) + 1 would be '1'

(in reply to michaelm75au)
Post #: 67
RE: Accelerating Japanese airframes - 9/18/2011 7:32:42 AM   
michaelm75au


Posts: 13500
Joined: 5/5/2001
From: Melbourne, Australia
Status: offline
To get 3 RD, you need to have:
(a) a fully repaired factory large enough to produce 20+ planes a day to get 2 RD points (1/10 of the daily output)
(b) +1 for no damaged devices.
That would require the factory to be about 570(0) to 600(0) depending on die roll.

A factory of 2(0) would usually only produce '1' plane on 2 days during the month. Those 2 days would be the only days in the month that R&D could occur.


< Message edited by michaelm -- 9/18/2011 7:33:24 AM >


_____________________________

Michael

(in reply to Numdydar)
Post #: 68
RE: Accelerating Japanese airframes - 9/18/2011 4:25:01 PM   
Numdydar

 

Posts: 3211
Joined: 2/13/2004
Status: offline
quote:

ORIGINAL: michaelm

To get 3 RD, you need to have:
(a) a fully repaired factory large enough to produce 20+ planes a day to get 2 RD points (1/10 of the daily output)
(b) +1 for no damaged devices.
That would require the factory to be about 570(0) to 600(0) depending on die roll.
A factory of 2(0) would usually only produce '1' plane on 2 days during the month. Those 2 days would be the only days in the month that R&D could occur.



This is absurd. Am I the olny one that thinks this is messed up beyond belief? A fully repaired 570(0) R&D factory is NEVER going to happen. With the routine you have laid out, the Japanese player has absolutly NO reason to increase R&D factories as the MOST you can get out of any of them will be 1 point.

< Message edited by Numdydar -- 9/18/2011 4:29:10 PM >

(in reply to michaelm75au)
Post #: 69
RE: Accelerating Japanese airframes - 9/18/2011 5:10:38 PM   
kmitahj

 

Posts: 100
Joined: 4/25/2011
Status: offline

quote:

the Japanese player has absolutly NO reason to increase R&D factories as the MOST you can get out of any of them will be 1 point

Yep. 30(0) is a sweetspot and there is little point to expand r&d factories beyond. But hey, I told you so long time ago!

Is it absurd? I'm not so sure. I guess Garry Grisby would be more appropriate debater for you but let me ask why you think that more of this Factory scene from Modern Times should speed up research process?

Or let me requote myself (yuk) from other thread:
quote:

Think about factory as an entity with large number of hands working at assembly line (workers) and a few brains taking care to make the assemby process smooth (engineers). In research number of hands is not so important as number of brains working on plane improvements - engineers are perfectly happy to get one new plane prototype a day for their tests (hence factory of size 30 is enough to give steady stream of r&d pts day by day). But because the number of engineers in factory is limited you may use thier skills either on research of new aircraft model or on supervision of assembly lines expansion but not on both at the same time.

(in reply to Numdydar)
Post #: 70
RE: Accelerating Japanese airframes - 9/18/2011 7:12:56 PM   
darbycmcd

 

Posts: 394
Joined: 12/6/2005
Status: offline
So Nymdalar, you are getting results that are somewhat at odds with this setup, yes? Is this a version issue, ie are you using one of the betas that has the incorrect formula?

(in reply to kmitahj)
Post #: 71
RE: Accelerating Japanese airframes - 9/19/2011 2:07:02 AM   
Numdydar

 

Posts: 3211
Joined: 2/13/2004
Status: offline
Well after playing the game over a year in RL terms and setting up my production plans at the start of the game based on a very incorrect understanding of how R&D really worked I can say that yes I am at odds with the system as it will be in q4 and beyond. As I said before, why cap the R&D at 3 as there is no way you can ever get there. Why go through this convoluted section of code when all it's going to do is produce 1 point in 99.9999% of the time? Makes no sense.

I did notice that starting with p8 beta that my planes were accelerating better, but thought that was because my surperior planning was starting to pay off lol. Not because of a bug in the program Part of me feels that I should not get this issue fixed by using that q4 beta as all the HI and supplies that I have put into the R&D system to expand the factories would be totally wasted if I did. At least with the q2 version I am using now will give me some payback for the points I have put in, not to mention spending the time playing the game to get to this point.

(in reply to darbycmcd)
Post #: 72
RE: Accelerating Japanese airframes - 9/19/2011 1:46:33 PM   
SuluSea


Posts: 2358
Joined: 11/17/2006
Status: offline
My R&D plans start off with no factories researching anything after 12/43 it just takes too long to repair that far out and I've consider'd replanning for a 18 month window but feel I have enough factories to focus on the airframes I want badly. 

 Time is of the essence and factories sitting around for a year not doing nothing is too tough a pill to swallow, not to mention when Allied Bombers are taking up airspace in Japan research could get difficult.  Like with anything the more work and planning you put into anything the more you'll reap the rewards. The Japanese player has 77 airframe R&D factories to start the war , for instance the 55 R&D Frank factory at Maebashi goes straight into Ki-44a research and although the factory loses some points upon conversion I'll let that one convert into building as soon as the plane comes on line which will be around 38 IIRC , every factory will be built up to 30 if they aren't already at that level.

I'm always looking to refine , thanks for the reading & tips on an interesting part of the Japanese economy guys.









_____________________________

"There’s no such thing as a bitter person who keeps the bitterness to himself.” ~ Erwin Lutzer

(in reply to Numdydar)
Post #: 73
RE: Accelerating Japanese airframes - 9/20/2011 12:49:17 PM   
viberpol


Posts: 838
Joined: 10/20/2005
From: Gizycko, Poland, EU
Status: offline

quote:

ORIGINAL: Numdydar

quote:

ORIGINAL: michaelm

To get 3 RD, you need to have:
(a) a fully repaired factory large enough to produce 20+ planes a day to get 2 RD points (1/10 of the daily output)
(b) +1 for no damaged devices.
That would require the factory to be about 570(0) to 600(0) depending on die roll.
A factory of 2(0) would usually only produce '1' plane on 2 days during the month. Those 2 days would be the only days in the month that R&D could occur.



This is absurd. Am I the olny one that thinks this is messed up beyond belief? A fully repaired 570(0) R&D factory is NEVER going to happen. With the routine you have laid out, the Japanese player has absolutly NO reason to increase R&D factories as the MOST you can get out of any of them will be 1 point.


Hmm... that seems weird to me.
Maybe there are some other bonuses for the R&D that we don't really know...
and the formula IRL is little bit more complex?

This is a printscreeen from a real PBEM going from day one.
Not a mathematical theory nor the "clean" test.
I have 200 fully repaired N1K2 from turn 728 or so in three factories of 65+65+70.
Notice no upgrade from turn 728 till 769... and then? Every 10 or so days -- acceleration! And between 728 - 806 number of repaired/repairing factories didn't change a point.




Attachment (1)

_____________________________

Przy lackim orle, przy koniu Kiejstuta Archanioł Rusi na proporcach błysł

(in reply to Numdydar)
Post #: 74
RE: Accelerating Japanese airframes - 9/21/2011 4:06:31 AM   
Numdydar

 

Posts: 3211
Joined: 2/13/2004
Status: offline
If you're playing with any beta patch between p8 and q3, there is NO check for the first steps of the R&D routine. So with those size of factories you are getting 3 points each/turn towards acceleration for a total of 9 points a day. Which works out to about every 11 days as you indicate above.

Beta q4 and above, you will be reduced to 1 point per day each as the max you can get. This is because the first step in the R&D routine is now fixed. So you will be reduced from 9 points/day to 3. So about every 4 -5 weeks you will accelerate. As stated above to get to 2 points per day one factory would need to be 270(0) or better. This is what I have an issue with.

< Message edited by Numdydar -- 9/21/2011 4:07:15 AM >

(in reply to viberpol)
Post #: 75
RE: Accelerationg Japanese airframes - 10/9/2017 6:21:35 AM   
Numdydar

 

Posts: 3211
Joined: 2/13/2004
Status: offline
Thanks to Timotheus for pointing my spelling mistake in the title. I have corrected this so others my find this topic more easily when using the correct spelling versus my own unique way

But most of my OP was flawed as noted. To find out the real way R&D works for Japan, you can visit the link below. But you my learn more than just R&D. But that is the risk you take

http://www.matrixgames.com/forums/tm.asp?m=3329605

(in reply to Numdydar)
Post #: 76
Page:   <<   < prev  1 2 [3]
All Forums >> [New Releases from Matrix Games] >> War in the Pacific: Admiral's Edition >> RE: Accelerating Japanese airframes Page: <<   < prev  1 2 [3]
Jump to:





New Messages No New Messages
Hot Topic w/ New Messages Hot Topic w/o New Messages
Locked w/ New Messages Locked w/o New Messages
 Post New Thread
 Reply to Message
 Post New Poll
 Submit Vote
 Delete My Own Post
 Delete My Own Thread
 Rate Posts


Forum Software © ASPPlayground.NET Advanced Edition 2.4.5 ANSI

8.188