Planetarion Forums

SYMM · 15 Sep 2003, 17:34

Fr-research is 48 ticks.
As Xan, i need to add 10%, which is 53 (rounding up all the way through...)
I have Res set to Eng-#1, so take 15%, which is 46.
I have 7 constructions, 2 of which are res-labs, so take another 20%, giving 37.
I just started the res this tick, and its taking 38 ticks.
If anyone suggests rounding i will shoot them, as doing it as one calculation gives 35.9.
Doing it (with rounding at each step) in different ways, gives various other numbers, all lower than 38.

Another thought i had was that the %ages are done as one, so 100+10-20-15=75%, which gives 36.

Is there something i'm missing?

zenopus · 15 Sep 2003, 19:29

The only way I can get 38 is by dividing by (1 - 10%) instead of multiplying by (1 + 10%)

Code:

48 / (1 - 10%) round up to 54
54 * (1 - 20%) round up to 44
44 * (1 - 15%) round up to 38

Odd way to do it though.

SYMM · 17 Sep 2003, 09:34

With no reply from "staff", are we supposed to assume Zenopus' reasoning is correct? (which is indeed, a very odd way to do it!)

NEWSBOT3 · 17 Sep 2003, 10:58

not idea tbh :\

Gerbie · 17 Sep 2003, 11:47

As I said before: the manual says 10%. That's wrong. It's approx. 1/6th extra research time for xan. They changed it and did not say that. You should have checked before the first tick as at that point you could still change race. :P

NEWSBOT3 · 17 Sep 2003, 12:55

Quote:

Originally posted by Gerbie
As I said before: the manual says 10%. That's wrong. It's approx. 1/6th extra research time for xan. They changed it and did not say that. You should have checked before the first tick as at that point you could still change race. :P

hmm, i shall verify this with spinner as this was not known to me :\

zenopus · 17 Sep 2003, 13:37

Quote:

Originally posted by Gerbie
... the manual says 10%. That's wrong. They changed it and did not say that. You should have checked before the first tick ...

Checked where?

Spinner · 17 Sep 2003, 20:02

I am afraid the manual is to blame here, very unfortunate.

The Races have the following research time modifications:
Terran : Normal
Cathaar : 20 % faster
Xandathrii: 20 % slower
Zikonian: Normal

For your specific example:

48 * 120 / 100 = 57,6 = 57 as it is forced to an Integer.
57 * 85 / 100 = 48,45 = 48 ^^^
48 * 80 / 100 = 38,4 = 38 ^^^^

So 38 it is.

at0mic.c0w · 18 Sep 2003, 00:04

:duh:

Tactitus · 18 Sep 2003, 03:56

Quote:

Originally posted by Spinner
For your specific example:

48 * 120 / 100 = 57,6 = 57 as it is forced to an Integer.
57 * 85 / 100 = 48,45 = 48 ^^^
48 * 80 / 100 = 38,4 = 38 ^^^^

So 38 it is.

:duh:

/me suggests Spinner use a proper rounding function

Doing 3 truncations can yield a worse-case error up to 3 ticks from the correct answer (as defined by the formula in the manual).

Also, you can combine the terms and do a single division, which will reduce the error regardless of whether you use a round-to-nearest or a straight truncation. In this case:

(48 * 120 * 85 * 80) / (100 * 100 *100) = 39.16 => round/truncation => 39

zenopus · 18 Sep 2003, 08:01

Quote:

Originally posted by Tactitus
Also, you can combine the terms and do a single division, which will reduce the error regardless of whether you use a round-to-nearest or a straight truncation. In this case:

(48 * 120 * 85 * 80) / (100 * 100 *100) = 39.16 => round/truncation => 39

Or he could just add up the percentages

Code:

48 * (100 - -20 - 15 - 20) / 100 = 40.8 truncate to 40

Tactitus · 18 Sep 2003, 14:46

Quote:

Originally posted by zenopus
Or he could just add up the percentages

Code:

48 * (100 - -20 - 15 - 20) / 100 = 40.8 truncate to 40

Yes; depending on how you define the formula.

The current algorithm applies the varioius research bonuses and penalties sequentially (ala compound interest), which should yield a result of 39.168--which gives 39 ticks. You could define them as all being applied concurrently (ala simple interest), in which case the result should be 40.8--which gives 40 ticks if you truncate or 41 ticks if you round-to-nearest.

SYMM · 18 Sep 2003, 14:58

I think we'll stick with 39....its long enough

Tactitus · 18 Sep 2003, 15:19

Quote:

Originally posted by SYMM
I think we'll stick with 39....its long enough

You mean 38?

I don't care what it is, personally, as I'm not playing. But I think it would be nice if one could figure it out from the manual.

zenopus · 18 Sep 2003, 21:51

This brings on another question.

For research variations are applied (and the intermediate result truncated) in this order

racial bonus
engineering priority bonus
research lab bonus

However, for production, looking at Zik with production as 1st eng.pri. it must be applied in this order

engineering priority bonus
racial bonus

Code:

 4 = 12 * 50% - 2

because the other way around would give 5 ticks

racial bonus
engineering priority bonus

Code:

 5 = (12 - 2) * 50%

Or am I missing something?

NEWSBOT3 · 18 Sep 2003, 23:30

Quote:

Originally posted by Tactitus
You mean 38?

I don't care what it is, personally, as I'm not playing. But I think it would be nice if one could figure it out from the manual.

will try and update manual before i leave tomorrow.

Gerbie · 19 Sep 2003, 08:11

Quote:

Originally posted by zenopus
This brings on another question.

For research variations are applied (and the intermediate result truncated) in this order
racial bonus

engineering priority bonus

research lab bonus

However, for production, looking at Zik with production as 1st eng.pri. it must be applied in this order
engineering priority bonus

racial bonus

Code:

4 = 12 * 50% - 2

because the other way around would give 5 ticks
racial bonus

engineering priority bonus

Code:

5 = (12 - 2) * 50%

Or am I missing something?

Nope, you calc the engineering bonus first, then apply percentages. It used to be like that with research too, when the racial bonusses there were still ticks instead of percentages.