More on xp gained by killing monsters

G

Guest

Guest
Archived from groups: rec.games.roguelike.adom (More info?)

Wherein Shyssiryxius gets killed a lot in the interests of science....

comments refer to v1.1.1
spoiler space
spoiler space
xp
experience
spd
speed
slow
shyssiryxius
greater
water
dragon
spoiler space
spoiler space

In the archives there's a formula (if I recall correctly) for how xp
diminishes with increasing number of kills of the same type of
creature. I think it was divide by 1+[n/20] where [] means round down.
Anyway I wanted to look at how xp varies with the more specific
attributes of a creature. In this case the Great Water Dragon
Shyssiryxius seemed like a good choice because you get a lot of xp for
her. She does not vary wrt PV, DV level and attack but hp and spd do
vary (measured by blessed stethoscope) from game to game. The data
tables have her down as hp=1140 spd=120. I found a variation of
1102-1197 hp and 108-131 spd. I think +/- 10% is typical for spd
variation in monsters. The hp variation seemed slanted - I was
expecting to find some values in the 1080-1100 range but it never
happened in about 100 trials. OTOH there were only 2 in the range
1180-1200.

XP variation by spd has been documented already - the ratio of your
speed to the victim acts as a factor in reducing your xp. Just to add
here that there is a limit to this. You cannot multiply the xp by more
than 3.2 -- so it's pointless slowing yourself below 5/16ths of the
speed of monster. Also note that if the monster is slowed there is an
*additional* 50% cut in xp on top of any effect of reducing it's speed.
ie typically casting slow on a creature will reduce xp to one quater.
The x3.2 cap is applied before the additional 50% penalty for slow. I
used the slow spell itself and if you could slow down a monster some
other way this penalty may not apply.

As for variation by hp the good news is that every time the GWD had the
same amount of hp it would yield the same xp (after taking into account
spd). Unfortunately the formula hp -> xp even for this one creature is
pretty odd. An increase in 1hp would increase the total xp by some
ratio of 104.75 xp (typically 209 xp) but the exact amount varied. The
best rule of thumb might be that the overall xp range (as "normalized"
for a spd 100 character and spd 100 monster) appeared to be 2 million -
1.9 million (assuming that the hp range is 1100 to 1200) or put another
way 1000 xp per hp on average (specific values were off as much as 10k
from that linear model though, and the variation had a sort of pattern
that repeated every 20 hp).

If this is correct it suggests that there's a value in the code
specifically for the GWD's xp value which is modified slightly by the
actual hp, rather than the xp being calculated using a generic formula
(same as with all monsters) based on a set of tables, her abilities,
hps, danger level (30) and so on (which would be the AD&D method). It
seems odd that such a value wouldn't appear in the data table dumps
we've had though...? I had expected to find a linear increase in XP
for each +1hp per the AD&D model but this certainly doesn't happen for
Shyssiryxius.

Not sure how monster experience level effects xp for killing them.
There have been reports of the ACW being spawned with levels greater
than 1. Has anyone seen that? Usually other than entering the SMC at
high levels you don't see high level monsters unless you killed so many
their xp is reduced.
 
G

Guest

Guest
Archived from groups: rec.games.roguelike.adom (More info?)

context free data as spoiler space

1102
1109 18 7
1110 3 1
1111 0 1
1113 133 2
1114 4 1
1121 15 7
1123 6 2
1125 6 2
1126 2 1
1128 5 2
1129 2 1
1135 139 6
1136 2 1
1139 7 3
1140 4 1
1141 0 1
1144 10 3
1145 2 1
1147 3 2
1148 4 1
1149 2 1
1156 143 7
1157 2 1
1159 5 2
1160 4 1
1167 15 7
1168 4 1
1169 2 1
1171 4 2
1172 127 1
1173 2 1
1175 6 2
1179 9 4
1186 17 7
1197 154 11

frobnoid wrote:
> On Thu, 23 Jun 2005 17:08:06 -0700, DavidByron wrote:
> Have you also ruled out variation due to alignment?

Methodology: save a character with the right equipment just before they
enter the GWD lair for the first time. In this case I used a drakeling
wizard to allow the use of a wand of cold to control spd, although for
these tests I didn't need it. They also had a blessed stethoscope and
boots of the slow shuffle. From saved game enter the lair, cast
darkness, summon the GWD by droping and picking up an item, note stats
using stethoscope, assign show_experience = true, kill GWD.

So there wasn't any variation in the character at all. Exactly the
same key strokes. I don't know if alignment effect xp. I haven't
heard that rumoured.

> I'd be interested in seeing people post the raw data they're using for the
> various calculations being done when possible. Anyone else have interest?

LOL, I doubt it. Anyway the table above is: col 1=hp of GWD. col
2=manipulated xp increase. col3 = partial diff of col 1. Specifically
col 2 is calculated by (1) multiply xp by 54, which was the spd of the
character in all cases, (2) divide by the spd of the GWD (3) subtract
the corresponding result for the row above (ie take the difference) (4)
divide by 104.725 xp rounding to the nearest integer. 104.725 was
chosen after noticing most of the increases were multiples of about 104
or 105. There wasn't much rounding to do (all results were within .025
of an integer) and if I added more digits I could have got it even more
accurate. So that was nice - 8 significant figures reduced to numbers
less than 200, and in most cases less than 10. Anyway you can see I
didn't carry on until I had all the hp values so there are gaps. col 3
represents the size of the gap so in general you'd expect col 2 to be
proportional to col 3. Ha. Still it looks like some kind of weird
rounding might explain most of it apart from the blips. at 1172, and
approximately 1152, 1132 and 1112. Note that where duplicate hp
results came in the xp value after step 1 and 2 was identical to within
1 in a million. 1140 and 1141 really did work out exactly the same xp.
1110 and 1111 were only a negligible 1 xp different out of 1.9
million. Otherwise the normalized xp values strictly increased as
you'd expect.

I was actually expecting something like the blips because of the AD&D
hypothesis. It's been a while but I think they categorised the
monsters broadly by hp total (this gave a base xp value) and then added
another amount per hp so that you'd expect to see a sloping graph with
periodical jumps. However I was expecting the jumps to appear at nice
round numbers like 1170, not 1172, and probably less jumps too. Maybe
on the hundreds or fifties. Not every 20.

if you want the unmanipulated data it's here (hp, spd of GWD, xp gain).
In many case there were more than one row for each listed hp value but
I've removed the duplicates, and (obviously) sorted the results by hp.

1102 123 4346804
1109 120 4244973
1110 124 4387193
1111 120 4245673
1113 108 3848963
1114 128 4562726
1121 111 3959968
1123 113 4032633
1125 111 3962551
1126 120 4284305
1128 119 4249757
1129 121 4321650
1135 110 3958426
1136 120 4318749
1139 126 4536398
1140 120 4321308
1141 116 4177265
1144 120 4323636
1145 119 4288067
1147 117 4216679
1148 127 4578064
1149 120 4326197
1156 123 4468462
1157 123 4468941
1159 111 4034022
1160 109 3962183
1167 130 4729321
1168 116 4220909
1169 120 4366923
1171 108 3931070
1172 129 4727217
1173 108 3958090
1175 120 4399272
1179 117 4291334
1186 131 4809146
1197 110 4071067
 
G

Guest

Guest
Archived from groups: rec.games.roguelike.adom (More info?)

On Thu, 23 Jun 2005 17:08:06 -0700, DavidByron wrote:

> Wherein Shyssiryxius gets killed a lot in the interests of science....
>
> comments refer to v1.1.1
> spoiler space
> spoiler space
> xp
> experience
> spd
> speed
> slow
> shyssiryxius
> greater
> water
> dragon
> spoiler space
> spoiler space

> If this is correct it suggests that there's a value in the code
> specifically for the GWD's xp value which is modified slightly by the
> actual hp, rather than the xp being calculated using a generic formula
> (same as with all monsters) based on a set of tables, her abilities, hps,
> danger level (30) and so on (which would be the AD&D method). It seems
> odd that such a value wouldn't appear in the data table dumps we've had
> though...? I had expected to find a linear increase in XP for each +1hp
> per the AD&D model but this certainly doesn't happen for Shyssiryxius.

Have you also ruled out variation due to alignment?

I'd be interested in seeing people post the raw data they're using for the
various calculations being done when possible. Anyone else have interest?