Difference between revisions of "Invoke"

From NetHackWiki
Jump to navigation Jump to search
m (cut words)
(I made an ash of myself)
Line 7: Line 7:
 
Invoking strategy is basically a trade-off between the risk of getting nothing and waiting too long. The turns to wait after a successful #invoke versus probability of success are graphed below (assuming your experience is <=17): <div class="thumb tcenter">[[Image:GraphWaitingTimeVersusInvokeSuccessProbability.svg|thumb|waiting time after successful #invoke versus probability of success of invocation]]</div>
 
Invoking strategy is basically a trade-off between the risk of getting nothing and waiting too long. The turns to wait after a successful #invoke versus probability of success are graphed below (assuming your experience is <=17): <div class="thumb tcenter">[[Image:GraphWaitingTimeVersusInvokeSuccessProbability.svg|thumb|waiting time after successful #invoke versus probability of success of invocation]]</div>
  
If you want to maximize the expected rate of successful invocations per time, you should wait 91 turns after a successful invoke, 121 turns after one consecutive failure, 211 turns after two consecutive failures and so on. In the table, after waiting for "wait" turns, "chance" means the probability of a successful invoke, and "rate" is 100 times the expected rate of successful invokes per turn.
+
Robert Schneck gives an invocation schedule in his [http://groups.google.com/group/rec.games.roguelike.nethack/msg/a987da11a8df6d6b ascension post of a liquid diet tourist]:
 
 
{| class="wikitable"
 
!Consecutive failures beforehand !! colspan="3" | your level <=17 !! colspan="3" | your level 18-20 !! colspan="3" | your level 21-23 !! colspan="3" | your level 24-26 !! colspan="3" | your level 27-29 !! colspan="3" | your level 30
 
|-
 
|&nbsp; || wait || chance || rate || wait || chance || rate || wait || chance || rate || wait || chance || rate || wait || chance || rate || wait || chance || rate ||
 
|-
 
|0||91||46.7||0.47||91||46.7||0.47||91||46.7||0.47||91||46.7||0.47||91||46.7||0.47||91||46.7||0.47
 
|-
 
|1||121||73.6||0.59||121||73.6||0.59||121||73.6||0.59||121||73.6||0.59||121||73.6||0.59||121||73.6||0.59

 
|-
 
|2||211||79.7||0.37||211||79.7||0.37||211||79.7||0.37||211||79.7||0.37||211||79.7||0.37||211||79.7||0.37

 
|-
 
|3||39||21.9||0.43||39||21.9||0.43||39||21.9||0.43||39||21.9||0.43||39||21.9||0.43||39||21.9||0.43

 
|-
 
|4||195||76.0||0.38||195||75.6||0.38||195||75.6||0.38||195||75.6||0.38||195||75.6||0.38||195||75.6||0.38

 
|-
 
|5||46||23.4||0.41||46||22.8||0.40||45||22.3||0.39||45||22.3||0.39||45||22.3||0.39||45||22.3||0.39

 
|-
 
|6||185||75.9||0.4||185||73.5||0.39||186||73.1||0.38||186||73.0||0.38||186||73.0||0.38||186||73.0||0.38

 
|-
 
|7||49||29.4||0.49||48||25.1||0.42||48||24.5||0.41||48||24.4||0.41||48||24.4||0.41||48||24.4||0.41

 
|-
 
|8||180||91.5||0.50||183||74.6||0.40||183||72.0||0.38||183||71.5||0.38||183||71.4||0.38||183||71.3||0.38

 
|-
 
|9||41||86.0||2.02||51||31.4||0.51||50||26.8||0.44||50||26.1||0.43||50||26.0||0.43||50||26.0||0.43

 
|-
 
|10||33||87.2||2.53||178||90.4||0.50||182||73.9||0.40||182||71.1||0.38||182||70.6||0.38||182||70.5||0.38
 
|-
 
|11||31||89.2||2.77||43||85.0||1.89||53||32.5||0.52||52||27.7||0.44||52||27.0||0.43||52||26.9||0.43
 
|-
 
|12||30||90.6||2.92||34||85.9||2.41||176||89.8||0.51||180||73.1||0.4||179||69.9||0.38||179||69.3||0.38
 
|-
 
|13||29||90.6||3.02||32||88.6||2.66||44||83.4||1.81||54||33.2||0.52||54||29.1||0.45||54||28.4||0.44
 
|-
 
|14||29||92.1||3.09||30||88.4||2.83||35||84.9||2.31||174||88.7||0.5||178||72.1||0.4||178||69.2||0.38
 
|-
 
|15||28||90.9||3.14||30||90.8||2.93||33||88.2||2.57||46||82.9||1.72||56||34.9||0.53||56||30.2||0.45
 
|-
 
|16||28||91.6||3.17||29||90.4||3.01||31||88.6||2.75||36||84.1||2.22||173||88.1||0.5||176||71.4||0.4
 
|-
 
|17||28||92.3||3.2||29||91.7||3.07||30||89.2||2.86||33||86.1||2.49||47||81.8||1.66||58||36.0||0.53
 
|-
 
|18||28||92.8||3.23||28||90.4||3.11||30||91.1||2.94||32||88.6||2.66||37||83.6||2.15||171||87.3||0.5
 
|-
 
|19||28||93.2||3.25||28||91.0||3.15||29||90.4||3.01||31||89.6||2.79||34||86.1||2.42||49||82.0||1.6
 
|-
 
|20||27||91.2||3.27||28||91.7||3.17||29||91.5||3.06||30||89.7||2.88||32||87.0||2.6||38||83.5||2.09
 
|}
 
 
 
At level 17, the probability of 20 fails is 0.067%. On condition that outcome doesn't occur, the expected rate of successful #invokes is once every 101.35 turns.
 
 
 
These numbers were computed using the exact probability distribution of rnz(100). The relevant waiting time is the turns you actually wait plus the 3d10 penalty for early invokes. For the first #invoke attempt, we maximize the probability of successful invocation, times the expectation of one over the waiting time. This quantity * 100 is listed in the table as "rate". The waiting time for the next attempt is then computed with the conditional distribution of the remaining artifact timeout, including the penalty from the previous attempt. Iteratively, the rate is again maximized and the conditional distribution updated. It isn't quite clear if each invocation attempt may indeed be optimized separately.
 
 
 
Robert Schneck gives a different invocation schedule in his [http://groups.google.com/group/rec.games.roguelike.nethack/msg/a987da11a8df6d6b ascension post of a liquid diet tourist]:
 
 
  82;81;68;81;88;67;54;96;78;62;88;81;65;63;81;65  
 
  82;81;68;81;88;67;54;96;78;62;88;81;65;63;81;65  
Unfortunately, he does not justify these numbers. He only [http://groups.google.com/group/rec.games.roguelike.nethack/browse_thread/thread/6dd7a0a1a6e90d96/61344bead5d498d6#61344bead5d498d6 hints] they could be the expected waiting time before the artifact regenerates on its own.
+
Unfortunately, he does not justify these numbers. He only [http://groups.google.com/group/rec.games.roguelike.nethack/browse_thread/thread/6dd7a0a1a6e90d96/61344bead5d498d6#61344bead5d498d6 hints] they could be the expected waiting time before the artifact regenerates on its own (which would be close to the solution).
  
 
== List of possible invocations ==
 
== List of possible invocations ==

Revision as of 11:15, 18 July 2010

The #invoke extended command allows you to activate certain objects. The only objects that can be activated this way are crystal balls and most quest artifacts.[1]

Artifacts tire after you invoke them, and you cannot do so again for rnz(100) turns (averages 100; can be up to 2000, but 95% of the time it is less than 400). If you invoke again too soon, "You feel that the <artifact> is ignoring you" and 3d10 turns are tacked on to the wait time.[2]


Optimum invocation schedule

Invoking strategy is basically a trade-off between the risk of getting nothing and waiting too long. The turns to wait after a successful #invoke versus probability of success are graphed below (assuming your experience is <=17):

waiting time after successful #invoke versus probability of success of invocation

Robert Schneck gives an invocation schedule in his ascension post of a liquid diet tourist:

82;81;68;81;88;67;54;96;78;62;88;81;65;63;81;65 

Unfortunately, he does not justify these numbers. He only hints they could be the expected waiting time before the artifact regenerates on its own (which would be close to the solution).

List of possible invocations

Invoking a non-artifact crystal ball is exactly the same as applying it.

List of possible invocations for new SLASH'EM artifacts[3]

  • The Candle of Eternal Flame: summons a (tame) fire elemental.
  • The Eye of the Beholder: reduces all non-undead monsters in eye's line of sight to 1/3 HP and reduces luck and alignment by 3 with luck above -9. Instadeath with luck below -9.
  • The Gauntlets of Defense: toggles invisibility on/off.
  • The Great Dagger of Glaurgnaa: boost in spell energy.
  • The Hand of Vecna: summons 4-7 tame graveyard creatures and reduces alignment and luck by 3 with luck above -9. Does 5-24 damage to player with luck below -9.
  • Holy Spear of Light: creates a lit field of a 12 square radius. This damages undead and demons in a 9 square radius.
  • The Storm Whistle: summons a (tame) water elemental.

References