Multiple instances of NaNs from (negative number)^(non-integer)

[mcpower]

A bugfix in break_eternity.js (Patashu/break_eternity.js@8b6efe7) for Patashu/break_eternity.js#43 is causing a few instances of NaNs in my post-game save when cherry-picked into Prestige Tree Rewritten:

• ba12 softcap:
  

  Prestige-Tree/js/sc.js

  Lines 273 to 280 in cec9198

  	ba12: {
  	title: "Top-Right Balance Upgrade (Positive Ion)",
  	type: "log",
  	start: new Decimal(0.75),
  	exp: new Decimal(0.25),
  	display() { return hasUpgrade("ba", 12) && upgradeEffect("ba", 12).gte(this.start) },
  	info() { return "Starts at "+format(this.start.times(100))+"%, logarithmic and brought to the "+format(this.exp.pow(-1))+"th root" },
  	},

  
  is causing errors here:
  

  Prestige-Tree/js/sc.js

  Lines 386 to 387 in cec9198

  	let exp = getSoftcapData(name, "exp");
  	return val.log10().pow(exp).times(start.div(start.log10().pow(exp)));

  
  specifically this portion:

  start.log10().pow(exp)

  start.log10() is -0.12493873660829993, so this calculates (-0.12493873660829993 ^ 0.25), causing a NaN. This has no real roots.

• nextMinibot:
  

  Prestige-Tree/js/layers.js

  Lines 8271 to 8274 in cec9198

  	nextMinibot() {
  	if (player.r.allotted.breeders.lt(1)||tmp.r.totalMinibots.gte(tmp.r.minibotCap.plus(player.r.spentMinibots))) return new Decimal(1/0);
  	else return Decimal.pow(10, tmp.r.totalMinibots.sub(player.r.grownMinibots.times(tmp.r.reduceMinibotReqMult)).plus(1).pow(1.5)).times(1e5).div(player.r.allotted.breeders.max(1).pow(tmp.r.breederExp));
  	},

  
  specifically, this portion:

  tmp.r.totalMinibots.sub(player.r.grownMinibots.times(tmp.r.reduceMinibotReqMult)).plus(1).pow(1.5)

  At one point in time:

  • tmp.r.totalMinibots = 1
  • player.r.grownMinibots = 627
  • tmp.r.reduceMinibotReqMult = 1

  so this calculates (-625 ^ 1.5), causing a NaN. This has no real roots.

• The Brain "next":
  

  Prestige-Tree/js/layers.js

  Line 7874 in cec9198

  next() { return Decimal.pow(10, Decimal.pow(10, new Decimal((player.ne.activeChallenge==11||hasAchievement("a", 151))?tmp.ne.challenges[11].amt:0).plus(1).div(tmp.ne.challenges[11].gainMult).root(tmp.ne.buyables[11].effect).log10().root(3).times(11)).sub(1)).sub(1) },

  That's a lot of code. Formatted using prettier, it looks like:

    Decimal.pow(
    10,
    Decimal.pow(
      10,
      new Decimal(
        player.ne.activeChallenge == 11 || hasAchievement("a", 151)
          ? tmp.ne.challenges[11].amt
          : 0
      )
        .plus(1)
        .div(tmp.ne.challenges[11].gainMult)
        .root(tmp.ne.buyables[11].effect)
        .log10()
        .root(3)
        .times(11)
    ).sub(1)
  ).sub(1);

  and the problem is in the .root(3) after the .log10, which ends up calculating (-2.6340821214933543 ^ (1/3)). This technically does have a real root of -1.3810511702377193, as (-x)^(1/3) = -(x^(1/3)).

• Wraith buyMax (spotted while going into The Brain in an end-game save):
  

  Prestige-Tree/js/layers.js

  Lines 5003 to 5006 in cec9198

  	buyMax() {
  	let target = player.ps.points.sub(1).div(2).min(player.ps.souls.sub(200).div(174).root(4).sub(1)).times(tmp.ps.buyables[this.id].scaleSlow).plus(1).floor().max(0)
  	player[this.layer].buyables[this.id] = player[this.layer].buyables[this.id].max(target)
  	},

  
  specifically this portion:

  player.ps.souls.sub(200).div(174).root(4).sub(1)

  player.ps.souls is zero, so this calculates (-1.1494252873563218 ^ (1/4)). This has no real roots.

There might be more instances that surface if you're not at end-game.

The easiest way of resolving this problem is to call .abs() before calling .root or .pow, as that should keep the behaviour identical to the currently used version of break_eternity.js if break_eternity.js is updated (and is a no-op if break_eternity.js isn't updated). However, in some cases a .max call might be more appropriate.