naruyoko / expantanum.js Goto Github PK

ExpantaNum('10^^1e30').toString() yields '10^^10^1e30' not '10^^1e30'

To reproduce:

1. Open https://naruyoko.github.io/ExpantaNum.js/test.html
2. In the code to be executed, input return ExpantaNum.arrow(3, 3, 3)
3. Leave the version at the default.
4. Click "Run".
5. Stderr output:

1670044697132ms, 23:18:17 GMT-0600 (Central Standard Time)

ReferenceError: OmegaNum is not defined
ReferenceError: OmegaNum is not defined
    at eval (eval at <anonymous> (https://naruyoko.github.io/ExpantaNum.js/test.js:102:12), <anonymous>:938:32)
    at Q.arrow (eval at <anonymous> (https://naruyoko.github.io/ExpantaNum.js/test.js:102:12), <anonymous>:956:38)
    at eval (eval at dg.onclick (https://naruyoko.github.io/ExpantaNum.js/test.js:7:51), <anonymous>:3:19)
    at dg.onclick (https://naruyoko.github.io/ExpantaNum.js/test.js:7:99)

Likely caused by this commit: d96a3cb

10^10^10^10^10^10^10^10^NaN

Reproduction Steps:
ExpantaNum.ZERO.toStringWithDecimalPlaces(3,true)
raises the following error:

Uncaught RangeError: toFixed() digits argument must be between 0 and 100
    at Number.toFixed (<anonymous>)
    at decimalPlaces (eval at <anonymous> (test.js:102:12), <anonymous>:1213:31)
    at P.toStringWithDecimalPlaces (eval at <anonymous> (test.js:102:12), <anonymous>:1250:28)
    at <anonymous>:1:17

This works in the console on https://naruyoko.github.io/ExpantaNum.js/test.html, and also still errors when replacing 3 with any other integer.

Input: ExpantaNum.pent(0,2) // 0^^^2
Output: NaN

Input: ExpantaNum.pent(0,Infinity) // 0^^^Infinity
Output: Infinity

I enter return new ExpantaNum("10{1000000}10").toHyperE();
It use few second and result: E10#0#1...(a lot of #1)...#1#11#9...(a lot of #9)...#9

https://naruyoko.github.io/ExpantaNum.js/docs.html

#4

It lose precision by a lot

This should be (10^^)^22 (10^)^150 153.15327796708874 for correct calculation or (10^^)^22 (10^)^75 153.15327796708874 for correct display

When running:

ExpantaNum.lambertw("ee4992767175317001")

ExpantaNum.js errors out with the following:

Uncaught Error: Iteration failed to converge: e1e4992767175317001
d_lambertw@https://naruyoko.github.io/ExpantaNum.js/test.js line 102 > Function:673:11
anonymous/P.lambertw@https://naruyoko.github.io/ExpantaNum.js/test.js line 102 > Function:688:51
anonymous/Q.lambertw@https://naruyoko.github.io/ExpantaNum.js/test.js line 102 > Function:692:30
anonymous@https://naruyoko.github.io/ExpantaNum.js/test.js line 7 > Function:3:19
@https://naruyoko.github.io/ExpantaNum.js/test.js:7:99

See also Patashu/break_eternity.js#152.

When the library standardizes numbers such as [[0, a],[1, x], ...] (x is bigger than the max safe integer), it crashes on line 1165, x.splice(1,i);.

I hope the bug will be fixed soon.

ExpantaNum.fromString("1.000F10")
// Malformed input: 1.000F10

ExpantaNum.HyperE("1.000F10")
// Malformed input: 1.000F10

Seems like this line:

needs to include [E-Ze]+ (or whatever the max range is):

-    isExpantaNum = /^[-\+]*(Infinity|NaN|(J+|J\^\d+ )?(10(\^+|\{[1-9]\d*\})|\(10(\^+|\{[1-9]\d*\})\)\^[1-9]\d* )*((\d+(\.\d*)?|\d*\.\d+)?([Ee][-\+]*))*(0|\d+(\.\d*)?|\d*\.\d+))$/,
+    isExpantaNum = /^[-\+]*(Infinity|NaN|(J+|J\^\d+ )?(10(\^+|\{[1-9]\d*\})|\(10(\^+|\{[1-9]\d*\})\)\^[1-9]\d* )*((\d+(\.\d*)?|\d*\.\d+)?([E-Ze]+[-\+]*))*(0|\d+(\.\d*)?|\d*\.\d+))$/,
                                                                                                                                           ^^^^^^^

And it seems like that works now:

isExpantaNum.test("1.000e10")
// true 

isExpantaNum.test("1.000e10e25")
// true 

isExpantaNum.test("1.000F10")
// true

isExpantaNum.test("Ge1.025e52")
// true

isExpantaNum.test("1.00FF125")
// true 

isExpantaNum.test("1.00A125e25")
// false

However, it doesn't parse the strings correctly still, just doesn't say it's malformed input.

fromHyperE looks to use a similar regex, but not isExpantaNum, I'm not sure what that's for.

I created a PR with jest tests to demonstrate the issue: #24

> ExpantaNum.pent("(10^^)^5 288",41.1)
< '(10^^)^42 (10^)^286 10000000000'
> ExpantaNum.pent("(10^^)^5 288",41)
< '(10^^)^44 (10^)^286 10000000000'

The first result should be bigger than the second result.

is there a function that I haven't seen that let's you use hyperoperations larger than tetration and pentation?

• The 1.5 in new ExpantaNum("J1.5") can be replaced by any positive decimal number and it will still result in a hang.
• The hang occurs in the P.standardize function.
• Reducing ExpantaNum.maxOps to 10 didn't help much.

ExpantaNum(2).factorial.toString();
"function (){\n var x=this.clone();\n var f=ExpantaNum.factorials;\n if (x.lt(ExpantaNum.ZERO)||!x.isint()) return x.add(1).gamma();\n if (x.lte(170)) return new ExpantaNum(f[+x]);\n var errorFixer=1;\n var e=+x;\n if (e<500) e+=163879/209018880Math.pow(e,5);\n if (e<1000) e+=-571/2488320Math.pow(e,4);\n if (e<50000) e+=-139/51840Math.pow(e,3);\n if (e<1e7) e+=1/288Math.pow(e,2);\n if (e<1e20) e+=1/12*e;\n return x.div(ExpantaNum.E).pow(x).mul(x.mul(ExpantaNum.PI).mul(2).sqrt()).times(errorFixer);\n }"