0

I'm looking for a website or something like that which helps me find the exact value of a decimal value of a number:

  • Of course, we cannot determine the exact value from a decimal as it is finite, but for example if I type in $1.414\ldots$, it should me all probable forms which I am looking for, one would simply be it's rational form, one could be $\sqrt{2}$ and many others$\ldots$ and we can get our exact form by typing in more digits.
  • I believe such a thing would be possible because I have seen something like that in past, but now I have forgotten what was it.
  • Another example would be if I type $1.155727349\ldots$, it shows $\mbox{me}\ \pi/{\rm e}$.

More simply put:

Is there any way in which I can get to the exact value of a number from it's decimal value if I know which constants would be appearing in it ?.

Thank you !.

Felix Marin
  • 94,079
V.G
  • 4,262
  • I tried WolframaAlpha and it works! However, if there are any other sites/software specifically designed for this purpose, I'd certainly like to know. – V.G May 16 '21 at 06:56
  • 6
    There's the original Inverse Symbolic Calculator: http://wayback.cecm.sfu.ca/projects/ISC/ISCmain.html It's ok, but not fantastic. There was a better one, but it's offline. :( But just in case it comes back, here's its URL: https://isc.carma.newcastle.edu.au/ – PM 2Ring May 16 '21 at 06:58
  • @PM2Ring Yes! I was talking about this only... I saw it before, but forgot about it. Thank you! – V.G May 16 '21 at 07:12
  • 1
    If you can use Python, the 3rd party library mpmath has a function for identifying numbers, which may be helpful: https://mpmath.org/doc/current/identification.html You can play with it on the SageMathCell server ; here's a tiny demo. It's not very smart, but it's better than nothing, I guess. ;) – PM 2Ring May 16 '21 at 07:17
  • Strictly speaking, we can never know which expression matches since the next digits could be different. But there are symbolic calculators finding possible matches (Wolfram Alpha is one of the tools). Of course, the more digits are given , the more likely is that the match is actually the intented number. – Peter Mar 11 '22 at 08:41
  • Also note that none of the tools will find every imaginable combination of known irrational numbers. But expressions you have in mind should be found without difficulty. – Peter Mar 11 '22 at 08:44
  • 3

2 Answers2

1

For these kind of tasks, I love to use the program Ries by Robert Munafo. It was essentially made for such tasks. Let me show you how it gives outputs (on the examples you ask).

For $1.414$, it gives

Your target value: T = 1.414                                  mrob.com/ries
            x^2 = 2                      for x = T + 0.000213562 {37}
            x^5 = e^sqrt(3)              for x = T - 1.75419e-05 {75}
   1/(log_5(x)) = sqrt(7)+2              for x = T + 1.36403e-05 {92}
        e^(x^5) = 8^e                    for x = T - 3.80911e-06 {92}
1/(x-sqrt(phi)) = e^phi+2                for x = T + 1.25986e-06 {102}
      1/ln(x)-2 = 1/4"/phi               for x = T - 6.46979e-07 {101}
      (x^2-1)^2 = cospi(1/8^2)           for x = T + 5.46898e-07 {96}
  x^2-sqrt(phi) = 1/e^(1/pi)             for x = T + 3.53186e-07 {97}

sinpi(1/(e^x)^2) = 1/(sqrt(2)+4) for x = T - 2.92598e-07 {106} 1/sinpi(ln(sqrt(x))) = 8/(1+pi) for x = T + 8.71324e-08 {109} log_(2/x)(x) = sqrt(sinpi(ln(phi))) for x = T + 7.7742e-09 {110} sinpi(-x+sqrt(2)) = 2/e^8 for x = T + 2.1526e-10 {111} (3-1/7) x = 1/5^2+4 ('exact' match) {117} (for more results, use the option '-l3')

NOTE: 'exact' match may result from floating-point roundoff error.

log_A(B) = logarithm to base A of B = ln(B) / ln(A) cospi(X) = cos(pi * x) e = base of natural logarithms, 2.71828... sinpi(X) = sin(pi * x) ln(x) = natural logarithm or log base e sqrt(x) = square root phi = the golden ratio, (1+sqrt(5))/2 A"/B = Ath root of B pi = 3.14159...

                 --LHS--      --RHS--      -Total-

max complexity: 67 62 129 dead-ends: 2747395 5223537 7970932 Time: 0.209 expressions: 196190 334648 530838 distinct: 102990 99793 202783 Memory: 14272KiB

Total equations tested:             10277681070 (1.028e+10)

For $1.155727349$, it gives

   Your target value: T = 1.155727349                            mrob.com/ries
            e^x = pi                     for x = T - 0.0109975   {42}
           x/pi = 1/e                    for x = T + 7.90922e-10 {57}
         x-pi/e = 1/-(e^(e^pi))          for x = T + 7.01771e-10 {109}
         pi-e x = 1/e^(e^3)              for x = T + 9.48281e-11 {108}
       pi 1/x-e = 1/e^(e^3)              for x = T - 1.35727e-11 {109}
    (e x)^(4^2) = pi^(4^2)-1             for x = T - 1.13793e-11 {126}
     1/(pi-e x) = 6^(8+pi)               for x = T + 4.12737e-12 {128}
          (for more results, use the option '-l3')

e = base of natural logarithms, 2.71828... pi = 3.14159...

                 --LHS--      --RHS--      -Total-

max complexity: 67 62 129 dead-ends: 2754425 5223469 7977894 Time: 0.212 expressions: 196844 334628 531472 distinct: 102092 99784 201876 Memory: 14208KiB

Total equations tested:             10187148128 (1.019e+10)

It gives you a lot of examples to pick and choose from. Hope this helps!


Obligatory xkcd

xkcd

Flavour text : Two tips: 1) $8675309$ is not just prime, it's a twin prime, and 2) if you ever find yourself raising $\log$(anything)^$e$ or taking the $\pi$-th root of anything, set down the marker and back away from the whiteboard; something has gone horribly wrong.

HackR
  • 3,940
0

As already pointed out in comments ISC ( Inverse Symbolic Calculator ) is the usual tool to find some mathematical expressions which numerical results are close to a given number.

As far as I know a first version referred as "Plouffe's Inverter" was built by Simon Plouffe. Now a more advanced version of ISC is online in the Canadian Centre for Experimental and Constructive Mathematics (Burnaby, Canada) : http://wayback.cecm.sfu.ca/projects/ISC/ISCmain.html More information was already given in comments.

A paper for general public on the subject : https://fr.scribd.com/doc/14161596/Mathematiques-experimentales . It is shown how one can built his own "Toy-ISC" working on his own PC.

More important is to consider how the results are affected by the number of digits used in the numerical calculus. For example comming back to the OP question they are infinity manny numbers close to $$1.155727349$$ Thus any ISC will give many different mathematical expressions corresponding to those numbers. One of them is $$\frac{\pi}{e}\simeq 1.155727349790922$$ It is not far from $1.155727349$ but the ISC will find many others closer to $1.155727349$. For example fancy ones : $$\frac{\sinh\left(\pi^2+\ln(\pi)\right)}{\cosh(\pi^2+1)} \simeq 1.155727349059204$$ $$\exp\left(-\frac{\cos(e)}{\sqrt{K}} \right)-\frac{3^{1/4}}{K}\simeq 1.155727349006754$$ $K$ is the Catalan constant. https://mathworld.wolfram.com/CatalansConstant.html

That is why the imput number into the ISC must have as many digits as possible in order to reduce the choice of mathematical expressions as output.

JJacquelin
  • 68,401
  • 4
  • 40
  • 91