2
$\begingroup$

How is it possible that Mathematica doesn't recognize it's own definitions of special functions ? I tried as input: 

Integrate[Cosh[I2t]/E^(6 Cosh[t]), {t, 0, Infinity}]

The above should give $K_{i2}(6)$ according to this. Unfortunately, it just happily reproduces my input, and leaves me ignorant of how it should have been phrased.

Does someone know how I could get Mathematica recognize to evaluate such input to something useful?

Moreover, when I plot

Plot[BesselK[2 I, z], {z, 0, 7}]

I get a graph, but when I evaluate

BesselK[2 I, 6]

again, Mathematica repeats my input.

I'm not really used to Mathematica, so sometimes I get this sort of frustrating output, Mathematica not understanding what I'm trying to say.

Improve this question
$\endgroup$
  • 3
    $\begingroup$ 1. I2t is treated as a single variable; put spaces in between variables that you intend to multiply. 2. BesselK[2 I, 6] is an exact number without an apparently simpler closed form; to see an approximation, use N[]: N[BesselK[2 I, 6]]. $\endgroup$ – J. M.'s ennui♦ Jun 3 '15 at 0:17
  • 1
    $\begingroup$ What makes you think this transformation rule is in Mathematica? While related, the Wolfram Functions site is not a source for Mathematica transformation rules per se... $\endgroup$ – ciao Jun 3 '15 at 0:35
  • 2
    $\begingroup$ @Guesswhoitis. You'll become the first portable computerless computer language interpreter $\endgroup$ – Dr. belisarius Jun 3 '15 at 2:03
  • $\begingroup$ Not quite yet there, @bel. :) $\endgroup$ – J. M.'s ennui♦ Jun 3 '15 at 2:14
  • $\begingroup$ Closely related: Incorrect results for elementary integrals when using Integrate. Unfortunately the solution there (Integrate`InverseIntegrate) doesn't work in general. Also related: Why does Mathematica give the wrong answer when integrating?. $\endgroup$ – Jens Jun 3 '15 at 4:07
2
$\begingroup$

Mathematica does not recognize your integral is BesselK[2 I, 6] even though it is. All the identities shown on http://functions.wolfram.com are not necessarily incorporated into Mathematica. The numeric equivalence of BesselK[2 I, 6] and the integral can be demonstrated by 

N[BesselK[2 I, 6], 16] == 
  NIntegrate[Cosh[2 I t]/E^(6 Cosh[t]), {t, 0, ∞},
    PrecisionGoal -> 16, 
    Method -> {"LocalAdaptive", Method -> "MultipanelRule"}]

True
Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.