1

Hi I'm asking this in context of this:

Suppose we have non-elementary integral $\chi(x) = \int K(x)dx$ and we want to find

$$\int^b_aK(x)dx$$

Now lets say there is some function $F(x)$ which has elementary integral $\int F(x)$

such that $K(x) + F(x) $ has elementary integral so we can do this :

$$\int^b_aK(x)dx = \int^b_a(K(x)+0)dx$$

$$ = \int^b_a(K(x)+ F(x) - F(x))dx$$

$$=\int^b_aK(x)+ F(x)dx + \int^b_a-F(x)dx$$

Maybe $\int^b_a K(x) + F(x)dx$ is not elementary but lets say I dilute this condition to there exist some definite integral property transformation or substitution which allows us to arrive at numerical value for e.g just like $\int \ln(\sin(x))dx$ which is non-elementary according to WolframAlpha but it gives into some computable value for appropriate limits like
$\int^{\pi/2}_0 \ln(\sin(x))dx$

So my end goal goal is converting $\int^b_a K(x) + F(x)dx$ into some function like that atleast.

  • 1
    I am not sure if the sum of an elementary plus a non elementary falls in the category of elementary, but the sum of two non elementary can become elementary. Not sure if that helps your specific problem – imranfat Dec 25 '20 at 06:23
  • There does not seem to be a well-formed question here. Is it possible that a function with non-elementary antiderivative produces an elementary definite integral within some special limits? Yes, but you already knew that. – Ivan Neretin Dec 25 '20 at 08:37
  • @IvanNeretin Let me rephrase this and try to more clear:Is it possible to find a function $F(x)$ such that adding it to $K(x)$ within some special limits , if not in general , change it to integral which is computable without resorting to special function? – Integral_spirit Dec 25 '20 at 09:34
  • 1
    Still, there is not much to be said, except for the trivial fact: if a definite integral within certain limits is elementary, then it is elementary, and if it isn't, then it isn't. An addition of a function F with elementary antiderivative will not change that. – Ivan Neretin Dec 25 '20 at 10:30
  • The fact that an indefinite integral is expressible as an elementary function is a different thing from a definite integral being expressed as a combination of well known numbers. These are two different problems and don't mix the two. – Paramanand Singh Dec 25 '20 at 11:18
  • @IvanNeretin As imranfat claims sum of two non-elementary integrals can become elementary(they didn't specify definite or indefinite), what's the magic here?. That sum of two non-elementary can be elementary but sum of one elementary and other non-elementary can't be? How can I verify that a given definite integral within a certain limit is elementary or computable without special functions? – Integral_spirit Dec 25 '20 at 13:43
  • @ParamanandSingh Now that I think about it, should I edit the question to : Can a sum of an elementary and non-elementary definite integral be evaluated without special functions? End goal is calculating the definite integral of such non-elementary integral without resorting to special functions.(or maybe just find the special limits withing which it is possible if not much can be done) – Integral_spirit Dec 25 '20 at 13:49
  • The magic here is rather trivial, too. It is better explained by the way of analogy. Can a sum of a rational number and an irrational number be rational? No way, for obvious reasons. What about the sum of two irrational numbers? Well, it can: $\sqrt2+(-\sqrt2)=0$. Same thing with elementary and non-elementary integrals. – Ivan Neretin Dec 25 '20 at 16:14

2 Answers2

3

I'm not familiar with the deeper details of definite integration. I only want to show that one of the assumptions in your question is wrong and that there are at least some cases (maybe single cases) where a definite non-elementary integral can be evaluated without special functions.

1.) The set of all elementary functions is closed regarding addition (It is an algebraically closed field.). $\int K(x)+F(x)\ dx=\int K(x)\ dx+\int F(x)\ dx$ cannot be elementary therefore. Your prerequisite "such that $K(x)+F(x)$ has elementary integral" is wrong therefore.

2.) Definite integrals can be calculated also by other methods than indefinite integration.
Lichtblau, D.: Symbolic definite integration: methods and open issues. ACM Communications in Computer Algebra 45 (2011) (1/2) 1-16
Raab, C. G.: Definite Integration in Differential Fields. PhD thesis Johannes Kepler University Linz, Austria, 2012
Algorithms for symbolic definite integration?
Davenport, G.: An Exploration of Three Related Parametric Definite Integrals. 2016
Davenport, G.: The Difficulties of Definite Integration
A non-elementary definite integral and therefore a sum of an elementary and a non-elementary definite integral can therefore be evaluated in some cases without special functions.
Nyblom, M. A.: On the evaluation of a definite integral involving nested square root functions. Rocky Mountain J. Math. 37 (2007) 4 1301-1304

3.) One could ask: Is there a theory of integration in elementary terms for definite integrals? But it seems there is no such theory.

4.) The value of a definite integral is a number. One could ask if this number can be an elementary number. But that's a completely different mathematical problem.

IV_
  • 6,964
  • Hi can you throw light on what kind of cases for which this is possible?, what condition or constraints can be used to identify such definite integrals? Maybe, cite a few examples. That paper is way above my mathematical training, I'm relieved to know that somebody has answered such question and I can get back to it once I have developed a proper mathematical background.Thank you – Integral_spirit Dec 26 '20 at 07:01
  • @Integral_spirit If there is no answer in here, you should ask this in another question. https://math.stackexchange.com/questions/176032/good-book-on-evaluating-difficult-definite-integrals-without-elementary-antider https://math.stackexchange.com/questions/101298/can-every-definite-integral-be-expressed-as-a-combination-of-elementary-function https://math.stackexchange.com/questions/902320/is-there-a-theory-of-integration-in-elementary-terms-for-definite-integrals https://math.stackexchange.com/questions/2006635/is-the-question-whether-the-value-of-a-given-definite-integral-has-a-closed-form – IV_ Dec 26 '20 at 10:24
  • @Integral_spirit https://math.stackexchange.com/questions/908603/can-you-prove-a-definite-integral-has-no-closed-form https://math.stackexchange.com/questions/2401302/closed-forms-of-definite-integrals https://mathoverflow.net/questions/92688/closed-form-finite-sum https://mathoverflow.net/questions/105689/techniques-to-solve-equations-involving-a-definite-integral/171979#171979 – IV_ Dec 26 '20 at 10:37
  • @Integral_spirit http://www-elsa.physik.uni-bonn.de/~dieckman/IntegralsDefinite/DefInt.html – IV_ Dec 26 '20 at 11:51
  • High thanks a lot for such an informative and well cited answer, thank you for spending your time to trackdown those references. Some of them are way out of my reach and others are manageable by me, nonetheless they all have been very informative. I'll get back to this answer again after going through them. I saw your profile and it was a treasure trove of answers related to such aspects of integration. Thank you a lot :) – Integral_spirit Dec 27 '20 at 06:08
1

$$\int_2^3 \Gamma'(x)dx=1$$

Does this fit your question?

Anixx
  • 9,119