Skip to main content
Log in

Generating-function representation for scalar products

  • Original Paper
  • Published:
Journal of the Korean Physical Society Aims and scope Submit manuscript

Abstract

We employ the generating-function representation for an n-dimensional vector in Euclidean or Hilbert space to evaluate scalar products. The generating function is constructed as a power series in a complex variable weighted by the components of a vector. The scalar product is represented by a convolution of the generating functions for the vectors integrated over a closed contour in the complex plane. The analyticity of the generating functions associated with the Laurent theorem reduces the evaluation of the scalar product into counting combinatoric multiplicity factors. As applications, we provide two exemplary computations: the sum of the squares of integers and the normalization of normal modes in a vibrating loaded string. As a byproduct of the latter example, we find a new alternative proof of a famous trigonometric identity that is essential for Fourier analyses.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Similar content being viewed by others

References

  1. G.B. Arfken, H.-J. Weber, F.E. Harris, Mathematical Methods for Physicists, 7th edn. (Elsevier, India, 2012)

    MATH  Google Scholar 

  2. G.N. Watson, J. Lond. Math. Soc. S1–8, 189 (1933)

    Article  Google Scholar 

  3. G.N. Watson, J. Lond. Math. Soc. S1–8, 194 (1933)

    Article  Google Scholar 

  4. G.N. Watson, J. Lond. Math. Soc. S1–8, 289 (1933)

    Article  Google Scholar 

  5. M. Abramowitz, I. Stegun, Handbook of Mathematical Functions (Dover, New York, 1965), p. 785

    Google Scholar 

  6. U-R. Kim, D.-W. Jung, J. Lee, and C. Yu, Solving eigenproblem with analyticity of generating function. J. Korean Phys. Soc. (in press)

  7. J. Brown, R. Churchill, Complex Variables and Applications, 9th edn. (McGraw-Hill Education, New York City, 2013). (See, for example)

    Google Scholar 

  8. S.K. Lando, Lectures on Generating Functions (American Mathematical Society, Rhode Island, 2003), p. 23

    Google Scholar 

  9. J. Beery, Sums of Powers of Positive Integers-Introduction (Mathematical Association of America Publications, Washington, D.C., 2009)

    Book  Google Scholar 

  10. A.K. Svinin, S.V. Svinina, Russ. Math. 63, 25 (2019)

    Article  Google Scholar 

  11. J.B. Marion, Classical Dynamics of Particles and Systems, 5th edn. (Cengage Learning, Boston, 2003)

    Google Scholar 

  12. A.L. Fetter, J.D. Walecka, Theoretical Mechanics of Particles and Continua (McGraw-Hill, New York City, 1980). (Chapter 24)

    MATH  Google Scholar 

  13. R.A. Matzner, L.C. Shepley, Classical Mechanics (Prentice-Hall, Hoboken, 1991), pp. 232–239

    MATH  Google Scholar 

  14. P.D. Ritger, N.J. Rose, Differential Equations with Applications (McGraw-Hill, New York City, 1968), pp. 367–372

    MATH  Google Scholar 

  15. D.-W. Jung, W. Han, U-R. Kim, J. Lee, C. Yu, Finding normal modes of loaded string with Lagrange multipliers, KPOP\({\mathscr {E}}\)-2020-08

Download references

Acknowledgements

As members of the Korea Pragmatist Organization for Physics Education (KPOP\({\mathscr {E}}\)), the authors thank the remaining members of KPOP\({\mathscr {E}}\) for useful discussions. The work is supported in part by the National Research Foundation of Korea (NRF) under the BK21 FOUR program at Korea University, Initiative for science frontiers on upcoming challenges. The work of JL is supported in part by grants funded by the Korea government (MSIT) under Contract Nos. NRF-2020R1A2C3009918 and NRF-2017R1E1A1A01074699. The work of DWJ and CY is supported in part by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education 2018R1D1A1B07047812 (DWJ) and 2020R1I1A1A01073770 (CY), respectively. All authors contributed equally to this work.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Jungil Lee.

Appendix A: Proof of Eq. (55)

Appendix A: Proof of Eq. (55)

In this appendix, we prove the identity in Eq. (55) by applying the generating-function approach. We define a sequence \(a^{(r)}\) as

$$\begin{aligned} a^{(r)}\equiv & {} \left\{ \left. \sin \frac{rk\pi }{n+1} \right| k=1,~2,~\ldots ,~n. \right\} ,\nonumber \\ r= & {} 1,~2,~\ldots ,~n. \end{aligned}$$
(A1)

The corresponding generating function can be constructed from Eqs. (43) and (54) as

$$\begin{aligned} g\left( z; a^{(r)} \right)= & {} \displaystyle \sum _{k=1}^n z^k\sin \frac{rk\pi }{n+1}= \sin \theta _{r}\frac{z+(-1)^{r-1} z^{n+2}}{(z-z_r^+)(z-z_r^-)}, \nonumber \\ \theta _r= & {} \frac{r\pi }{n+1}, \end{aligned}$$
(A2)

where \(z_r^\pm \) are defined in Eq. (44). As is stated in Ref. [6], \(g\left( z ,a^{(r)} \right) \) is an entire function: analytic in the entire complex plane. Thus, the function does not have any poles. Then, the left-hand side of Eq. (55) can be expressed in terms of \(a^{(r)}\) and \(a^{(s)}\) as

$$\begin{aligned}&\sum _{k=1}^n \sin \frac{rk\pi }{n+1}\sin \frac{sk\pi }{n+1}=\sum _{k=1}^n a^{(r)}a^{(s)},\nonumber \\&r,\,s=1,~2,~\ldots ,~n. \end{aligned}$$
(A3)

Substituting \(\psi ^\star =a^{(r)}\) and \(\chi =a^{(s)}\) into Eq. (18), we can express the series in Eq. (A3) as

$$\begin{aligned}&\sum _{k=1}^n a^{(r)}a^{(s)} = \frac{1}{2\pi i}\oint _C \frac{\mathrm{d}z}{z} g\left( z; a^{(r)} \right) g\left( \frac{1}{z}; a^{(s)}\right) . \end{aligned}$$
(A4)

Substituting the generating function in Eq. (A2) into Eq. (A4) and dividing the expression by \(\sin \theta _{r}\sin \theta _{s}\), we find that

$$\begin{aligned} \displaystyle&\!\!\!\frac{\sum _{k=1}^n a^{(r)}a^{(s)}}{\sin \theta _{r}\sin \theta _{s}}\nonumber \\&\,\,=\displaystyle \frac{1}{2\pi i}\oint _C \frac{\mathrm{d}z}{z} \left[ \frac{ z+(-1)^{r-1} z^{n+2} }{(z-z_r^+)(z-z_r^-)} \right] \left[ \frac{ \frac{1}{z}+(-1)^{s-1} \frac{1}{ z^{n+2}} }{\frac{1}{z^2}(z-z_s^+)(z-z_s^-)} \right] \nonumber \\ \,&\,\,=\displaystyle \frac{1}{2\pi i}\oint _C \frac{\mathrm{d}z}{z} \frac{z^2\left[ 1 +(-1)^{r+s} + (-1)^{s-1} z^{-n-1}+ (-1)^{r-1} z^{n+1} \right] }{(z-z^+_r)(z-z^-_r)(z-z^+_s)(z-z^-_s)}.\nonumber \\ \end{aligned}$$
(A5)

According to Eq. (44), \(z^\pm _j\) are complex conjugates on the unit circle centered at the complex 0:

$$\begin{aligned} z^\pm _j=e^{\pm i\theta _{j}},\quad \theta _{j}=\frac{j\pi }{n+1},\quad j=r,~s\in \{1,~2,~\ldots ,~n\}.\nonumber \\ \end{aligned}$$
(A6)

Because \(g\left( z ,a^{(r)} \right) \) is an entire function, the integrand in Eq. (A5) has a pole only at \(z=0\). Thus, the contour integral is invariant under deformation as long as C encloses 0 once counterclockwise. If we choose a contour C entirely inside the unit circle centered at the complex 0, then \(|z/z_j^\pm |<1\) at any point z on the contour. Hence, we can make a Taylor-series expansion about \(z=0\) as

$$\begin{aligned} \frac{1}{z-z_j^\pm }=-\frac{1}{z_j^\pm } \sum _{k=0}^\infty e^{\mp ik\theta _{j}}z^k,\quad j=r,~s\in \{1,~2,~\ldots ,~n\}.\nonumber \\ \end{aligned}$$
(A7)

In the integrand of Eq. (A5), we can discard any contributions that are analytic at \(w=0\) because they do not contribute to the residue by applying Cauchy integral formula in Eq. (3). As a result, Eq. (A5) collapses into the sum of the following elementary residues:

$$\begin{aligned}&\!\!\! \frac{\sum _{k=1}^n a^{(r)}a^{(s)}}{\sin \theta _{r}\sin \theta _{s}}\nonumber \\&\,\!\!\!\quad =\displaystyle (-1)^{s-1} \sum _{a=0}^\infty \sum _{b=0}^\infty \sum _{c=0}^\infty \sum _{d=0}^\infty \frac{1}{2\pi i}\oint _C \frac{\mathrm{d}z}{z} z^{1-n+a+b+c+d}e^{i\theta _{r}(a-b)} e^{i\theta _{s}(c-d)}\nonumber \\&\,\!\!\!\quad =\displaystyle (-1)^{s-1} \sum _{a=0}^\infty \sum _{b=0}^\infty \sum _{c=0}^\infty \sum _{d=0}^\infty \delta _{a+b+c+d,n-1}e^{i\theta _{r}(a-b)} e^{i\theta _{s}(c-d)} \nonumber \\&\,\quad = \displaystyle (-1)^{s-1} \sum _{a=0}^{n-1}\quad \sum _{b=0}^{n-1-a}\quad \sum _{c=0}^{n-1-a-b}e^{i\theta _{r}(a-b)} e^{i\theta _{s}(2c+a+b-n+1)} \nonumber \\&\,\quad = \displaystyle \frac{(-1)^{s-1}}{2[\cos \theta _{r}-\cos \theta _{s}]}\left[ \frac{\sin (n+1)\theta _{r}}{\sin \theta _{r}} - \frac{\sin (n+1)\theta _{s}}{\sin \theta _{s}}\right] . \end{aligned}$$
(A8)

If \(r\ne s\), then the denominator factors for the two terms \([\cos \theta _{r}-\cos \theta _{s}]\sin \theta _{r}\) and \([\cos \theta _{r}-\cos \theta _{s}]\sin \theta _{s}\) are both non-vanishing. Because the numerator is always vanishing due to the constraints \(\sin (n+1)\theta _{r}=\sin (n+1)\theta _{s}=0\), we find that

$$\begin{aligned} \left. \frac{\sum _{k=1}^n a^{(r)}a^{(s)}}{\sin \theta _{r}\sin \theta _{s}} \right| _{r\ne s} =0. \end{aligned}$$
(A9)

If \(r= s\), then both the numerator and the denominator vanish simultaneously. Thus, we apply L’Hôpital’s rule by taking the ratio of the first-order derivatives of the numerator and the denominator and applying the identities \(\sin (n+1)\theta _{r}=0\) and \(\cos (n+1)\theta _{s}=(-1)^s\) in Eq. (36). Equivalently, we can replace \(\theta _{s}\) with \(\theta _{s} +x\) and take the Taylor-series expansion of both the numerator and the denominator to find that

$$\begin{aligned}&\displaystyle \left. \frac{\sum _{k=1}^n a^{(r)}a^{(s)}}{\sin \theta _{r}\sin \theta _{s}} \right| _{r= s}\nonumber \\&\quad = \displaystyle \frac{(-1)^{s-1} }{2}\lim _{x\rightarrow 0} \frac{ \displaystyle \frac{\sin [(n+1)(\theta _{r}+x)]}{\sin (\theta _{r}+x)} - \frac{\sin (n+1)\theta _{r}}{\sin \theta _{r}} }{ \cos (\theta _{r}+x)-\cos \theta _{r} } \nonumber \\&\quad =\displaystyle \frac{(-1)^{s-1} }{2}\lim _{x\rightarrow 0} \frac{ (n+1)x \sin \theta _{r} \cos [(n+1)\theta _{r}] }{\displaystyle -x\sin ^3\theta _{r} } \nonumber \\&\quad = \displaystyle \frac{ n+1 }{2\sin ^2\theta _{r} }. \end{aligned}$$
(A10)

Combining the results in Eqs. (A9) and (A10), we conclude that

$$\begin{aligned} \sum _{k=1}^n&\sin \frac{rk\pi }{n+1}\sin \frac{sk\pi }{n+1}=\frac{n+1}{2}\,\delta _{rs},\nonumber \\&r,\,s=1,~2,~\ldots ,~n. \end{aligned}$$
(A11)

This completes the proof of Eq. (55).

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Kim, UR., Jung, DW., Kim, D. et al. Generating-function representation for scalar products. J. Korean Phys. Soc. 79, 429–437 (2021). https://doi.org/10.1007/s40042-021-00227-7

Download citation

  • Received:

  • Revised:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s40042-021-00227-7

Keywords

Navigation