Real lines on random cubic surfaces

From MaRDI portal



Abstract: We give an explicit formula for the expectation of the number of real lines on a random invariant cubic surface, i.e. a surface ZsubsetmathbbRP3 defined by a random gaussian polynomial whose probability distribution is invariant under the action of the orthogonal group O(4) by change of variables. Such invariant distributions are completely described by one parameter lambdain[0,1] and as a function of this parameter the expected number of real lines equals: �egin{equation} E_lambda=frac{9(8lambda^2+(1-lambda)^2)}{2lambda^2+(1-lambda)^2}left(frac{2lambda^2}{8lambda^2+(1-lambda)^2}-frac{1}{3}+frac{2}{3}sqrt{frac{8lambda^2+(1-lambda)^2}{20lambda^2+(1-lambda)^2}} ight). end{equation} This result generalizes previous results by Basu, Lerario, Lundberg and Peterson for the case of a Kostlan polynomial, which corresponds to lambda=frac13 and for which Efrac13=6sqrt2−3. Moreover, we show that the expectation of the number of real lines is maximized by random purely harmonic cubic polynomials, which corresponds to the case lambda=1 and for which E1=24sqrtfrac25−3.


The authors' paper belongs to the rapidly developing field of random real algebraic geometry that has its origin in classical works of \textit{M. Kac} [Bull. Am. Math. Soc. 49, 314--320 (1943; Zbl 0060.28602)], \textit{A. Edelman} and \textit{E. Kostlan} [Bull. Am. Math. Soc., New Ser. 32, No. 1, 1--37 (1995; Zbl 0820.34038)] answering the question how many zeros of a random polynomial are real, and that was continued by \textit{M. Shub} and \textit{S. Smale} [Prog. Math. 109, 267--285 (1993; Zbl 0851.65031); SIAM J. Numer. Anal. 33, No. 1, 128--148 (1996; Zbl 0843.65035)] with their program of studying the real zeros of random polynomial systems. The paper at hand gives an explicit formula for the expectation \( E \) of the number of real lines on a real cubic surface in \( \mathbb{R}P^3 \) defined by a random gaussian polynomial \( f \) whose probability distribution is invariant under the action of the orthogonal group \( O(4) \) by change of variables. Such invariant distributions are completely described by one parameter \( \lambda\in [0, 1] \) and the expectation \( E \), as a function of \( \lambda \), equals \( E(\lambda) =\dfrac{9(8\lambda^2+(1-\lambda)^2)}{2\lambda^2+(1-\lambda)^2}\left(\dfrac{2\lambda^2}{8\lambda^2+(1-\lambda)^2}-\dfrac{1}{3}+\dfrac{2}{3}\sqrt{\dfrac{8\lambda^2+(1-\lambda)^2}{20\lambda^2+(1-\lambda)^2}}\right) . \) If \( \lambda=1/3 \) the polynomial \( f \) is [\textit{E. Kostlan}, In: From topology to computation: proceedings of the Smalefest (Berkeley, CA, 1990). Springer, New York, 419--431 (1993; Zbl 0788.60069)]. In this case \( E(1/3)=6\sqrt{2}-3 \). This result was obtained earlier by \textit{S. Basu} et al. [Math. Ann. 374, No. 3--4, 1773--1810 (2019; Zbl 1423.14303)]. The authors show that the function \( E(\lambda) \) is monotone increasing and is maximized by random purely harmonic cubic polynomials for which \( E(1) = 24\sqrt{2/5}- 3\).



Cites work









This page was built for publication: Real lines on random cubic surfaces

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2066765)