The zeros $c_k\left(\nu \right)$ of the solution $z(t,\nu )$ of the differential equation $z^{\text{'}\text{'}}+q(t,\nu )z=0$ are investigated when $\underset{t\to \infty }{lim}q(t,\nu )=1$, ${\int }^{\infty }|q(t,\nu )-1|dt<\infty$ and $q(t,\nu )$ has some monotonicity properties as $t\to \infty$. The notion $c_{\kappa }\left(\nu \right)$ is introduced also for $\kappa$ real, too. We are particularly interested in solutions $z(t,\nu )$ which are “close" to the functions $sint$, $cost$ when $t$ is large. We derive a formula for $d{c}_{\kappa }\left(\nu \right)/d\nu$ and apply the result to Bessel differential equation, where we introduce new pair of linearly independent solutions replacing the usual pair $J_{\nu }\left(t\right)$, $Y_{\nu }\left(t\right)$. We show the concavity of $c_{\kappa }\left(\nu \right)$ for...

The non-conforming linear ($P_1$) triangular FEM can be viewed as a kind of the discontinuous Galerkin method, and is attractive in both the theoretical and practical purposes. Since various error constants must be quantitatively evaluated for its accurate a priori and a posteriori error estimates, we derive their theoretical upper bounds and some computational results. In particular, the Babuška-Aziz maximum angle condition is required just as in the case of the conforming $P_1$ triangle. Some...

In this paper, we investigate the a priori and the a posteriori error analysis for the finite element approximation to a regularization version of the variational inequality of the second kind. We prove the abstract optimal error estimates in the $H^1$- and $L_2$-norms, respectively, and also derive the optimal order error estimate in the $L_{\infty }$-norm under the strongly regular triangulation condition. Moreover, some residual–based a posteriori error estimators are established, which can provide the...

If $\left\{v_i\right\}$ and $\left\{w_j\right\}$ are two families of unitary bases for ${\mathbf{C}}^n$, and $\theta$ is a fixed number, let $V^n$ and $W^n$ be subspaces of ${\mathbf{C}}^n$ spanned by $[\theta ·n]$ vectors in $\left\{v_i\right\}$ and $\left\{w_j\right\}$ respectively. We study the angle between $V^n$ and $W^n$ as $n$ goes to infinity. We show that when $\left\{v_i\right\}$ and $\left\{w_j\right\}$ arise in certain arithmetically defined families, the angles between $V^n$ and $W^n$ may either tend to $0$ or be bounded away from zero, depending on the behavior of an associated eigenvalue problem.

We prove that the error term ${\sum }_{n\le x}\Lambda \left(n\right)/n-logx+\gamma$ differs from (ψ(x)-x)/x by a well controlled function. We deduce very precise numerical results from the formula obtained.

Soit $F$ un sous-ensemble compact de ${\mathbf{R}}^m$ ayant des points intérieurs et soit ${\mu }_{\alpha }^F$ la distribution d’équilibre sur $F$ de masse totale 1 par rapport au noyau $r^{\alpha -m}$ avec $0\&lt;\alpha \&lt;2$ pour $m\ge 2$, et $0\&lt;\alpha \&lt;1$ pour $m=1$. La restriction de ${\mu }_{\alpha }^F$ à l’intérieur de $F$ est absolument continue et a pour densité $f_{\alpha }^F$. On donne une formule explicite pour $f_{\alpha }^F$ et, pour une classe générale d’ensembles $F$, on démontre que $f_{\alpha }^F$, définie en réalité sur un ensemble de mesure de Lebesgue nulle, croît comme la distance à la frontière $\partial F$ de $F$ élevée à la puissance...