1: \begin{abstract}

2: We develop algorithms for multivariate integration and approximation in

3: the weighted half-period cosine space of smooth non-periodic functions.

4: %of $d$-variables.

5: %The parameter $\alpha$ characterizes the smoothness of the space by

6: %controlling the decay of the spectral coefficients in the $L_2$ norm. The

7: %weights $\bsgamma = (\gamma_j)$ moderate the relative importance between

8: %successive coordinate directions.

9: We use specially constructed tent-transformed rank-$1$ lattice points as

10: cubature nodes for integration and as sampling points for approximation.

11: For both integration and approximation, we study the connection between

12: the worst-case errors of our algorithms in the cosine space and the

13: worst-case errors of some related algorithms in the well-known weighted

14: Korobov space of smooth periodic functions. By exploiting this connection,

15: we are able to obtain constructive worst-case error bounds with good

16: convergence rates for the cosine space.

17: 

18: \smallskip

19: \noindent \textbf{Keywords:} \textit{Quasi-Monte Carlo methods, Cosine series, Function approximation, Hyperbolic crosses, Rank-$1$ lattice rules, Spectral methods, Component-by-component construction.}

20: 

21: \smallskip

22: \noindent \textbf{Subject Classification:} 65D30, %Numerical integration

23: 65D32, %Quadrature and cubature formulas

24: 65C05,  %Monte Carlo methods

25: 65M70   %	Spectral, collocation and related methods

26: 65T40  % 	Trigonometric approximation and interpolation

27: %65T50   %	Discrete and fast Fourier transforms

28: \end{abstract}

29: