1: \begin{abstract}

2: The strong-form asymmetric kernel-based collocation method, commonly referred to

3: as the Kansa method, is easy to implement and hence is widely used for solving

4: engineering problems and partial differential equations despite the lack of

5: theoretical support. The simple least-squares (LS) formulation, on the other hand,

6: makes the  {study}

7: of its solvability and convergence

8:  rather nontrivial.

9: %In our previous works, we showed how  Kansa's formulation should be modified to

10: %guarantee its solvability. We also showed

11: %\blue{its convergence in $L^2(\Omega)$}.

12: %%the convergence of an $L^\infty$-minimization formulation, whose solution process requires linear programming solvers.

13: %In this paper, we focus on the more welcoming and numerically accessible least-squares approaches. 

14: %Our analysis is carried out for

15: In this paper, we focus on

16: general second order linear elliptic differential equations in $\Omega \subset

17: \R^d$

18: {under} Dirichlet boundary conditions.

19: With kernels that reproduce $H^m(\Omega)$ and some smoothness assumptions

20:  {on the solution}, we provide  denseness

21: {conditions}

22: for a constrained least-squares  {method} and

23: a  {class of}  weighted least-squares

24:  {algorithms} to be convergent.

25: %Our analysis covers two trial spaces. 

26: Theoretically,

27: we identify  some $H^2(\Omega)$ convergent  LS formulations

28: that {have}  {an} optimal error

29: {behavior like } $h^{m-2}$.

30: We also demonstrate the effects of various collocation settings

31:  {on the respective}

32: convergence rates, as well as how these

33: formulations perform with high order

34: kernels and when coupled with the stable

35: evaluation technique for the Gaussian kernel.

36: 

37: \end{abstract}