This paper basically does what the title says. It makes two distinct uses of MATLAB, one symbolic and one numeric, to solve, in a meshless manner, the plate-bending problem. The symbolic code writes 50 MATLAB functions, typically a few hundred characters of dense formula each, which are used by the numeric code.
The fundamental conclusion: “The Symbolic.m program takes a few seconds to run and a few minutes for the programmer to adapt it to a new shear deformation theory,” as contrasted with “many days” by hand. The time penalty in the numerical part is negligible for large problems. While the author finds the fact that he can use MATLAB for both phases significant, I feel that this is largely incidental. Indeed, the numerical time penalty could probably be obviated by compiling the numeric code in, say, Fortran.
There are various problems with the rendering of the MATLAB: continuation symbols are missing, and what should be single quote characters are printed as forward and backward quotes in some MATLAB fragments. Also, the word “resultant,” which has a technical meaning in algebra, especially computer algebra, is being used non-technically to mean, essentially, intermediate quantities.