A Dimension Abstraction Approach to Vectorization in Matlab

Abstract

Matlab is a matrix-processing language that offers very efficient built-in operations for

Compile-Time Disambiguation of MATLAB Types through Concrete Interpretation with Automatic Run-Time Fallback

“…Internally, however, MATLAB documentation describes a much finer granularity of types, presumably designed to be compatible with the Java interface [8]. Because variables in MATLAB programs can be of varying size and type and MATLAB operators are often heavily overloaded, the benefits of performing static type inference whenever possible are well known [11], [1], [4], [10], [7], [9], [3]. Figure 2 shows the order of magnitude performance improvement in the whole program achieved by optimizing a function called dlaplacian that is the performance bottleneck in a MATLAB version of the NAS MG benchmark [2].…”

Optimizing Matlab through Just-In-Time Specialization

“…Previous compiler approaches to Matlab have mainly focused on numerical performance, primarily in the context of static language subsets or contexts. As well as more traditional loop and array optimizations, code restructuring can be performed to ensure programs take good advantage of optimized instrinics [7]. Good performance can also be achieved by translating Matlab code to other static languages, such as C [8] or Fortran 90 [6,9], where further aggressive optimization or parallelization can be performed.…”

Fast Quadratic Programming for Mean-Variance Portfolio Optimisation

“…The technique of vectorisation has gained popularity due to its attractive feature regarding the reduction of an algorithm's complexity. In practice, albeit some degree of automation can be achieved [7,10], vectorisation requires additional resources for the appropriate transformation of a dataset into matrices that can be readily used by the corresponding algorithm. Very often, vectorisation is used to calculate integrals via quadrature-based methods [26] and solve differential equations [25,27].…”