What’s New in MATLAB R2016b

MATLAB R2016b was released in the middle of September 2016. In this post I discuss some of its new features (I will not consider the toolboxes). This is personal selection of highlights; for a complete overview see the Release Notes.

The features below are discussed in greater detail in the third edition of MATLAB Guide, to be published by SIAM in December 2016.

Live Editor

The Live Editor, introduced in R2016a, provides an interactive environment for editing and running MATLAB code. When the code that you enter is executed the results (numerical or graphical) are displayed in the editor. The code is divided into sections that can be evaluated, and subsequently edited and re-evaluated, individually. The Live Editor works with live scripts, which have a .mlx extension. Live scripts can be published (exported) to HTML or PDF. R2016b adds more functionality to the Live Editor, including an improved equation editor and the ability to pan, zoom, and rotate axes in output figures​.

The Live Editor is particularly effective with the Symbolic Math Toolbox, thanks to the rendering of equations in typeset form, as the following image shows.

live-editor1.jpg

The Live Editor is a major development, with significant benefits for teaching and for script-based workflows. No doubt we will see it developed further in future releases.

Implicit Expansion

I have already written about this powerful generalization of the long-standing MATLAB feature of scalar expansion. See this blog post.

Local Functions

Local functions are what used to be called subfunctions: they are functions within functions or, to be more precise, functions that appear after the main function in a function file. What’s new in R2016b is that a script can have local functions. This is a capability I have long wanted. When writing a script I often find that a particular computation or task, such as printing certain statistics, needs to be repeated at several places in the script. Now I can put that code in a local function instead of repeating the code or having to create an external function for it.

String Arrays

MATLAB has always had strings, created with a single quote, as in

s = 'a string'

which sets up a 1-by-8 character vector. A new datatype string has been introduced. A string is an array for which each entry contains a character vector. The syntax

str = string('a string')

sets up a 1-by-1 string array, whereas

str = string({'Boston','Manchester'})

sets up a 1-by-2 string array via a cell array. String arrays are more memory efficient than cell arrays and allow for more flexible handling of strings. They are particularly useful in conjunction with tables. According to this MathWorks video, string arrays “have a multitude of new methods that have been optimized for performance”. At the moment, support for strings across MATLAB is limited and it is inconvenient to have to set up a string array by passing a cell array to the string function. No doubt string arrays will be integrated more seamlessly in future releases.

Tall Arrays

Tall arrays provide a way to work with data that does not fit into memory. Calculations on tall arrays are delayed until a result is explicitly requested with the gather function. MATLAB optimizes the calculations to try to minimize the number of passes through the data. Tall arrays are created with the tall function, which take as argument an array (numeric, string, datetime, or one of several other data types) or a datastore. Many MATLAB functions (but not the linear algebra functions, for example) work the same way with tall arrays as they do with in-memory arrays.

Timetables

Tables were introduced in R2013b. They store tabular data, with columns representing variables of possibly different types. The newly introduced timetable is a table for which each row has an associated date and time, stored in the first column. The timerange function produces a subscript that can be used to select rows corresponding to a particular time interval. These new features make MATLAB an even more powerful tool for data analysis.

Missing Values

MATLAB has new capabilities for dealing with missing values, which are defined according to the data type: NaNs (Not-a-Number) for double and single data, NaTs (Not-a-Time) for datetime data, <undefined> for categorical data, and so on. For example, the function ismissing detects missing data and fillmissing fills in missing data according to one of several rules.

Mac OS Version

I have Mac OS X Version 10.9.5 (Mavericks) on my Mac. Although this OS is not officially supported, R2016b installed without any problem and runs fine. The pre-release versions of R2016a and R2016b would not install on Mavericks, so it seems that compatibility with older operating systems is ensured only for the actual release.

At the time of writing, there are some compatibility problems with Mac OS Version 10.12 (Sierra) for certain settings of Language & Region.

This entry was posted in software and tagged . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s