When calling a class (say, to convert types), are we calling a function?
Show older comments
I know that we can call built-in MATLAB classes and input values of any other built-in class into it with the same inputting syntax that we use for functions e.g. double('a'). Does this mean a class is a function? Or is it that a separate function is invoked when we call a class?
For example, what is the behind-the-scenes difference between calling sin as a function, and calling logical as a function? They both act like functions, and to make matters worse, both become function handles if we type @ in front of their identifiers:
isSinAfunction = isa(@sin,'function_handle')
isLogicalAfunction = isa(@logical, 'function_handle')
So is logical a function?
Accepted Answer
For example, what is the behind-the-scenes difference between calling sin as a function, and calling logical as a function?
Nothing, really. [There may be some technical scenarios that I'm not thinking of offhand where there are differences, but for the most part there should be no difference.]
They both act like functions, and to make matters worse, both become function handles if we type @ in front of their identifiers:
There's a good reason for that. The reason is that they're both functions, in this case both built-in functions.
which sin
which logical
So is logical a function?
Yes. Its purpose is to convert whatever you pass into it into a logical array. You can see this by looking at the documentation page for the logical function. Sometimes it does this by calling a built-in function, sometimes (if the class of its input has overloaded the function) it calls a class method.
I suspect what you may be trying to figure out is whether logical is a class constructor or a function. The answer is it falls more into the "conversion method" (or "conversion function") category, as described in the "Kinds of Methods" section on this documentation page. Some of the functions with the same name as a built-in class straddle the line between constructor and converter, the struct function jumps immediately to mind. But logical is IMO just a converter. If you were trying to construct a logical array (from size data) you'd be more likely to call true, false, or one of the array creation functions like ones or zeros.
9 Comments
Kyle
on 18 Dec 2023
[There may be some technical scenarios that I'm not thinking of offhand where there are differences, but for the most part there should be no difference.]
For those especially curious about these technical scenarios, the only thing that comes to mind is Dot Syntax vs Function Syntax. It doesn't apply in the examples that Ardy listed, but it does matter for object methods with more than one input, such as plus. If you want to dive in deeper to how MATLAB determines which functions to call, you can explore this example.
I'll also leave a note about my favorite way to allocate a large array of logicals, indexing! a(100,100) = false;
madhan ravi
on 18 Dec 2023
Hi Kyle,
a(100, 100) = false
or
a = false(100)
Is there any difference within MATLAB? or it’s just a matter of preference?
Kyle
on 18 Dec 2023
Hi madhan,
Great question! In general, if there is a builtin MATLAB function to initialize an array, it is best to use it. Syntax like a = false(100) is also less ambiguous to someone reading your code. You know that it will always create a new variable named a. Whereas using indexing to initialize the array will have different behavior if a already exists.
The times where using indexing might be better are if your algorithm actually wants different behavior the first time executing the assignment or if you are working with a data type like an object without its own builtin array creation function. But most people don't know this is even a valid way of initializing an array, so I always enjoy pointing it out!
Is there any difference within MATLAB?
Yes.
a = 42
a(2, 2) = false
class(a)
If the array already exists, extending it by assigning to a value in a location that doesn't yet exist doesn't change its type. It changes what you're trying to store in the array into the type of the array (in this case, false becomes 0.) For logical in most cases this doesn't make a difference, except if you were to try to use a for indexing. I'll put that code at the end of this comment as it will throw an error.
Compare:
b = 42
b = false(2)
This overwrites b (even though it existed already) and the new b is a logical array.
Indexing example:
a = magic(4)
ind = true(4)
ind([1 6 11 16]) = false
a(ind)
ind2 = ones(4)
ind2([1 16 11 16]) = false
a(ind2) % throws an error
madhan ravi
on 19 Dec 2023
Thanks Kyle and Steven.
Ardy
on 20 Dec 2023
Doesn't this shake your confidence that logical is just a converter function like any other function?
Why? The datatypes directory is not restricted to only containing data types. For example none of the functions whose names start with "is" are data types. They are querying functions. func2str and str2func are essentially conversion functions between text (char or string) and function handles. cast and typecast are general converter functions.
what(fullfile(matlabroot, 'toolbox', 'matlab', 'datatypes'))
Don't read that much into the directory names. Generally they group related functionality. But that can still lead to somewhat varying functions all being lumped in the same directory. For example, the elfun directory contains the trigonometric functions, the exponential and logarithm functions, functions for assembling and disassembling complex numbers, and rounding functions.
help elfun
The source of my confusion is when I compare MATLAB to how objects are constructed in Ptyhon, and over there, they "call" classes. In fact they say classes are "callable".
Okay. Do they mean that classes are "callable" (aka constructible, able to be instantiated as long as they're concrete?) via calling their constructor method or via factory functions/methods? If so, then they're behaving the same way as MATLAB. And if they mean something different, well if all programming languages were required to behave exactly the same way we'd all be programming in Short Code, Autocode, or FORTRAN.
Stephen23
on 20 Dec 2023
"In fact they say classes are "callable"."
They are not really.
In Python everything is an object. Those that are callable simply have a defined __call__ method:
Quoting that last page: "The __call__ method enables Python programmers to write classes where the instances behave like functions and can be called like a function. When the instance is called as a function; if this method is defined, x(arg1, arg2, ...) is a shorthand for x.__call__(arg1, arg2, ...)."
So even in Python, the class itself in not called, the __call__ method is.
Ardy
on 20 Dec 2023
More Answers (1)
Image Analyst
on 18 Dec 2023
3 votes
I like to think of it like this. In standard object oriented programming (OOP) you have a "class" which is an object that is a collection of "properties" and "methods". The "properties" are your variable names. And the "methods" are what MATLAB calls "functions".
So you don't really "call" a class. You "call" methods/functions of the class, and "assign" properties/variables of the class.
Categories
Find more on Call Python from MATLAB in Help Center and File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!
