Results for
In a previous discussion,
we looked at a variety of infallible tests for primality, but all of them were too slow to be viable for large numbers. In fact, all of the methods discussed there will fail miserably for even moderately large numbers, those with just a few dozen decimal digits. That does not even begin to push into the realm of large numbers. In turn, that forces us into a different realm of tests - tests which are usually and even almost always correct, but can sometimes incorrectly predict primality.
In this discussion, I will be trying to convince you that the Fermat test for primality can be a quite good test when the number is sufficiently large. Except of course, when it is really bad. Even so, the Fermat test for primality is both a useful tool as well as a necessary underpinning for several other better tests.
The Fermat test for primality relies on Fermat's little theorem, a perhaps under-appreciated tool. Even the name implies it is of little interest. I'm not taking about the famous last theorem, only proven in recent years, but his little theorem.
If you want to look over some nice ways to prove the little theorem, take a read in this link:
I will readily admit that long ago, when I learned about the little theorem in a nearly forgotten class, I thought it was interesting, but why would I care? Not until I learned more mathematics and saw Fermat’s little theorem appearing in different places did I begin to appreciate it. Fermat tells us that, IF P is a prime, AND w is co-prime with P (so the two are relatively prime, sharing no common factors except 1), then it must be true that
mod(w^(P-1),P) == 1
Try it out. Does it work? Be careful though as too large of an exponent will cause problems in double precision, and that is not difficult to do. As a test case that will not overwhelm doubles, note that 13 is prime, and 3 shares no common factors with 13, so we satisfy the requirements for Fermat's little theorem.
mod(3^12,13)
We can even verify that any co-prime of 13 will yield the same result.
mod((1:12).^12,13)
Indeed it worked, suggesting what we knew all along, that 13 is prime. The little Fermat test for primality of the number P uses a converse form of Fermat's little theorem, thus given a co-prime number w known as the witness, is that if
mod(w^(P-1),P)==1
then we have evidence that P is indeed prime. This is not conclusive evidence, but still it is evidence. It is not conclusive because the converse of a true statement is not always true.
The analogy I like here is if we lived in a universe where all crows are black. (I'll ask you to pretend this is true. In fact, some crows have a mutation, making them essentially albino crows. For the purposes of this thought experiment, pretend this cannot happen.) Now, suppose I show you a picture of a bird which happens to be black. Do you know the bird to be a crow? Of course not, as the bird may be a raven, or a redwing blackbird (until you see the splash of red on the wing), a common grackle, a European starling in summer plumage, a condor, etc. But having seen black plumage, it is now more likely the bird is indeed a crow. I would call this a moderately weak evidentiary test for crow-ness.
Little Fermat may seem to be of little value when testing for primality for two reasons. First, computing the remainder would seem to be highly CPU intensive for large P. In the example above, I had only to compute 3^12=531441, which is not that large. But for numbers with many thousands or millions of digits, directly raising even a number as small as 2 to that power will overwhelm any computer. Secondly, if we do that computation, little Fermat does not conclusively prove P to be prime.
Our savior in one respect is the powermod tool. And that helps greatly, since we can compute the remainder in a reasonable time. A powermod call is quite fast even for huge powers. (I won't get into how powermod works here, since that alone is probably worth a discussion. I could though, if I see some interest because there are some very pretty variations of the powermod algorithm. I hope to show you one of them when I discuss the Fibonacci test for primality in a future post.) Trying the little Fermat test using powermod on a number with 1207 decimal digits, I’ll first do a time check.
P = 4000*sym(2)^3999 - 1;
timeit(@() powermod(2,P-1,P))
As you can see, powermod really is pretty fast. Compared to an isprime test on that number it would show a significant difference.
I have said before that little Fermat is not a conclusive test. In fact, a good trick is to perform a second little Fermat test, using a different witness. If the second test also indicates primality, then we have additional evidence that P is in fact prime.
w = [2 3]; % Two parallel witnesses
powermod(w,P-1,P)
This value for P is indeed prime, and little Fermat suggests it is, doubly suggestive in that test since I actually performed two parallel tests. Here however, we need to understand when it will fail, and how often it will fail.
If we perform a little Fermat test for primality, we will never see false negatives, that is, if any test with any witness ever indicates a number is composite, then it is certainly composite. (The contrapositive of a true statement is always true.) The alternate class of failure is the false positive, where little Fermat indicates a number is prime when it was actually composite.
If P is composite, and w co-prime with P, we call P a Fermat pseudo-prime for the witness w if we see a remainder of 1 when P was in fact composite. When that happens, the witness (w) is called a Fermat liar for the primality of P. (A list of some Fermat pseudo-primes where 2 is a Fermat liar can be found in sequence A001567 of the OEIS.)
In the case of 4000*2^3999-1, I claimed the number to be in fact prime, and it was identified so (as PROBABLY prime by little Fermat. Next, consider another number from that same family. I’ll perform three parallel tests on it, with witnesses 2, 3, and 5. This will suggest the value of doing parallel tests on a number to reduce the failure rate from little Fermat.
P2 = 1024*sym(2)^1023 - 1;
w = [2; 3; 5];
gcd(w,P2)
F2 = powermod(w,P2-1,P2)
logical(F2 == 1)
As you can see, P2 is co-prime with all of 2, 3 and 5, but 2 is a Fermat liar, whilst 3 and 5 are Fermat truth tellers, identifying P2 as certainly composite. So the little Fermat test can definitely fail for SOME witnesses, since we see a disagreement. However, an interesting fact about P2 above is it is also a Mersenne number with prime exponent, thus it can be written as 2^1033-1, where 1033 is prime. I can go into more detail about this case later, but we can show that 2 is always a Fermat liar for composite Mersenne numbers when the exponent is itself prime. I’ll try to leave more detail on this matter in a future discussion, or perhaps a comment.
Next, consider the composite integer 51=3*17. As the product of two primes, it is clearly not itself prime.
P51 = 51;
w0 = 2:P51-2;
w = w0(gcd(w0,P51) == 1)
Note that I did not include 1 or 50 in that set, since 1 raised to any power is 1, and 50 is congruent to -1, mod 51. -1 raised to any even power is also always 1, and 51-1 is an even number. And so when we are working modulo 51, both 1 and 50 are not useful witnesses in terms of the little Fermat test.
R = powermod(w,P-1,P)
w(R == 1)
This teaches us that when 51 is tested for primality using the little Fermat test, there are 2 distinct witnesses w (16 and 35) we could have chosen which would have been Fermat liars, but all 28 other potential co-prime witnesses would have been truth tellers, accurately showing 51 to be composite. Proportionally, little Fermat will have been correct roughly 93% of the time, since only 2 of these 30 possible tests returned a false positive. (I’ll add that for any modulus P, if w<P is not co-prime with the modulus, then the computation mod(w^(P-1),P) will always return a non-unit result, and therefore we can theoretically use any integer w from the set 2:P-2 as a witness. However if w is co-prime with P then P is clearly not prime, and the entire problem becomes a little less interesting. As such, I will only consider co-prime witnesses for this discussion.) Regardless, that would make the little Fermat test for P=51 even more often correct, since it returns the correct result of composite for 46 out of the 48 possible witnesses 2:49. Does this mean Little Fermat is indeed the basis for a good test to rely on to learn if a number is prime? Well, yes. And no.
Little Fermat forms a very good test most of the time, but reliance is a strong word. This means we need to explore the little Fermat test in more depth, focusing on Fermat liars and the case of false positives. To offer some appreciation of the false positive rate, offline, I have tested all composite integers between 4 and 10000, for all their possible co-prime witnesses.
load FermatLiarsData
In that .mat file, I've saved three vectors, X, witnessCount, and liarCount. X is there just to use for plotting purposes and is NaN for all non-composite entries.
whos X witnessCount liarCount
The vector witnessCount is the number of valid witnesses for the corresponding number in X. Corresponding to that is the vector liarCount, which is the number of Fermat liars I found for each composite in X.
How many Fermat test useful witnesses are there for any integer X? This is just 2 less than the number of coprimes of X. The number of coprimes is given by the Euler totient function, commonly called phi(X). (I’ll be going into more depth on the totient in the next chapter of this series, because the Euler totient is a crucial part of understanding how all of this works.)
The witness count is phi(X)-2. Why subtract 2? 1 can never be a witness, but 1 is technically coprime to everything. The same applies to X-1 (which is congruent to -1 mod X.) As such, there are phi(X)-2 coprimes to consider. (I've posted a function called totient on the FEX, but it is easily computed if you know the factorization of X. Or for small numbers, you can just use GCD to identify all co-primes, and count them.)
plot(X,witnessCount,'.')
From that plot, you can learn a few interesting things. (As a mathematician, this is what I love the most, thus to look at whay may be the simplest, most boring plot, and try to find something of value, something I had never thought of before.) For example, we know that when X is prime, then everything from the set 2:X-2 is a valid witness. So the upper boundary on that plot will be the line y==x. As well, there are a few numbers where the order of the set of witnesses will be close to the maximum possible. For example, 961=31*31, has 928 valid witnesses. That makes some sense, as 961 is the square of a prime (31), so we know 961 is divisible only by 31. Only multiples of 31 will not be coprime with 961.
But how about the lower boundary? The least number of valid witnesses will always come from highly composite numbers, because they will share common factors with almost everything. For example 30 = 2*3*5, or 210=2*3*5*7.
witnessCount([30 210 420])
A good discussion about the lower bound for that plot can be found here:
What really matters to us though, is the fraction of the useful witnesses for a little Fermat test that yield a false positive.
plot(X,liarCount./witnessCount,'b.')
Cind = liarCount == witnessCount;
hold on
plot(X(Cind),1,'ro')
ylabel('Liar fraction')
xlabel('X')
title('Fermat pseudo-prime strength')
hold off
A look at this plot shows seven circles in red, corresponding to X from the list {561, 1105, 1729, 2465, 2821, 6601, 8911} which are collectively known as Carmichael numbers. These are numbers where all witnesses return a false positive. Carmichael numbers are themselves fairly rare. You can find a list of them as sequence A002997 in the OEIS. And for those of you who have never wandered around the OEIS, please take this opportunity to do so now. The OEIS stands for Online Encyclopedia of Integer Sequences. It contains a wealth of interesting knowledge about integers and integer sequences.)
There are a few other interesting numbers we can find in that plot, like 91 and 703, where roughly 50% of the valid witnesses yield false positives. Of the complete set, which numbers did return at least a 25% false positive rate for primality? These numbers would be known as strong pseudo-primes for the little Fermat test, because they are pseudo-primes for at least 25% of the potential witnesses. These strong pseudo-primes have some interesting similarities to the Carmichael numbers. (My next post will go into more depth on Carmichael numbers and strong pseudo-primes. At the moment, I am merely interested in looking at the how often the little Fermat test fails overall.)
find(liarCount./witnessCount> 0.25)
You should notice the spacing between successive strong Fermat pseudo-primes is growing slowly, with a spacing of roughly 800 on average in the vicinity of 10000. If I step out beyond by a factor of 10, the next strong Fermat pseudo-primes after 1e5 are {101101, 104653, 107185, 109061, 111361, 114589, 115921 126217, 126673}, so that average spacing is definitely growing.
Given that set, now we can look at the prime factorizations of each of those strong pseudo-primes. Can we learn something about them?
arrayfun(@factor,find(liarCount./witnessCount> 0.25),'UniformOutput',false)
Perhaps the most glaring thing I see in that set of factors is almost all of those strong Fermat pseudo-primes are square free. That is, in that list, only 45=3*3*5 had any replicated factor at all. That property of being square free is something we will see is necessary to be a Carmichael number, but it also suggests that a simple roughness test applied in advance would have eliminated almost all of those strong pseudo-primes as obviously not prime, even at a very low level of roughness.
In fact, for most composite integers, most witnesses do indeed return a negative, indicating the number is not prime, and therefore composite. Little Fermat does not commonly tell falsehoods, even though it can do so.
semilogy(X,movmedian(liarCount./witnessCount,20,'omitnan'),'b-')
title('False positive fraction for composites')
yline(0.0003,'r')
We can learn from this last plot that as the number to be tested grows large, the median false positive rate for little Fermat, even for X as low as only 10000, is roughly 0.0003. (It continues to decrease for larger X too. In fact, I’ve read that when X is on the order of 2^256, the relative fraction of Fermat liars is on the order of 1 in 1e12, and it continues to decrease as X grows in magnitude. In my eyes, that seems pretty good for an imperfect test. Not perfect, but not bad when paired with roughness and perhaps a second little Fermat test using a different witness, and we will start to see tests which bear a higher degree of strength.)
I’ll stop at this point in this post because the post is getting lengthy. In my next post, I’d like to visit some questions about what are Carmichael numbers, about whether some witnesses are better than others, and if there are any numbers which lack any Fermat liars. However, in order to dive more deeply, I will need to explain how/why/when the little Fermat test works, and what causes Fermat liars. Stay tuned, because this starts to get interesting.
Dear all,
Recently I started working on a VS Code-style integrated terminal for the MATLAB IDE.
The terminal is installed as an app and runs inside a docked figure. You can launch the terminal by clicking on the app icon, running the command integratedTerminal or via keyboard shortcut.
It's possible to change the shell which is used. For example, I can set the shell path to C://Git//bin//bash.exe and use Git Bash on Windows. You can also change the theme. You can run multiple terminals.
I hope you like it and any feedback will be much appreciated. As soon as it's stable enough I can release it as a toolbox.
I was reading Yann Debray's recent post on automating documentation with agentic AI and ended up spending more time than expected in the comments section. Not because of the comments themselves, but because of something small I noticed while trying to write one. There is no writing assistance of any kind before you post. You type, you submit, and whatever you wrote is live.
For a lot of people that is fine. But MATLAB Central has users from all over the world, and I have seen questions on MATLAB Answers where the technical reasoning is clearly correct but the phrasing makes it hard to follow. The person knew exactly what they meant. The platform just did not help them say it clearly.
I want to share a few ideas around this. They are not fully formed proposals but I think the direction is worth discussing, especially given how much AI tooling MathWorks has built recently.
What the platform has today
When you write a post in Discussions or an answer in MATLAB Answers, the editor gives you basic formatting options. Code blocks, some text styling, that is mostly it. The AI Chat Playground exists as a separate tool, and MATLAB Copilot landed in R2025a for the desktop. But none of that is inside the editor where people actually write community content.
Four things are missing that I think would make a real difference.
Grammar and clarity checking before you post
Not a forced rewrite. Just an optional Check My Draft button that highlights unclear sentences or anything that might trip a reader up. The user reviews it, decides what to change, then posts.
What makes this different from plugging in Grammarly is that a general-purpose tool does not know that readtable is a MATLAB function. It does not know that NaN, inf, or linspace are not errors. A MATLAB-aware checker could flag things that generic tools miss, like someone writing readTable instead of readtable in a solution post.
The llms-with-matlab package already exists on GitHub. Something like this could be built on top of it with a prompt that includes MATLAB function vocabulary as context. That is not a large lift given what is already there.
Translation support
MATLAB Central already has a Japanese-language Discussions channel. That tells you something about the community. The platform is global but most of the technical content is in English, and there is a real gap there.
Two options that would help without being intrusive:
- Write in your language, click Translate, review the English version, then post. The user is still responsible for what goes live.
- A per-post Translate button so readers can view content in a language they are more comfortable with, without changing what is stored on the platform.
A student who has the right answer to a MATLAB Answers question might not post it because they are not confident writing in English. Translation support changes that. The community gets the answer and the contributor gets the credit.
In-editor code suggestions
When someone writes a solution post they usually write the code somewhere else, test it, copy it, paste it, and format it manually. An in-editor assistant that generates a starting scaffold from a plain-text description would cut that loop down.
The key word is scaffold, not a finished answer. The label should say something like AI-generated draft, verify before posting so it is clear the person writing is still accountable. MATLAB Copilot already does something close to this inside the desktop editor. Bringing a lighter version of it into the community editor feels like a natural extension of what already exists.
A note on feasibility
These ideas are not asking for something from scratch. MathWorks already has llms-with-matlab, the MCP Core Server, and MATLAB Copilot as infrastructure. Grammar checking and translation are well-solved problems at the API level. The MATLAB-specific vocabulary awareness is the part worth investing in. None of it should be on by default. All of it should be opt-in and clearly labeled when it runs.
One more thing: diagrams in posts
Right now the only way to include a diagram in a post is to make it externally and upload an image. A lightweight drag-and-drop diagram tool inside the editor would let people show a process or structure quickly without leaving the platform. Nothing complex, just boxes and arrows. For technical explanations it is often faster to draw than to write three paragraphs.
What I am curious about
I am a Data Science student at CU Boulder and an active MATLAB user. These ideas came up while using the platform, not from a product roadmap. I do not know what is already being discussed internally at MathWorks, so it is entirely possible some of this is in progress.
Has anyone else run into the same friction points when writing on MATLAB Central? And for anyone at MathWorks who works on the community platform, is the editor something that gets investment alongside the product tools?
Happy to hear where I am wrong on the feasibility side too.
AI assisted with grammar and framing. All ideas and editorial decisions are my own.
Over the past few days I noticed a minor change on the MATLAB File Exchange:
For a FEX repository, if you click the 'Files' tab you now get a file-tree–style online manager layout with an 'Open in new tab' hyperlink near the top-left. This is very useful:
If you want to share that specific page externally (e.g., on GitHub), you can simply copy that hyperlink. For .mlx files it provides a perfect preview. I'd love to hear your thoughts.
EXAMPLE:
🤗🤗🤗
I recently created a short 5-minute video covering 10 tips for students learning MATLAB. I hope this helps!
A coworker shared with me a hilarious Instagram post today. A brave bro posted a short video showing his MATLAB code… casually throwing 49,000 errors!
Surprisingly, the video went virial and recieved 250,000+ likes and 800+ comments. You really never know what the Instagram algorithm is thinking, but apparently “my code is absolutely cooked” is a universal developer experience 😂
Last note: Can someone please help this Bro fix his code?
Currently, the open-source MATLAB Community is accessed via the desktop web interface, and the experience on mobile devices is not very good—especially switching between sections like Discussion, FEX, Answers, and Cody is awkward. Having a dedicated app would make using the community much more convenient on phones.
Similarty,github has Mobile APP, It's convient for me.
You may have come across code that looks like that in some languages:
stubFor(get(urlPathEqualTo("/quotes"))
.withHeader("Accept", equalTo("application/json"))
.withQueryParam("s", equalTo(monitoredStock))
.willReturn(aResponse())
.withStatus(200)
.withHeader("Content-Type", "application/json")
.withBody("{\\"symbol\\": \\"XYZ\\", \\"bid\\": 20.2, " + "\\"ask\\": 20.6}")))
That’s Java. Even if you can’t fully decipher it, you can get a rough idea of what it is supposed to do, build a rather complex API query.
Or you may be familiar with the following similar and frequent syntax in Python:
import seaborn as sns
sns.load_dataset('tips').sample(10, random_state=42).groupby('day').mean()
Here’s is how it works: multiple method calls are linked together in a single statement, spanning over one or several lines, usually because each method returns the same object or another object that supports further calls.
That technique is called method chaining and is popular in Object-Oriented Programming.
A few years ago, I looked for a way to write code like that in MATLAB too. And the answer is that it can be done in MATLAB as well, whevener you write your own class!
Implementing a method that can be chained is simply a matter of writing a method that returns the object itself.
In this article, I would like to show how to do it and what we can gain from such a syntax.
Example
A few years ago, I first sought how to implement that technique for a simulation launcher that had lots of parameters (far too many):
lauchSimulation(2014:2020, true, 'template', 'TmplProd', 'Priority', '+1', 'Memory', '+6000')
As you can see, that function takes 2 required inputs, and 3 named parameters (whose names aren’t even consistent, with ‘Priority’ and ‘Memory’ starting with an uppercase letter when ‘template’ doesn’t).
(The original function had many more parameters that I omit for the sake of brevity. You may also know of such functions in your own code that take a dozen parameters which you can remember the exact order.)
I thought it would be nice to replace that with:
SimulationLauncher() ...
.onYears(2014:2020) ...
.onDistributedCluster() ... % = equivalent of the previous "true"
.withTemplate('TmplProd') ...
.withPriority('+1') ...
.withReservedMemory('+6000') ...
.launch();
The first 6 lines create an object of class SimulationLauncher, calls several methods on that object to set the parameters, and lastly the method launch() is called, when all desired parameters have been set.
To make it cleared, the syntax previously shown could also be rewritten as:
launcher = SimulationLauncher();
launcher = launcher.onYears(2014:2020);
launcher = launcher.onDistributedCluster();
launcher = launcher.withTemplate('TmplProd');
launcher = launcher.withPriority('+1');
launcher = launcher.withReservedMemory('+6000');
launcher.launch();
Before we dive into how to implement that code, let’s examine the advantages and drawbacks of that syntax.
Benefits and drawbacks
Because I have extended the chained methods over several lines, it makes it easier to comment out or uncomment any one desired option, should the need arise. Furthermore, we need not bother any more with the order in which we set the parameters, whereas the usual syntax required that we memorize or check the documentation carefully for the order of the inputs.
More generally, chaining methods has the following benefits and a few drawbacks:
Benefits:
- Conciseness: Code becomes shorter and easier to write, by reducing visual noise compared to repeating the object name.
- Readability: Chained methods create a fluent, human-readable structure that makes intent clear.
- Reduced Temporary Variables: There's no need to create intermediary variables, as the methods directly operate on the object.
Drawbacks:
- Debugging Difficulty: If one method in a chain fails, it can be harder to isolate the issue. It effectively prevents setting breakpoints, inspecting intermediate values, and identifying which method failed.
- Readability Issues: Overly long and dense method chains can become hard to follow, reducing clarity.
- Side Effects: Methods that modify objects in place can lead to unintended side effects when used in long chains.
Implementation
In the SimulationLauncher class, the method lauch performs the main operation, while the other methods just serve as parameter setters. They take the object as input and return the object itself, after modifying it, so that other methods can be chained.
classdef SimulationLauncher
properties (GetAccess = private, SetAccess = private)
years_
isDistributed_ = false;
template_ = 'TestTemplate';
priority_ = '+2';
memory_ = '+5000';
end
methods
function varargout = launch(obj)
% perform whatever needs to be launched
% using the values of the properties stored in the object:
% obj.years_
% obj.template_
% etc.
end
function obj = onYears(obj, years)
assert(isnumeric(years))
obj.years_ = years;
end
function obj = onDistributedCluster(obj)
obj.isDistributed_ = true;
end
function obj = withTemplate(obj, template)
obj.template_ = template;
end
function obj = withPriority(obj, priority)
obj.priority_ = priority;
end
function obj = withMemory( obj, memory)
obj.memory_ = memory;
end
end
end
As you can see, each method can be in charge of verifying the correctness of its input, independantly. And what they do is just store the value of parameter inside the object. The class can define default values in the properties block.
You can configure different launchers from the same initial object, such as:
launcher = SimulationLauncher();
launcher = launcher.onYears(2014:2020);
launcher1 = launcher ...
.onDistributedCluster() ...
.withReservedMemory('+6000');
launcher2 = launcher ...
.withTemplate('TmplProd') ...
.withPriority('+1') ...
.withReservedMemory('+7000');
If you call the same method several times, only the last recorded value of the parameter will be taken into acount:
launcher = SimulationLauncher();
launcher = launcher ...
.withReservedMemory('+6000') ...
.onDistributedCluster() ...
.onYears(2014:2020) ...
.withReservedMemory('+7000') ...
.withReservedMemory('+8000');
% The value of "memory" will be '+8000'.
If the logic is still not clear to you, I advise you play a bit with the debugger to better understand what’s going on!
Conclusion
I love how the method chaining technique hides the minute detail that we don’t want to bother with when trying to understand what a piece of code does.
I hope this simple example has shown you how to apply it to write and organise your code in a more readable and convenient way.
Let me know if you have other questions, comments or suggestions. I may post other examples of that technique for other useful uses that I encountered in my experience.
I struggle with animations. I often want a simple scrollable animation and wind up having to export to some external viewer in some supported format. The new Live Script automation of animations fails and sabotages other methods and it is not well documented so even AIs are clueless how to resolve issues. Often an animation works natively but not with MATLAB Online. Animation of results seems to me rather basic and should be easier!
Frequently, I find myself doing things like the following,
xyz=rand(100,3);
XYZ=num2cell(xyz,1);
scatter3(XYZ{:,1:3})
But num2cell is time-consuming, not to mention that requiring it means extra lines of code. Is there any reason not to enable this syntax,
scatter3(xyz{:,1:3})
so that I one doesn't have to go through num2cell? Here, I adopt the rule that only dimensions that are not ':' will be comma-expanded.
If you use tables extensively to perform data analysis, you may at some point have wanted to add new functionalities suited to your specific applications. One straightforward idea is to create a new class that subclasses the built-in table class. You would then benefit from all inherited existing methods.
One workaround is to create a new class that wraps a table as a Property, and re-implement all the methods that you need and are already defined for table. The is not too difficult, except for the subsref method, for which I’ll provide the code below.
Class definition
Defining a wrapper of the table class is quite straightforward. In this example, I call the class “Report” because that is what I intend to use the class for, to compute and store reports. The constructor just takes a table as input:
classdef Rapport
methods
function obj = Report(t)
if isa(t, 'Report')
obj = t;
else
obj.t_ = t;
end
end
end
properties (GetAccess = private, SetAccess = private)
t_ table = table();
end
end
I designed the constructor so that it converts a table into a Report object, but also so that if we accidentally provide it with a Report object instead of a table, it will not generate an error.
Reproducing the behaviour of the table class
Implementing the existing methods of the table class for the Report class if pretty easy in most cases.
I made use of a method called “table” in order to be able to get the data back in table format instead of a Report, instead of accessing the property t_ of the object. That method can also be useful whenever you wish to use the methods or functions already existing for tables (such as writetable, rowfun, groupsummary…).
classdef Rapport
...
methods
function t = table(obj)
t = obj.t_;
end
function r = eq(obj1,obj2)
r = isequaln(table(obj1), table(obj2));
end
function ind = size(obj, varargin)
ind = size(table(obj), varargin{:});
end
function ind = height(obj, varargin)
ind = height(table(obj), varargin{:});
end
function ind = width(obj, varargin)
ind = width(table(obj), varargin{:});
end
function ind = end(A,k,n)
% ind = end(A.t_,k,n);
sz = size(table(A));
if k < n
ind = sz(k);
else
ind = prod(sz(k:end));
end
end
end
end
In the case of horzcat (same principle for vertcat), it is just a matter of converting back and forth between the table and Report classes:
classdef Rapport
...
methods
function r = horzcat(obj1,varargin)
listT = cell(1, nargin);
listT{1} = table(obj1);
for k = 1:numel(varargin)
kth = varargin{k};
if isa(kth, 'Report')
listT{k+1} = table(kth);
elseif isa(kth, 'table')
listT{k+1} = kth;
else
error('Input must be a table or a Report');
end
end
res = horzcat(listT{:});
r = Report(res);
end
end
end
Adding a new method
The plus operator already exists for the table class and works when the table contains all numeric values. It sums columns as long as the tables have the same length.
Something I think would be nice would be to be able to write t1 + t2, and that would perform an outerjoin operation between the tables and any sizes having similar indexing columns.
That would be so concise, and that's what we’re going to implement for the Report class as an example. That is called “plus operator overloading”. Of course, you could imagine that the “+” operator is used to compute something else, for example adding columns together with regard to the keys index. That depends on your needs.
Here’s a unittest example:
classdef ReportTest < matlab.unittest.TestCase
methods (Test)
function testPlusOperatorOverload(testCase)
t1 = array2table( ...
{ 'Smith', 'Male' ...
; 'JACKSON', 'Male' ...
; 'Williams', 'Female' ...
} , 'VariableNames', {'LastName' 'Gender'} ...
);
t2 = array2table( ...
{ 'Smith', 13 ...
; 'Williams', 6 ...
; 'JACKSON', 4 ...
}, 'VariableNames', {'LastName' 'Age'} ...
);
r1 = Report(t1);
r2 = Report(t2);
tRes = r1 + r2;
tExpected = Report( array2table( ...
{ 'JACKSON' , 'Male', 4 ...
; 'Smith' , 'Male', 13 ...
; 'Williams', 'Female', 6 ...
} , 'VariableNames', {'LastName' 'Gender' 'Age'} ...
) );
testCase.verifyEqual(tRes, tExpected);
end
end
end
And here’s how I’d implement the plus operator in the Report class definition, so that it also works if I add a table and a Report:
classdef Rapport
...
methods
function r = plus(obj1,obj2)
table1 = table(obj1);
table2 = table(obj2);
result = outerjoin(table1, table2 ...
, 'Type', 'full', 'MergeKeys', true);
r = reportingits.dom.Rapport(result);
end
end
end
The case of the subsref method
If we wish to access the elements of an instance the same way we would with regular tables, whether with parentheses, curly braces or directly with the name of the column, we need to implement the subsref and subsasgn methods. The second one, subsasgn is pretty easy, but subsref is a bit tricky, because we need to detect whether we’re directing towards existing methods or not.
Here’s the code:
classdef Rapport
...
methods
function A = subsasgn(A,S,B)
A.t_ = subsasgn(A.t_,S,B);
end
function B = subsref(A,S)
isTableMethod = @(m) ismember(m, methods('table'));
isReportMethod = @(m) ismember(m, methods('Report'));
switch true
case strcmp(S(1).type, '.') && isReportMethod(S(1).subs)
methodName = S(1).subs;
B = A.(methodName)(S(2).subs{:});
if numel(S) > 2
B = subsref(B, S(3:end));
end
case strcmp(S(1).type, '.') && isTableMethod (S(1).subs)
methodName = S(1).subs;
if ~isReportMethod(methodName)
error('The method "%s" needs to be implemented!', methodName)
end
otherwise
B = subsref(table(A),S(1));
if istable(B)
B = Report(B);
end
if numel(S) > 1
B = subsref(B, S(2:end));
end
end
end
end
end
Conclusion
I believe that the table class is Sealed because is case new methods are introduced in MATLAB in the future, the subclass might not be compatible if we created any or generate unexpected complexity.
The table class is a really powerful feature.
I hope this example has shown you how it is possible to extend the use of tables by adding new functionalities and maybe given you some ideas to simplify some usages. I’ve only happened to find it useful in very restricted cases, but was still happy to be able to do so.
In case you need to add other methods of the table class, you can see the list simply by calling methods(’table’).
Feel free to share your thoughts or any questions you might have! Maybe you’ll decide that doing so is a bad idea in the end and opt for another solution.
(Requested for newer MATLAB releases (e.g. R2026B), MATLAB Parallel Processing toolbox.)
Lower precision array types have been gaining more popularity over the years for deep learning. The current lowest precision built-in array type offered by MATLAB are 8-bit precision arrays, e.g. int8 and uint8. A good thing is that these 8-bit array types do have gpuArray support, meaning that one is able to design GPU MEX codes that take in these 8-bit arrays and reinterpret them bit-wise as other 8-bit array types, e.g. FP8, which is especially common array type used in modern day deep learning applications. I myself have used this to develop forward pass operations with 8-bit precision that are around twice as fast as 16-bit operations and with output arrays that still agree well with 16-bit outputs (measured with high cosine similarity). So the 8-bit support that MATLAB offers is already quite sufficient.
Recently, 4-bit precision array types have been shown also capable of being very useful in deep learning. These array types can be processed with Tensor Cores of more modern GPUs, such as NVIDIA's Blackwell architecture. However, MATLAB does not yet have a built-in 4-bit precision array type.
Just like MATLAB has int8 and uint8, both also with gpuArray support, it would also be nice to have MATLAB have int4 and uint4, also with gpuArray support.
I can't believe someone put time into this ;-)
The formula comes from @yuruyurau. (https://x.com/yuruyurau)
digital life 1
figure('Position',[300,50,900,900], 'Color','k');
axes(gcf, 'NextPlot','add', 'Position',[0,0,1,1], 'Color','k');
axis([0, 400, 0, 400])
SHdl = scatter([], [], 2, 'filled','o','w', 'MarkerEdgeColor','none', 'MarkerFaceAlpha',.4);
t = 0;
i = 0:2e4;
x = mod(i, 100);
y = floor(i./100);
k = x./4 - 12.5;
e = y./9 + 5;
o = vecnorm([k; e])./9;
while true
t = t + pi/90;
q = x + 99 + tan(1./k) + o.*k.*(cos(e.*9)./4 + cos(y./2)).*sin(o.*4 - t);
c = o.*e./30 - t./8;
SHdl.XData = (q.*0.7.*sin(c)) + 9.*cos(y./19 + t) + 200;
SHdl.YData = 200 + (q./2.*cos(c));
drawnow
end
digital life 2
figure('Position',[300,50,900,900], 'Color','k');
axes(gcf, 'NextPlot','add', 'Position',[0,0,1,1], 'Color','k');
axis([0, 400, 0, 400])
SHdl = scatter([], [], 2, 'filled','o','w', 'MarkerEdgeColor','none', 'MarkerFaceAlpha',.4);
t = 0;
i = 0:1e4;
x = i;
y = i./235;
e = y./8 - 13;
while true
t = t + pi/240;
k = (4 + sin(y.*2 - t).*3).*cos(x./29);
d = vecnorm([k; e]);
q = 3.*sin(k.*2) + 0.3./k + sin(y./25).*k.*(9 + 4.*sin(e.*9 - d.*3 + t.*2));
SHdl.XData = q + 30.*cos(d - t) + 200;
SHdl.YData = 620 - q.*sin(d - t) - d.*39;
drawnow
end
digital life 3
figure('Position',[300,50,900,900], 'Color','k');
axes(gcf, 'NextPlot','add', 'Position',[0,0,1,1], 'Color','k');
axis([0, 400, 0, 400])
SHdl = scatter([], [], 1, 'filled','o','w', 'MarkerEdgeColor','none', 'MarkerFaceAlpha',.4);
t = 0;
i = 0:1e4;
x = mod(i, 200);
y = i./43;
k = 5.*cos(x./14).*cos(y./30);
e = y./8 - 13;
d = (k.^2 + e.^2)./59 + 4;
a = atan2(k, e);
while true
t = t + pi/20;
q = 60 - 3.*sin(a.*e) + k.*(3 + 4./d.*sin(d.^2 - t.*2));
c = d./2 + e./99 - t./18;
SHdl.XData = q.*sin(c) + 200;
SHdl.YData = (q + d.*9).*cos(c) + 200;
drawnow; pause(1e-2)
end
digital life 4
figure('Position',[300,50,900,900], 'Color','k');
axes(gcf, 'NextPlot','add', 'Position',[0,0,1,1], 'Color','k');
axis([0, 400, 0, 400])
SHdl = scatter([], [], 1, 'filled','o','w', 'MarkerEdgeColor','none', 'MarkerFaceAlpha',.4);
t = 0;
i = 0:4e4;
x = mod(i, 200);
y = i./200;
k = x./8 - 12.5;
e = y./8 - 12.5;
o = (k.^2 + e.^2)./169;
d = .5 + 5.*cos(o);
while true
t = t + pi/120;
SHdl.XData = x + d.*k.*sin(d.*2 + o + t) + e.*cos(e + t) + 100;
SHdl.YData = y./4 - o.*135 + d.*6.*cos(d.*3 + o.*9 + t) + 275;
SHdl.CData = ((d.*sin(k).*sin(t.*4 + e)).^2).'.*[1,1,1];
drawnow;
end
digital life 5
figure('Position',[300,50,900,900], 'Color','k');
axes(gcf, 'NextPlot','add', 'Position',[0,0,1,1], 'Color','k');
axis([0, 400, 0, 400])
SHdl = scatter([], [], 1, 'filled','o','w',...
'MarkerEdgeColor','none', 'MarkerFaceAlpha',.4);
t = 0;
i = 0:1e4;
x = mod(i, 200);
y = i./55;
k = 9.*cos(x./8);
e = y./8 - 12.5;
while true
t = t + pi/120;
d = (k.^2 + e.^2)./99 + sin(t)./6 + .5;
q = 99 - e.*sin(atan2(k, e).*7)./d + k.*(3 + cos(d.^2 - t).*2);
c = d./2 + e./69 - t./16;
SHdl.XData = q.*sin(c) + 200;
SHdl.YData = (q + 19.*d).*cos(c) + 200;
drawnow;
end
digital life 6
clc; clear
figure('Position',[300,50,900,900], 'Color','k');
axes(gcf, 'NextPlot','add', 'Position',[0,0,1,1], 'Color','k');
axis([0, 400, 0, 400])
SHdl = scatter([], [], 2, 'filled','o','w', 'MarkerEdgeColor','none', 'MarkerFaceAlpha',.4);
t = 0;
i = 1:1e4;
y = i./790;
k = y; idx = y < 5;
k(idx) = 6 + sin(bitxor(floor(y(idx)), 1)).*6;
k(~idx) = 4 + cos(y(~idx));
while true
t = t + pi/90;
d = sqrt((k.*cos(i + t./4)).^2 + (y/3-13).^2);
q = y.*k.*cos(i + t./4)./5.*(2 + sin(d.*2 + y - t.*4));
c = d./3 - t./2 + mod(i, 2);
SHdl.XData = q + 90.*cos(c) + 200;
SHdl.YData = 400 - (q.*sin(c) + d.*29 - 170);
drawnow; pause(1e-2)
end
digital life 7
clc; clear
figure('Position',[300,50,900,900], 'Color','k');
axes(gcf, 'NextPlot','add', 'Position',[0,0,1,1], 'Color','k');
axis([0, 400, 0, 400])
SHdl = scatter([], [], 2, 'filled','o','w', 'MarkerEdgeColor','none', 'MarkerFaceAlpha',.4);
t = 0;
i = 1:1e4;
y = i./345;
x = y; idx = y < 11;
x(idx) = 6 + sin(bitxor(floor(x(idx)), 8))*6;
x(~idx) = x(~idx)./5 + cos(x(~idx)./2);
e = y./7 - 13;
while true
t = t + pi/120;
k = x.*cos(i - t./4);
d = sqrt(k.^2 + e.^2) + sin(e./4 + t)./2;
q = y.*k./d.*(3 + sin(d.*2 + y./2 - t.*4));
c = d./2 + 1 - t./2;
SHdl.XData = q + 60.*cos(c) + 200;
SHdl.YData = 400 - (q.*sin(c) + d.*29 - 170);
drawnow; pause(5e-3)
end
digital life 8
clc; clear
figure('Position',[300,50,900,900], 'Color','k');
axes(gcf, 'NextPlot','add', 'Position',[0,0,1,1], 'Color','k');
axis([0, 400, 0, 400])
SHdl{6} = [];
for j = 1:6
SHdl{j} = scatter([], [], 2, 'filled','o','w', 'MarkerEdgeColor','none', 'MarkerFaceAlpha',.3);
end
t = 0;
i = 1:2e4;
k = mod(i, 25) - 12;
e = i./800; m = 200;
theta = pi/3;
R = [cos(theta) -sin(theta); sin(theta) cos(theta)];
while true
t = t + pi/240;
d = 7.*cos(sqrt(k.^2 + e.^2)./3 + t./2);
XY = [k.*4 + d.*k.*sin(d + e./9 + t);
e.*2 - d.*9 - d.*9.*cos(d + t)];
for j = 1:6
XY = R*XY;
SHdl{j}.XData = XY(1,:) + m;
SHdl{j}.YData = XY(2,:) + m;
end
drawnow;
end
digital life 9
clc; clear
figure('Position',[300,50,900,900], 'Color','k');
axes(gcf, 'NextPlot','add', 'Position',[0,0,1,1], 'Color','k');
axis([0, 400, 0, 400])
SHdl{14} = [];
for j = 1:14
SHdl{j} = scatter([], [], 2, 'filled','o','w', 'MarkerEdgeColor','none', 'MarkerFaceAlpha',.1);
end
t = 0;
i = 1:2e4;
k = mod(i, 50) - 25;
e = i./1100; m = 200;
theta = pi/7;
R = [cos(theta) -sin(theta); sin(theta) cos(theta)];
while true
t = t + pi/240;
d = 5.*cos(sqrt(k.^2 + e.^2) - t + mod(i, 2));
XY = [k + k.*d./6.*sin(d + e./3 + t);
90 + e.*d - e./d.*2.*cos(d + t)];
for j = 1:14
XY = R*XY;
SHdl{j}.XData = XY(1,:) + m;
SHdl{j}.YData = XY(2,:) + m;
end
drawnow;
end
It’s exciting to dive into a new dataset full of unfamiliar variables but it can also be overwhelming if you’re not sure where to start. Recently, I discovered some new interactive features in MATLAB live scripts that make it much easier to get an overview of your data. With just a few clicks, you can display sparklines and summary statistics using table variables, sort and filter variables, and even have MATLAB generate the corresponding code for reproducibility.
The Graphics and App Building blog published an article that walks through these features showing how to explore, clean, and analyze data—all without writing any code.
If you’re interested in streamlining your exploratory data analysis or want to see what’s new in live scripts, you might find it helpful:
If you’ve tried these features or have your own tips for quick data exploration in MATLAB, I’d love to hear your thoughts!
Pure Matlab
82%
Simulink
18%
11 votes
Jorge Bernal-AlvizJorge Bernal-Alviz shared the following code that requires R2025a or later:
Test()
function Test()
duration = 10;
numFrames = 800;
frameInterval = duration / numFrames;
w = 400;
t = 0;
i_vals = 1:10000;
x_vals = i_vals;
y_vals = i_vals / 235;
r = linspace(0, 1, 300)';
g = linspace(0, 0.1, 300)';
b = linspace(1, 0, 300)';
r = r * 0.8 + 0.1;
g = g * 0.6 + 0.1;
b = b * 0.9 + 0.1;
customColormap = [r, g, b];
figure('Position', [100, 100, w, w], 'Color', [0, 0, 0]);
axis equal;
axis off;
xlim([0, w]);
ylim([0, w]);
hold on;
colormap default;
colormap(customColormap);
plothandle = scatter([], [], 1, 'filled', 'MarkerFaceAlpha', 0.12);
for i = 1:numFrames
t = t + pi/240;
k = (4 + 3 * sin(y_vals * 2 - t)) .* cos(x_vals / 29);
e = y_vals / 8 - 13;
d = sqrt(k.^2 + e.^2);
c = d - t;
q = 3 * sin(2 * k) + 0.3 ./ (k + 1e-10) + ...
sin(y_vals / 25) .* k .* (9 + 4 * sin(9 * e - 3 * d + 2 * t));
points_x = q + 30 * cos(c) + 200;
points_y = q .* sin(c) + 39 * d - 220;
points_y = w - points_y;
CData = (1 + sin(0.1 * (d - t))) / 3;
CData = max(0, min(1, CData));
set(plothandle, 'XData', points_x, 'YData', points_y, 'CData', CData);
brightness = 0.5 + 0.3 * sin(t * 0.2);
set(plothandle, 'MarkerFaceAlpha', brightness);
drawnow;
pause(frameInterval);
end
end
From my experience, MATLAB's Deep Learning Toolbox is quite user-friendly, but it still falls short of libraries like PyTorch in many respects. Most users tend to choose PyTorch because of its flexibility, efficiency, and rich support for many mathematical operators. In recent years, the number of dlarray-compatible mathematical functions added to the toolbox has been very limited, which makes it difficult to experiment with many custom networks. For example, svd is currently not supported for dlarray inputs.
This link (List of Functions with dlarray Support - MATLAB & Simulink) lists all functions that support dlarray as of R2026a — only around 200 functions (including toolbox-specific ones). I would like to see support for many more fundamental mathematical functions so that users have greater freedom when building and researching custom models. For context, the core MATLAB mathematics module contains roughly 600 functions, and many application domains build on that foundation.
I hope MathWorks will prioritize and accelerate expanding dlarray support for basic math functions. Doing so would significantly increase the Deep Learning Toolbox's utility and appeal for researchers and practitioners.
Thank you.
I just learned you can access MATLAB Online from the following shortcut in your web browser: https://matlab.new
Thanks @Yann Debray
From his recent blog post: pip & uv in MATLAB Online » Artificial Intelligence - MATLAB & Simulink
I'm working on training neural networks without backpropagation / automatic differentiation, using locally derived analytic forms of update rules. Given that this allows a direct formula to be derived for the update rule, it removes alot of the overhead that is otherwise required from automatic differentiation.
However, matlab's functionalities for neural networks are currently solely based around backpropagation and automatic differentiation, such as the dlgradient function and requiring everything to be dlarrays during training.
I have two main requests, specifically for functions that perform a single operation within a single layer of a neural network, such as "dlconv", "fullyconnect", "maxpool", "avgpool", "relu", etc:
- these functions should also allow normal gpuArray data instead of requiring everything to be dlarrays.
- these functions are currently designed to only perform the forward pass. I request that these also be designed to perform the backward pass if user requests. There can be another input user flag that can be "forward" (default) or "backward", and then the function should have all the necessary inputs to perform that operation (e.g. for "avgpool" forward pass it only needs the avgpool input data and the avgpool parameters, but for the "avgpool" backward pass it needs the deriviative w.r.t. the avgpool output data, the avgpool parameters, and the original data dimensions). I know that there is a maxunpool function that achieves this for maxpool, but it has significant issues when trying to use it this way instead of by backpropagation in a dlgradient type layer, see (https://www.mathworks.com/matlabcentral/answers/2179587-making-a-custom-way-to-train-cnns-and-i-am-noticing-that-avgpool-is-significantly-faster-than-maxpo?s_tid=srchtitle).
I don't know how many people would benefit from this feature, and someone could always spend their time creating these functionalities themselves by matlab scripts, cuDNN mex, etc., but regardless it would be nice for matlab to have this allowable for more customizable neural net training.
