We are so far away from each other Steve!
@walkingrandomly
I’m here to discuss Mathematics, Research Software Engineering, Machine Learning and #HPC. I was @walkingrandomly on twitter Now a Community Developer Advocate at MathWorks. Author of The MATLAB Blog.
We are so far away from each other Steve!
Zooming right in, you can see that I am part of a cluster called 'HPC and Supercomputing professionals'. Zooming out, it's fun to see what communities are clustered near us such as 'Ceramic artists and potters'
Happy Valentines from MATLAB to you. blogs.mathworks.com/matlab/2026/...
I just bench pressed 120kg for the first time in my life. I’m 48.
I’m happy!
Interesting article on AI Coding Assistants where the author suggests that they are getting worse and that maybe its partially the fault of all the newbies using them leading to bad training data.
spectrum.ieee.org/ai-coding-de...
What's your experience?
The free AI-powered workspace for scientists has drawn immediate skepticism from researchers who fear the tool will accelerate the already overwhelming flood of low-quality papers into scientific journals.
I live in the UK but was in Boston last week. Man alive that place is cold!
My desktop PC is 5 years old and showing its age.
What do you suggest I look at for a new machine for scientific computing?
The current specs are
- Intel i7-11700 with 32GB RAM
- NVIDIA RTX 3070 GPU
- 1TB SSD + 1TB HDD disk space
Looks like a nice bit of kit
We will give it more chances then! Currently just assuming that we are clueless noobs at using it
I caved in and bought an air fryer.
So far I am deeply unimpressed. Deeply!
Hoping that its because I haven’t figured out how to use it properly yet and not because it was an overrated piece of tat.
I have done absolutely nothing today! It was glorious and I highly recommend it.
This article from The Guardian discusses many of my own concerns about AI right now:
Artificial intelligence research has a slop problem, academics say: ‘It’s a mess’
www.theguardian.com/technology/2...
My current use for ChatGPT is translating the gen Z slang that my step daughter uses when she texts me
I asked a few people that question at SC25….the ones standing underneath banners saying ‘AI factory blah blah’
I still don’t know!
Needed to do a computer thing so opened my notes file HowToDoThing.md it contained one line:
TODO: Write detailed notes about thing.
Sometimes I hate lazy, past Mike!
In what is now a solid tradition, the most important post-SC event ... publication of Glenn's SC notes!
#HPC #SC25 #supercomputing #HPCignites
OK, this turned out to be a totally unexpected and really cool restoration technique. Physicists! MATLAB! Transfer film! Laser cutter!
www.youtube.com/watch?v=R1vm...
Mike Croucher with Thor
#SC25 was the best one I’ve ever attended….and not just because I got to meet Thor.
Our toolbox got published in the @joss-openjournals.bsky.social!
The arch at St Louis
View from the arch at St Louis
If you’re in St Louis for #SC25 I highly recommend going up the arch. Here’s one of the views from the top where you can also see its shadow
Have spent most of the day travelling to #SC25 and am just about to start the final leg. Taking off from Atlanta on my way to St Louis
Are you going to be at #SC25? Want to connect with a large subset of the #HPC community with one click? The #HPC starter pack is what you need! bsky.app/starter-pack...
I'm from MathWorks. Yes, this is normal and expected. When I teach my High Performance Computing in MATLAB courses, its one of the things I mention : don't co 'clear all'
It is because of JIT compilation, function caching and so on. I'll see if I can find a reference. If not, I'll write one.
MathWorks release MATLAB MCP Core server
In the video below you are watching Claude read a file on my local machine and use my local install of MATLAB to analyze it. Claude writes the MATLAB code,puts it on my machine and runs it using my copy of MATLAB
Article: blogs.mathworks.com/matlab/2025/...
Unless AI is kept in check, it will not only eat its own homework but it will also eat its legibility. We will lose the skill to distinguish a composition of a type of work, a deliberate work, against a hybrid work, a piece of work, an argument, fabricated out of past fragments. Audiences will be conditioned to believe that good enough is good enough, until even mediocrity will seem like plenty. Culture will be turned into monocropped and deprived of nutrients. The creative commons will be exhausted. And as is soil when it is overfarmed, once lost, it will not be easily recovered. Let us not flatter ourselves. It is not that AI will soon surpass us in thinking, it is more likely that it will surpass us in boring. When we leave machines to eat their own homework we are not training a generation of geniuses but beige. The cry is straight: do not confuse abundance with richness, fluency with thought, content with creation. Surprise still matters. Originality is not luxury but lifeline. Unless we defend it, we will drown in competence without creation.
"We were raised to believe over the decades that machines will outgrow us[, but] automatons are not consuming the globe brilliantly. They are not outsmarting us—they are overfitting on us."
Snehamol Joseph & Jeena Joseph (2025)
Digital decay: when AI eats its own homework doi.org/10.1007/s001...
We also support and ship solvers from SUNDIALS blogs.mathworks.com/matlab/2024/...
Which of those 50 solvers are useful to you and why?
Noting that they are using MATLAB 2019a there. I'm not sure how it would affect those benchmark results but there has been a lot of work done on what MATLAB can do with ODEs since then.
New blog post: Improved Registration Algorithm for IMREGCORR
Substantial improvements for function IMREGCORR, which registers image pairs using similarity, rigid, and translation transformations.
#matlab
www.steveeddins.com/blog/matlab/...
Zoom in on that Venn diagram.