Engineering Persuadable Matter

My latest publication, commenting on a paper about agential chemistry, from my own computational perspective.  This topic falls in the intersection of Artificial Life and Unconventional Computing, forming a research area I am intensely interested in.

Susan Stepney. Engineering Persuadable Matter: A Comment on Armstrong’s ‘Life, Mind and Matter’. Social Epistemology Review and Reply Collective, 14(3):33-42, 2025.

Rachel Armstrong (2024) advocates for a new approach to ‘agential chemistry’, a form of ‘new materialism’ that allows matter to take an active role. Here I comment on some of these ideas through a computational lens: the consequences if agential chemistry can perform computation to advance its own agenda; how it might provide the structure and dynamics needed for computation, and the metadynamics for open ended systems; and how it opens the possibility of a new technological discipline of engineering ‘persuadable’ agential matter.

While searching for an image to use to spice up this post, I came across an interesting Medium piece that forms a nice overview of some of the issues, from a neural AI perspective.  And has the pretty image I use above (click to embiggen). 

misty morning

 Early morning view: the world has gone away...

07:33 GMT, looking west

It's always worth checking

I was writing some Python code today, and I had some logic best served by a case statement.  I remembered that Python doesn't have a case statement, but I decided to google to see if there was a suitably pythonic pattern I should use instead.

Aha!  Python v3.10 introduced a case statement, and I'm currently using v3.13.  Excellent.  I scanned the syntax, then added the relevant lines to my code.

After I'd finished that bit of coding, I went and read the official Python tutorial.  Of course, Python being Python, its 'case' statement is actually a very sophisticated and powerful 'structural pattern matching' statement.  I might have some fun with this...

So, every day in every way, at least Python is getting better and better.

Reservoir computing benchmarks: a tutorial review and critique

Our latest paper, reviewing a bunch of standard benchmarks for Reservoir Computing, digging into their histories, and why some of them might not be the best approach to be using.

Chester Wringe, Martin Trefzer, Susan Stepney. Reservoir computing benchmarks: a tutorial review and critique. International Journal of Parallel, Emergent and Distributed Systems, 1-39, 2025. doi:10.1080/17445760.2025.2472211

Reservoir Computing is an Unconventional Computation model to perform computation on various different substrates, such as recurrent neural networks or physical materials. The method takes a ‘black-box’ approach, training only the outputs of the system it is built on. As such, evaluating the computational capacity of these systems can be challenging. We review and critique the evaluation methods used in the field of reservoir computing. We introduce a categorisation of benchmark tasks. We review multiple examples of benchmarks from the literature as applied to reservoir computing, and note their strengths and shortcomings. We suggest ways in which benchmarks and their uses may be improved to the benefit of the reservoir computing community.

If you don't subscribe to that journal, you can find the same text (if not as prettily typeset) on the arXiv, at   arXiv:2405.06561 [cs.ET]

partial lunar eclipse

A photo, with a digital SLR and an ordinary lens, of the partial lunar eclipse this morning, looking nice and red:

By the time a longer lens was fitted to the camera, low cloud had bubbled up and hidden the moon.  (We should have been better prepared: we will be for the partial solar eclipse in a fortnight.) In fact, the moon was nearly set anyway, as can be seen by "enhancing" the picture above, making the horizon visible:

A bit later, the sky was light, with an amazing contrail disappearing into the distance, as if following the setting moon.

You can identify on this picture where the moon was above, by comparing horizon profiles.  (The spike pointing to the moon is the pylon on the right.)  And yes, that position is still in cloud.

sunset with a bonus

Gorgeous sunset, with a bonus Venus -- can you spot it?  (You will probably need to click on the image to make it full size.)

18:30GMT, looking west

Here's a clue:

king of the castle

We had a tree cut down recently. One of the neighbourhood cats clearly likes this new perch.

what a difference a day makes

Yesterday, we had had nine sunny days since the beginning of the month, and the solar PV plot looked amazing.

Today broke the streak.  It was not sunny: it was full cloud cover all day, with a bit of brightness around noon.  The solar PV plot now looks very different, with a lot of (pale) orange filling in the minimum to lower quartile range:

Nine sunny days in March

I make a variety of plots of data from our solar PV system.  One set is of the power generation over the day, plotted on a background of the month (so far) averages.  February looked like this:

The horizontal time axis runs from 3:00am to 9:00pm GMT. The vertical axis runs from zero to 8kW. The orange regions indicate the minimum, lower quartile, median, upper quartile, and maximum generation at that time, over the month. The line indicates the actual generation at that time.  Several days have very little generation, with the black generation line hugging the x-axis (February this year seemed a particularly dark month), so the orange fills the shape.

March so far has been somewhat different:

Nearly full sun every day; hardly any orange in sight on the plots.  Even on the 7th, originally forecast to be dull, there were only a few dips in an otherwise bright day.  It has been very mild, even warm.

The weather is due to break tomorrow.  Ah well, it was glorious while it lasted!

sequestering carbon, several books at a time CXLIX

The latest batch:

not in a pear tree

 View from our kitchen window:

snapped on my phone through the kitchen window, so a little bit of reflection on some pics

This red-legged partridge was pecking around under the birdfeeder, on which was a great tit flinging seed around.

By the time we got a proper camera with a bigger lens, this bird had flown.

four or five planet evening

Clear skies, cold still weather: good seeing.  So, about half past six this evening, we looked for the planetary alignment.  We saw (with the naked eye) faint Mercury close to the still faintly pink horizon, spectacular Venus dominating the western sky, bright Jupiter near the zenith, and red red Mars a little further round.  Binoculars made them brighter, but still point like, except for Venus.  I could convince myself I could see Venus’ phase, but it was so bright, it was hard to be sure.

Later, 8:30-9pm, it was much darker, and we used the big telescope.  Venus and Mercury had set by then.  Jupiter was lower down, but still visible: we could see all four Galilean moons, and some dark bands on the planet itself.  Mars by then had moved close to the zenith.  Last time we tried to view it at that angle, we couldn’t see through the finder, so we bought a 90 degree adapter for it at AstroFest a few weeks ago.  It works!  We got to see the red planet as a disc.  Beautiful.

We had a quick look at the Pleiades, as they were in the vicinity.  My god, it’s full of stars!

We then used the autofinder to try for Uranus.  We could see a bright object near the centre of the field of view.  It could have been a very small disc of a planet, or just a point source of a star.  I choose to believe it was Uranus.

So, definitely four planets, and maybe even five.

relaxing train

While waiting for my train from Glasgow Queen Street to Edinburgh, for the first leg of my journey home, I was amused to see the destination on this other train:

view from a hotel window

I'm in Glasgow, visiting some colleagues at the University of Strathclyde, for our new LoCoMo project that started after I retired!  Yesterday afternoon we had a good chat about a paper (from a different project) that has been stalled since pre-Covid times, and figured out how to solve the issue.  I have been tasked with writing the new section we agreed on, as I now have the time!  Today we talked about the new project.  Tomorrow I'm looking forward to a lab tour, to see what the experimentalists are up to in quantum computing with Rydberg atoms (we are doing some vaguely related theory, but looking at cool kit is always fun).

Meanwhile, this is the view from my hotel room window (nice hotel: very quiet!)

scanners upgrade in vain

 Last night, my Windows 11 machine did an update, to version 24H2.

Today, my Canon scanner didn't work.  After switching it off and on again, and unplugging and replugging the cables, I resorted to Google.  Apparently this is a "known problem" (since December 2024!), and the solution is ... undo the upgrade.

Start > Settings > System > Recovery > Go Back

I now have version 23H2.  And my scanner works again.

new toy

AstroFest has a lot of exhibitors.  Last year we bought a small SeeStar telescope, and a dome for the big telescope we bought back in 2020.  This year we were a little more restrained.  

We had recently been looking at Jupiter with the big telescope; we had also tried for Mars.  But Mars is high in the sky at the moment, which means the telescope is almost vertical.  We have a 90 degree adaptor for the eye piece, but not for the finder scope.  It was too difficult to limbo down to look up through the finder, so we didn't see Mars.  Thanks to this year's AstroFest, we now have a 90 degree adaptor for the finder, too.

A quarter of a century ago, August 1999, we viewed the total eclipse of the sun.  For that, we bought a Lightline Solar Projector, a clever cardboard tube and lens contraption that allowed us to project an image.  We have been using it since to view the sun, but it's not particularly good for seeing sunspots, and is getting a bit battered.  We have a solar filter for the new SeeStar telescope, which gives us a spectacular view, but takes a while to set up.  While wandering around the exhibits, I saw a variety of "solar binoculars", binoculars with built-in solar filters, so they can be used to directly view the sun (and nothing else!)  I was debating whether to get a pair, when one of the other attendees struck up a conversation.  She said she had bought a pair previously, and had spent many a happy time watching the sun.  She convinced me, so I bought a pair.

The weather forecast is cloudy for the next week.  So I'll have to wait to try them out!

view from a hotel window

Another year, another AstroFest.  We stayed at a different hotel this year (next door to the previous one), in the hope it would be quieter: it was!  But the view wasn't quite as interesting:

Two days of fascinating talks again, on aurorae (the ones last year were some of the most southerly viewable ever), Mars, forthcoming eclipses, astrophotography (the quality of "amateur" photography today is streets ahead of professional photography several decades ago), black holes, observatories on the moon, star surveys, the UK Space Agency, star formation, the Antikythera Mechanism, exoplanets, and Enceladus.  Something for everyone.

And the exhibitors had bigger than ever telescopes on offer, including one mount with four large telescopes on it!  A bit beyond our budget, but nevertheless, the shape of amateur things to come?

The train down (Thursday afternoon) was 12 nearly empty coaches.  The train back (Saturday evening) was eight packed coaches: we had to stand.  Two different people offered me their seat: I must look older than I think I do!  I politely (I hope) declined: I had after all been sitting down all day at the event.

sequestering carbon, several books at a time CXLVIII

 The latest batch:

book review: The Big Picture

Sean M. Carroll.
The Big Picture: on the origins of life, meaning and the universe itself.
Oneworld. 2016

---

In this ambitious work, Carroll covers an enormous range: fundamental physics, cosmology, epistemology, complexity, consciousness, morality.

He shows how (our best current understanding of) fundamental physics allows no room for any ghosts in the machine. He explains the philosophical position of “poetic naturalism”, and how it can be used to tell (well-founded, scientific) “stories” about the emergent macro-world that don’t need to reduce everything to quantum physics, but how this necessarily means we have to omit certain aspects when telling these stories. And he introduces Bayesian reason as a technique for improving understanding at all levels.

This is a bold endeavour, cramming much profound material into 50 chapters, each less than 10 pages, but adding up to over 400 pages of fascinating material. He has interesting insights on a wide range of topics, not just his own speciality of quantum physics, but also epistemology, emergence, complexity, and more. It is a deeply humanistic account, yet grounded in the cold hard light of the constraints of physical reality.

Highly recommended.

out of this world problems

Jupiter has been prominent recently.  We observed it using our SeeStar telescope.  It was a visible disc, but a bit small, with faint bands juuuust visible.

Last night was clear, so we decided to try using the big telescope in the dome to get a better view.

Last night was cold.  The dome was frozen shut.  Oh well.

Raman SOM

Daniel West, Susan Stepney, Y. Hancock.  Unsupervised self-organising map classification of Raman spectra from prostate cell lines uncovers substratified prostate cancer disease states.  Scientific Reports, 15:773, 2025. doi:10.1038/s41598-024-83708-6

This started out as a feasibility study, to see if Kohonen Self-Organising Maps (SOMs) could be used to cluster minimally preprocessed Raman spectroscopy data taken from individual cells.  SOM an unsupervised learning approach, and can cluster high dimensional data (here, over 1000D) down into a 2D visual representation.  We had Raman spectra of prostate cells, some cancerous, some not.  Could a SOM distinguish these two classes?

We blinded the data, so that the system did not know which spectrum was in which class, to ensure this was truly an unsupervised exercise.  After some fiddling about to understand what values several parameters should be, we fed the data in, and looked at the resulting map.  We could see three clusters.

Had it worked?  We unblinded the data, and yes, one of the clusters was the non-cancerous cells, and the other two clusters were cancerous cells.  Why two clusters?  Well, it turns out the mapping process had managed to discover two distinct classes of cancerous cells.  Further research is underway to investigate these differences.

So yes, it works, and better than we had hoped!