LLMs are taking over. We can't escape from that. So, yes, have you ever got lost in the context, with too long conversation, hours spend on trial and error so that your chat actually corrected something but ate its own tail at some point doing wrong what was good at the beginning.. Yup, this is where my story on Single Prompts.
When LLM's appeared I was struggling with the question about the value of our work. What will be valuable. What we can give to each other in this new era. Although this is now in the middle of change and middle of paradigm shift in programming in general, I think there is something. Prompts. New way of writing programs.
Imagine you want to write a piece of software and instead of starting from scratch you take a prompt at some level of abstraction that gets you to some specific point. It can be a program that you may start developming from. When I've recently came across the idea that Single Prompts will be of some value in the future I immediately thought about modelling, my lovely subject. I've sat down for a few months and yup, there it is, I was lucky to get through three taugh reviews and together with Taylor&Francis we will release the book on prompting just this July!
Book website: https://www.routledge.com/Computational-Modelling-with-Single-Prompts/Matyka/p/book/9781041223603
The concept of the book is simple. It covers 65 models from computational physics (from cellular automata, chaos, PDE's, particles, fluids, ants colonies, granular matter modelling of snowflakes, epidemia and even stalagmites growth). Each chapter consists of 1 model, full description (written by myself, no LLM was used here), then algorithm and single, full prompt. The objective of the prompt was, that it creates fully functional implementation in JavaScript for the model. Then, after the code was generated, results are included in the chapter. An exemplary result is on the book cover, in fact it was done using Claude Sonnet chat (I've used several other language models in the book, tested many).
Working on the book was a lot of fun, I've always dreamed of a book on modelling, now I was able to write one (many of the models I've implemented by myself in the past but for many I was not able to find time.. now with LLM's it took much less time!). There are many stories about models in the book, there is a specific reason I've included each of them, there are research ideas associated with many of them.
Ask me questions about the book here if you like, that may help me drive this blog as well as this new subject is getting me more deeper and deeper..
Maciej Matyka, 9-06-2026
maciejmatyka.github.io
