[eadan]

Congratulations on the launch! I've been excitedly waiting for the public release since hearing about HASH on Joel's blog a few months ago. I've been playing with some models over the past few hours and I have a few questions / comments.

First, would it be possible to perform sensitivity analysis on parameters of the model? For example, with the city infection model, run a number of simulations in parallel with different infection rate values, and see how it affects the model output. Relatedly, I think it would be useful to specify a probability distribution over a parameter rather than setting a fixed value. The state space of the model would become exponential in the number of parameters which can vary, so I understand if it's infeasible, at least in the public version.

The real-time spatial and plot views are great. I would like to try some network based simulations. Any plans to add visualizations for these?

[DavidWilkinson]

Thanks! We have some network features in the pipeline (including better viz) and will be introducing some stdlib functions to help in the coming days. :)

We have some users representing networks in the 3D viewer at present, and have seen three ways implementing networks to date:

1. Edges are represented as agents. They are used to store properties such as edge length, and to provide nodes with a way of accessing other nodes.

2. Edges are represented AND USED as agents. Edges not only store properties but themselves exhibit behaviors.

3. Nodes are given a network object which contains information about their network neighbors and all relevant properties (such as directed/undirected edge, edge length, etc...)

Re: your question around experiments... yes to all three (parameter sweeping, Monte Carlo, and sensitivity analysis), and a bunch more. We'll be shipping this alongside H-Cloud. More on that in the full explainer at https://hash.ai/about/mission

Thanks for giving the beta a spin! Happy to chat in more depth over on our public Slack.

[eadan]

Thanks for your response! This technology is very exciting. Since you're targeting WASM there's a whole host of possibilities. Python is already supported through Iodide [1], but I imagine other languages could also be ported, and even DSLs specifically tailored for agent based modelling. That's not the mention all the other things happening in the WASM space like webgpu [2] and WASI [3].

The potential for ABMs are huge now that we have access to cheap and massively parallel compute. Imagine arbitrarily complex models of each individual in an economy interacting with each other over a distributed network of thousands of machines -- that's possible now. Instead of trying to predict the future, we can compute it.

[1] https://github.com/iodide-project/pyodide

[2] https://github.com/gpuweb/gpuweb/wiki/Implementation-Status

[3] https://wasi.dev/

[DavidWilkinson]

The Python simulations we run in-browser are indeed using Pyodide in part! It's brilliant, but there's quite a performance penalty at present. We're working on some optimizations, but right now running the engine locally or offloading to the cloud are the only places real scale can be achieved. That said we're thrilled that Python-in-browser is possible at all, and excited at how many high-quality complementary projects are in the works. It really does feel like "everything is coming together" :)

[ijpsud]

Why is login required to view simulations? If I want to share my simulations with my followers, I don't want everyone to have to sign up for an account. Also, the "play" button in the editor UI should probably look more like a play button.

I'm excited by what you're building here!

[DavidWilkinson]

Hi! Fully appreciate that; we're looking into removing the gate following feedback here (https://news.ycombinator.com/item?id=23573136).

The run and play buttons being distinct are in our mind important in larger sims... but fully appreciate the confusion. We'll give this some more thought. Thanks for your patience and comments!

[Frost1x]

>Re: your question around experiments... yes to all three (parameter sweeping, Monte Carlo, and sensitivity analysis), and a bunch more. We'll be shipping this alongside H-Cloud.

Firstly, fantastic work.

I'd love to see more details on 'H-Cloud' regarding infrastructure, pricing, and feasibility/licensing options for self-deployment should I need to recreate/run a similar environment on premise vs. at a remotely hosted service.

I'm also quite interested in what's going to be FOSS'd in terms of the 'H-Engine' (written in Rust) and if some friendly interface (similar to H-Core) will still be provided.