arbor-sim/community

Welcome to the Arbor Community! What are you working on?

Brent Huisman

@brenthuisman

There have been no recent changes AFAIK to the ebrains lab, did it work before?

Don't know if you can see it, but here it is: https://gitlab.ebrains.eu/technical-coordination/project-internal/devops/platform/ebrains-spack-builds/-/issues/14

Lennart Landsmeer

@llandsmeer

Nope can't see it, but that doesnt matter

I used find / -name arbor-build-catalogue and some patience

Still the invalid JUNCTION_PROCESS thing is a problem

Is that a new arbor breaking change?

Or is the image misconfigured

Brent Huisman

@brenthuisman

The lab has Arbor v0.6. I'm not sure if this was changed post v0.6, but there have been changes to arbor-build-catalog.

Sebastian Schmitt

@schmitts

Could gap junctions transport ions/particles as well?

Brent Huisman

@brenthuisman

Actual gap junctions, or Arbor gap junctions?

Sebastian Schmitt

@schmitts

Arbor gap junctions :)

Brent Huisman

@brenthuisman

Currently they only transport events. What sort of transmission are you thinking of? Complete connection, as in the cytoplasm, so a continuous exchange of ions as per gradients, or particular ions over a threshold or sth?

Thorsten Hater

@thorstenhater

Errr, Arbor GJ are direct electrical connections according to

I = g(V_{here} - V_{there})

7 replies

So, they are transporting a non-specific current.

@schmitts You can try to model this using a Junction mechanism.

Sebastian Schmitt

@schmitts

@brent I was thinking about exchanging some proteins depending on some condition.

Brent Huisman

@brenthuisman

There is a ModelDB questionnaire: time to ask for more Arbor/simulator-independent
models :) https://yalesurvey.ca1.qualtrics.com/jfe/form/SV_3EkPlMkOUDV5xgW

Thorsten Hater

@thorstenhater

Ionic Diffusion is now part of Arbor :tada:
arbor-sim/arbor#1729

Brent Huisman

@brenthuisman

hurray!

Jannik Luboeinski

@jlubo

Nice! =D

Thorsten Hater

@thorstenhater

Github CoPilot for all. :tada: The function was auto-generated from the comment

// Get average runtime of successful runs in seconds
func averageRuntimeInSeconds(runs []Run) float64 {
    var totalTime int
    var failedRuns int
    for _, run := range runs {
        if run.Failed {
            failedRuns++
        } else {
            totalTime += run.Time
        }
    }
    averageRuntime := float64(totalTime) / float64(len(runs) - failedRuns) / 1000
    return averageRuntime
}

Oh ... wait ... this is broken and will return NAN or even throw div-by-zero.* :laughing:

Yes, this is the example they advertise

* if no runs or all fail

Sam Yates

@halfflat

Bit surprised that run.Time is integral too.

Also DIFFUSION!!!!! Awesomesauce.

Sebastian Schmitt

@schmitts

:fireworks:

Brent Huisman

@brenthuisman

Why does a meter_manager require a context? And why does the meter_report require it again? Does it only measure time executed in the Arbor .so?

1 reply

Sam Yates

@halfflat

With the caveat that I haven't looked at this code in probably four or five years, I believe the meter manager uses the context to grab the MPI communicator in order to collate time reports from across ranks.

Brent Huisman

@brenthuisman

OK, I assumed the report was only getting results for the current rank, I'll run it with MPI and see what I get.

Sebastian Schmitt

@schmitts

@halfflat hi! have you found some time to have a look at the proposal to have access to the discretized state?

Sam Yates

@halfflat

Hi! Alas, no. It's been a bit mad. Next week at the earliest I'm afraid.

Sebastian Schmitt

@schmitts

:cry:

Brent Huisman

@brenthuisman

FYI: with the merge of arbor-sim/arbor#1916, which enabled universal (AMD64+ARM64) macos builds, Python <3.8 support was dropped. I hadn't checked that universal wheels require 3.8 and up. If this is a problem for anyone, let me know.

Ben Cumming

@bcumming

About time we moved on from 3.6

Brent Huisman

@brenthuisman

Arbor v0.7 is released! It's up on PyPI: https://pypi.org/project/arbor/0.7/ If you've a minute, let me know if you run into any problems.

Sebastian Schmitt

@schmitts

:thumbsup:

Sebastian Schmitt

@schmitts

Which format for morphology and synapse/neuron dynamics is understood best by NEURON and Arbor?

35 replies

Thorsten Hater

@thorstenhater

This nmlcc ./. jnml, both on Arbor. Minor deviations, but the nmlcc simulation is roughly 2x faster

Sebastian Schmitt

@schmitts

nmlcc erstellt bessere mods?

Thorsten Hater

@thorstenhater

Faster ;)

Sebastian Schmitt

@schmitts

harder, better, faster, stronger?

Brent Huisman

@brenthuisman

Hello Arbor users! The next Arbor Developer Meeting will be held 30 August at 10am CEST. Please join and give your opinion on which features must be prioritized for the next release! Add to your calender and more details here.

3 replies

Brent Huisman

@brenthuisman

Hi everyone, it's spam-thursday :) Ebrains are sending around the news of our v0.7 release: https://ebrains.eu/news/new-update-for-arbor-simulation-library/

You can share, like or subscribe the social media posts here: https://twitter.com/EBRAINS_eu/status/1555123707194097665 https://www.linkedin.com/feed/update/urn:li:activity:6960889120785055744

Sebastian Schmitt

@schmitts

Thanks, @brenthuisman.

Thorsten Hater

@thorstenhater

Hi all,

I went through our issue tracker and did some clean-up. If you have an open issue or one assigned to you, please check whether it has been resolved, become obsolete, or you have pending questions. On the inverse, please re-open anything that's still relevant but was closed.

Sebastian Schmitt

@schmitts

@thorstenhater Thanks!

Sebastian Schmitt

@schmitts

Are there plans for implementing a variable time step integrator?

13 replies

Sebastian Schmitt

@schmitts

Can we have an explicit euler as well? I might need it (as usual) for comparison with existing simulations.

Sebastian Schmitt

@schmitts

What is the resolution of the spike times? dt? dt/2?

Where communities thrive

People

Repo info

Activity