SED-ML meeting at HARMONY

[Lucian Smith]

Greetings, everyone!  HARMONY is coming up, so we would like to invite everyone to an open meeting with the SED-ML editors and anyone else with a stake in SED-ML.  We'll discuss the current state of things, who needs help and what needs work, and plans for where to move in the future.

To try to accommodate both Europe and Pacific Coast time zones, we'll meet Thursday, April 17th at 18:00 Belgium time, just at the end of the conference.  Matthias will organize a room locally, and I'll post a zoom link for remote attendees.

If anyone would like to bring up something specific, please email the list, and we'll be sure to discuss it at the meeting.

-Lucian

[Alan Garny]

Hi Lucian (and Matthias), thanks for organising a SED-ML session.

One (minor) thing that I would like to see discussed is numberOfSteps in UniformTimeCourse (see Figure 2.12 in page 40 of the L1V5 specifications).

Indeed, in OpenCOR, we can set the equivalent of a uniform time course simulation using something like:

• Starting point = 0 (i.e. initialTime and outputStartTime are set to 0);
• Ending point = 1000 (i.e. outputEndTime is set to 1000); and
• Point interval = 1 (i.e. numberOfSteps is set to 1000).

(In CellML, the variable of  integration doesn't have to be time, hence starting point rather than outputStartTime for instance. Also, OpenCOR doesn't currently have the notion of initialTime, hence the same value is used for both initialTime and outputStartTime.)

By setting that point interval to 1, the user can change the value of the ending point without having to worry about what the value of numberOfSteps should be. It is similar to an experimental setup where you specify that you want data to be generated at 1 kHz (rather than having your setup somehow generating 1,000 data points no matter how long you are going to run your experiment for).

So, what about keeping numberOfSteps (for those who like/prefer to use it) and also having something like timeInterval?

Right now, numberOfSteps must be a positive integer, but we could allow it to be equal to zero. Thus, we could have a rule that says that either numberOfSteps or timeInterval must be different from zero and that the one which is different from zero is the one that should be used?

One reason I would welcome such a change is that if, in OpenCOR, someone was to set the point interval to 3 then numberOfSteps would have to be set to 333 (since it currently has to be a positive integer). This means that the next time the SED-ML file is open, OpenCOR would have to use a point interval of 3.003003003. This is the reason OpenCOR currently adds a one-step simulation to account for such a case. In other words, with a point interval of 3, your SED-ML file would end up having a uniform time course simulation from 0 to 999 with 333 steps followed by a one-step stimulation with a step of 1. Yuck!

Man, so much for a minor thing that I would like to discuss! 🙂

Alan

---

From: sed-ml-...@googlegroups.com <sed-ml-...@googlegroups.com> on behalf of Lucian Smith <luciano...@gmail.com>
Sent: Saturday, 5 April 2025 7:18 am
To: sed-ml-...@googlegroups.com <sed-ml-...@googlegroups.com>
Subject: [sed-ml-discuss] SED-ML meeting at HARMONY

 

--
You received this message because you are subscribed to the Google Groups "sed-ml-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sed-ml-discus...@googlegroups.com.
To view this discussion, visit https://groups.google.com/d/msgid/sed-ml-discuss/CAHLmBr38cx_MmjnvMuthJ%2BzrL8jV597-4bQ5oGX4zPreRGCGBg%40mail.gmail.com.

[Matthias König]

We are excited to invite you to the upcoming hybrid SED-ML meeting for HARMONY 2025!

Date & Time: April 17, 2025, at 06:00 PM Brussels Time

Join the Zoom Meeting:
https://hu-berlin.zoom-x.de/j/69231351515?pwd=7uOZK99zRBxN1hAPHeBbNocpTfgDZS.1
Meeting ID: 692 3135 1515
Password: 535474

This will be a hybrid meeting, and the information about the physical room and zoom will be added to the HARMONY agenda in the next days.
We look forward to seeing you there, whether in person or online!

[Lucian Smith]

I feel like there's two options for this.

One, we just change the list of UniformTimeCourse children to allow this and other configurations.  Another thing people sometimes want is the ability to have irregular outputs, so you just give it a start and and end, and the solver determines when to output points based on how much the state changes.  Or the ability to ask for an explicit list of particular time points as a vector (i.e. [0, 1, 2, 5, 10, 20, 50, 100]).

A second option is to use the 'analysis' class and just define everything with KiSAO terms.  You give the algorithm the 'uniform time course' term, then define however many algorithm parameters you want.

The nice thing about the second option is that it already works; the bad part is that fewer solvers will be able to import it without 'word of mouth' campaigning.  The nice thing about the first option is that it sets clear expectations for SED-ML interpretation going forward; the bad part is that we need a whole new spec out before it'd become official.

(Also, thanks Matthias for arranging things!)

-Lucian

To view this discussion, visit https://groups.google.com/d/msgid/sed-ml-discuss/SY0P300MB0755630A178B7354A504C6DADBA82%40SY0P300MB0755.AUSP300.PROD.OUTLOOK.COM.

[David Nickerson]

Hi all,

I've started an agenda / notes document here: https://docs.google.com/document/d/1wBkKvEx4cY1DFlBT2jG36HnKqTW_5LU3Q-A5vU9mBtk/edit?usp=sharing. Please add any other items that you want to make sure are discussed next week at HARMONY.

Cheers,

David.

To view this discussion, visit https://groups.google.com/d/msgid/sed-ml-discuss/CAHLmBr2XaOjf7k_ntCVaAzzWgVP8zAm3qgu6wz4AWxT%3Ds0Audw%40mail.gmail.com.