Potential topics to share in sciwork conference 2024

[yungyuc]

The sciwork community is preparing for their second conference in December 2024. Our recent work can be organized as the technical talks. I list what I can think of now, but there may be more topics:

1. C++/Python dual mode serializer (@tigercosmos @q40603 @ThreeMonth03)
   • issue Prototype profiler object serialization using JSON #343
   • issue Wrap the C++ profile object (CallProfile) to Python #344
   • issue Serializer/deserializer for hierarchical, multi-type data #348
2. Profile to speed up the 1D Euler solver (@tigercosmos @q40603 @ThreeMonth03 @j8xixo12)
   • The work may be extended to time-series code development (conversation Time series analysis #380)
   • issue Add profiling system for the shock tube problem #135
   • conversation Plan to improve the profiler #345
   • issue Prototype profiler object serialization using JSON #343
3. Qt camera control and testing in Python (@Tucchhaa)
   • issue Improve camera control #70
   • conversation Shortcut settings menu #412
4. Combine the 1D Euler solver with Qt for in-situ analysis (@Gene0315 @j8xixo12)
   • issue Enhance modmesh Euler 1D app GUI #202
   • issue Add more NRBC treatment in the 1D Euler solver #330
   • issue Adjust the boundary to improve the plotting #335
   • Extended work for any linear 1D solver is tracked with issue Add a pluggable linear solver of a vector equation #355
5. Time-series code development from scratch (@j8xixo12 @tigercosmos @xingularity)
   • conversation Time series analysis #380
6. Unstructured body mesh of mixed elements in two- and three-dimensional space (@yungyuc)
   • issue Port the 2/3D Euler solver from the legacy solvcon code base #404
7. Use Bézier curves operations interactively and in batch (@YenPeiChen07 @Gene0315)
   • issue Prototype drawing capability using Bezier curve #238
   • issue Use Bezier curve to fit NACA 4-digit airfoil  #320
   • issue Linear regression for cubic Bézier curve approximation #389
8. Package Windows executable (@terrychan999)
   • issue viewer Windows executable should include dll for portability #194
9. Building process (including dependency) for Mac and Windows

I added the names of recent contributors in each of the topics. We don't have to do a technical talk, but it usually is a productive way to organize our know-hows and becomes a stepping stone to sharpen our skills.

We can use this conversation to discuss the plan, and create new ones when needed.

[ThreeMonth03]

It sounds like a good idea. I'm interested in the issue #343, and I will attempt to solve it as my conference topic. If the remaining time allows, I'm willing to join the discussion and development in #380.

[ThreeMonth03]

Thank's for your useful suggestions, but I have some questions:

1. When should I finish the draft?

As soon as possible. The first draft is a starting point to discuss what to do for your presentation and code.

2. Should the draft include the topic of time series? If no, it seems that I only can talk about the reentrancy for the serializer.

It's up to you. It's not a bad idea to keep it simple in the beginning. More contents can be added after the first draft.

3. Sould I use Chinese/English for the oral presentation?

It's up to you.

Ok, I will write the draft this Friday.

[ThreeMonth03]

Know your enemy and know yourself. In the context of high-performance computing, it is crucial to identify and optimize performance bottlenecks. A C++ profiler is an effective tool for pinpointing these hotspots. In the modmesh project, I contributed to the development of a profiler, and I would like to share several key aspects of it. Here are the topics I will cover:

1. The data structure and basic principles of the profiler.
2. The statistical features of the profiler.
3. The reentrant serializer of the profiler.

[yungyuc]

@ThreeMonth03 what is the subject and what is the abstract?

An abstract without bullets will be preferred. Abstracts of well written sentences connect the rationale more clearly.

[ThreeMonth03]

@ThreeMonth03 what is the subject and what is the abstract?

An abstract without bullets will be preferred. Abstracts of well written sentences connect the rationale more clearly.

@yungyuc Sorry for my late reply. It seems that I cannot make a presentation because I'm busy for researching.
However, I can lead @Vincent-Lien to fulfill the serialization of the profiler.

[yungyuc]

That sounds good. We can discuss here and in discord.

[j8xixo12]

I'm interested in time series analysis, especially its application in analyzing sensor data in aerospace field.

I’m willing to give a presentation!

I will draft abstract and subject after studying some materials about FFT, I think it will take about 1 week.

[yungyuc]

Sounds, great, @j8xixo12 . One of the simplest form of the draft can be a comment under this thread or use a new conversation.

[Gene0315]

I think I need to finish the issue #391 (follow the issue #335) first. I should have done it long ago. And then, I want to contiune the issue #330 to add the NRBC treatment.
Also, I'm interested in the Bézier curves fitting.

I have a question. Do I need to write a draft?

[yungyuc]

Writing a draft will facilitate your development. I think you can write two for both of your topics for the 1D Euler solver and the Bezier curve fitting. What about starting with drafting the subjects for the presentation?

You do not need to give two presentations (but you have the right to).