UQWorld

Hello to our first 50 or so members (up to 18/10/2019)

A couple of weeks ago, UQWorld surpassed its first 50 members milestone. Other than the effort of @bsudret and @ste to introduce UQWorld to the UQ community at large during their speaking trips, we have largely allowed UQWorld grows rather organically, by inviting people from our circles and referring some UQLab users to come when they have usage questions. There’s even a small chance that you just stumbled here (it’s the internet).

But regardless of how you end up here, hello and welcome!

You will find we are friendly and open in this community. This is a safe space where you can get connected with people from around the world, from a wide range of scientific disciplines to discuss your UQ-related problems or share your thoughts on UQ concepts and practices.

In this topic, we would like to make it easy for you to introduce yourself to the community and ask your questions.

But before that, let’s recap…

Where to start? :beginner:

If you are still finding your way around UQWorld, take a look at these posts:

Learn a thing or two about UQ :open_book:

Across UQWorld you can find resources about UQ concepts and practice. Here are some examples:

For UQLab users

If you are a UQLab user, we have a dedicated section for you in UQWorld:

And now…

Say hello to our first 50 or so members :wave:

Please introduce yourself! How did you find us? What prompted you to join?
What are you working on right now? Would you like to share it? or perhaps need help with it? Go ahead and hit the reply button.

1 Like

Let me start myself:
I’m Damar, I’m a nuclear engineer by training, and currently a post-doc at the Chair of RSUQ.

I help the founders of UQLab (that’s @bsudret and @ste) to establish UQWorld;
we want it to be a community for UQ researchers and practitioners at large, and not just for UQLab users (I was not myself during my Ph.D. study :wink:). So if you have any suggestions, please let me know!

2 Likes

Hi

I’m a PhD Scholar at the IITB-Monash Research Academy. I’m a structural engineer by training.

My PhD research involves estimating the value of structural health monitoring for degrading structures using reliability methods. It involves lots of reliability estimation, bayesian updating, metamodelling etc. This is why UQLab is a very attractive toolbox for me. I like its black-box model structure a lot due, thus making itself very versatile and flexible.

I was introduced to UQLab by my supervisor, Prof Siddhartha Ghosh. I believe its a very valuable contribution to the scientific community.

EDIT: I’m interested in knowing how regular users can contribute to UQLab.

1 Like

Then let me continue:

My name is Paul and I am a third-year PhD student at the Chair of RSUQ. I joined the chair following the completion of my Master’s degree in Civil Engineering.

The topic of my PhD, quite generally, is the use of surrogate models in the context of Bayesian model calibration. I am also the main developer of the Bayesian module in UQLab which is why you will find me answering most of the questions related to that module here on UQWorld :thinking:.

If you have any comments about the module or suggestions for its improvement - don’t hesitate to contact me or directly post something in the feature request section of the forum.

2 Likes

My turn then:

I am Xujia, a third-year Ph.D. student at the Chair of RSUQ. I studied mechanics and mathematics in my master’s program.

My Ph.D. thesis is entitled Surrogate modeling for stochastic simulators using statistical approaches, which is a part of the SNSF project surrogate models for stochastic simulators. Therefore, my research focuses on developing appropriate methods to efficiently and accurately emulate stochastic simulators. I look forward to having fruitful discussions in UQWorld!

Regarding UQLab, I help the administrators to maintain and possibly enrich the registration and license systems. If you encounter any problems with your UQLab license, please do not hesitate to send a support request to uqlab@ethz.ch.

2 Likes

Hi @Shihab_Khan, welcome to UQWorld! Happy to see you and @paulremo having a productive time here. About that contribution you mentioned, thanks a lot for the interest! I think there might be something in the pipeline :slight_smile:

Thanks for being active here, @paulremo. And what a response time in the Q&A :clap:

@xujia, thanks for chipping in! That post on the stochastic simulator is great! :tada:

Who would pick up the baton next :question: :smiley:

1 Like

Hello,

I am Sophy, a 3rd year PhD student at Oxford University in the Earth Sciences department. My background is in Oceanography, and I’m now using “black box” optimisation methods to improve the parameters in ocean biogeochemical models. These models are very computationally expensive, so I’ve been using UQLab to replace them with Kriging / PCE surrogates, and then optimise the surrogate instead. I’m finding that the surrogates aren’t sufficiently accurate for parameter optimization (because I can’t afford to train them with too much real data and my function landscape is incredibly noisy), but they’re great for an initial global parameter sensitivity study using the Sobol Indices.

I was introduced to UQLab on a recent summer school (Modelling and Numerical Methods for Uncertainty Quantification), and I’ve found it very helpful and welcoming so far.

Hi @Sophy_Oliver, welcome to UQWorld!

I was wondering if you have explored some possibilities of using surrogate specifically for optimization purpose. As an example, there is a family of methods based on the so-called Efficient Global Optimization (EGO, based on the work of Jones, et al.). And there’s also the this work by Picheny, et al. that might give you some overview on noisy model. Perhaps you are already familiar with these?

These methods are not directly available in UQLab but you might be able to embed the surrogate created by UQLab in some of these algorithms and create a customized workflow.