Summer of HPC: Exploring Edinburgh, Python coding and much more

Greetings from Edinburgh’s sunny festival season.

During the past few weeks I have been both exploring this vibrant city and cultivating my Python programming skills. I will give you more details about how my work has been going, but first let me introduce you to some of the adventures I’ve had. From Edinburgh Castle to museums of modern art and street performances, this city has it all. The local pubs have their own vivid rhythm, and the traditional delicious fish and chips combination. We were even brave enough to try haggis and deep-fried Mars Bars here. 

But I think the most adventurous day I have had during this trip was hiking in the Pentland Hills Regional Park. The route was challenging and the weather unpredictable, but we managed to complete our mission and climb up and down the hills with the moor flowers.

Let’s take a hot chocolate, a cup of tea now and take a closer look at my journey with Python and Geology!!!

The idea to use automated algorithms to facilitate the work done by geologists (as far as geological interpretation is concerned) is not new, but a recent increase in research conducted in the field of machine-learning and the implementation of neural network methods proves that it is time to revisit the topic. That was the motivation for my SoHPC project at EPCC in collaboration with the British Geological Survey (BGS).

During my summer internship, I have been familiarizing myself with methods concerning the simultaneous correlation of multiple well logs and have also been understanding their current challenges and limitations. This research field has been very new to me but still I find it very exciting. In general, I have focused on implementing Python tools and libraries to the preprocessing of well log data from the Netherlands and the Dutch sector of the North Sea continental shelf. Understanding which parameters and measurements will be of greater use to geologists has also been part of my work conducted through publication and literature search.

After the preprocessing is complete I will implement already existing well log correlation methods based on the preprocessed data. In this way, geologists will be able to evaluate the limitations of the existing methods and the quality of specific well log measurements. The next step would be to create a neural network-based application for the automatic and simultaneous well log correlation. The preliminary design and concept on which I am currently working has proven to be challenging as well.

So I am really looking forward to the final results of this effort in the future and I am more than excited to be participating in such an interesting project.

And just to give you a glimpse of well log correlation results, please see below.

Left: Density distribution as a function of depth for various well logs. Right: After well log correlation the density of the well logs as a function of relative geologic time.

This post first appeared on the Summer of HPC blog.