The EPIC WIN Thread 3: SON OF EPIC

Today's been a good day on the tech scene. Got the new 6s for me. Mom gets my 5s to replace her worn out 4s. AT&T had a special on the iPad mini LTE for $50. I got that, and mom gets my old 3rd generation wifi only iPad.

Signed up on the Amazon Prime sale and linked accounts, so she gets all the perks without having to ask me to do all the ordering. The Kindle app is on the iPad for her, so now she can get all the books that weren't compatible with her old school kindle.

And did I mention an interview with a different hotel on Tuesday. A former coworker moved to a management position there, and was overjoyed when I put my application in. :) (He's not doing the interview, though.)
 

fade

Staff member
A project I have been working on for 2 years finally paid off today. This was a lot of work. Thousands of lines of code in multiple languages. Heavily computational, as in runs on hundreds of processors for days, consuming hundreds of GB of RAM. But it works.

 
A project I have been working on for 2 years finally paid off today. This was a lot of work. Thousands of lines of code in multiple languages. Heavily computational, as in runs on hundreds of processors for days, consuming hundreds of GB of RAM. But it works.
. . . I just realized how we could've solved that problem Douglas Adams passed. Several years ago, this is what should've happened:

Ken Jennings: "I'll take Life, the Universe and Everything for 2000, Alex"
Alex Trebek: "42"

Ken Jennings answers in the form of a question. The question.


Anyway. So your code does something, fade. What does it do?
 
I went on a ten hour road trip with my coworkers today. I took close to 1,200 pictures and saw a lot of wildlife including several herds of elk. Even though I'm in rough shape with my arthritis I managed to hike down to a waterfall and back up!

I'll pay for it tomorrow but it was a fantastic day.
 

fade

Staff member
Anyway. So your code does something, fade. What does it do?
I just saw this. Ooops. It's hard to explain in a short span, but it models and inverts electromagnetic geophysical data from deep ocean EM surveys. There are devices that function like metal detectors, only with antennas that are a kilometer long. My expertise is in these instruments. In particular, modeling the hypothetical response of a theoretical earth. Why is this important? Because modeling is part of something more important called inversion, which as the name implies is the opposite of modeling. You start with a measured response and try to find the hypothetical earth that produced it, and if you've done everything right, that hypothetical earth should approximate the real one. This is also called things like "tomography" or "imaging". It's very computationally expensive, and the answers are non-unique (one earth produces one response, but the same response could have come from multiple earths).

Another device that works the same way is an MRI. It's basically the same instrument, except it's less computationally intense because you have the luxury of putting sensors on all sides of the body and you also know what humans are made of.

The goal is to measure the electrical resistance (technically the resistivity, which is resistance/volume) of the earth beneath the ocean, because big, flat resistors that conform to certain geologies are possibly oil.

EDIT: Not sure if I made it clear, but the response you measure with one of these devices is not very useful on its own. You have to invert the response, producing an image of the earth in terms of electrical resistance, to make the best use of it.
 
Last edited:
EDIT: Not sure if I made it clear, but the response you measure with one of these devices is not very useful on its own. You have to invert the response, producing an image of the earth in terms of electrical resistance, to make the best use of it.
I used to work at TGS-NOPEC (a large speculative seismic company) like 15 years ago --I totally followed everything you wrote :)
 
I've read a lot about NVIDIA's GPGPU marketing and promotion, which means I followed about 60% of what you wrote.

--Patrick
 
You explained quite well, so while I felt lost at the beginning, by the end of your post, I followed about 60% of what you wrote and understood what you meant :p
 
layman's terms:

You can't really look under rock in the ocean to find oil. So you have to look indirectly. You can do things like shoot sound into the rock and see how it bounces back (siesmic) or shoot EM into the rock and see how it reacts.

Like this:


The data that is returned is just that: Data. It's not a picture of what's under the rock. It's just a stream of data that says what happened to the EM or sound once it entered the rock and returned. So they have to interpret it. The software looks at the results and tries to answer the question "what kind of land formation could produce this data". The data could imply multiple kinds of land formations, some oil bearing, and some not.

Here's an example of a 3d seismic survey and a possible interpretation.



Back when I worked at TGS, most of this work was done on several-feet-long printouts of seismic running up and down our office hallways, where geologists pored over them with colored pencils in hand. If you think IT guys are the high paid golden boys in most companies, we had nothing on these guys. They were making 4 or 5 times what I was at the time, and most of them had 5 large screen monitors side-by-side on their desks long before it was fashionable (or affordable) to have multiple monitors on one computer.
 
Last edited:

fade

Staff member
layman's terms:

You can't really look under rock in the ocean to find oil. So you have to look indirectly. You can do things like shoot sound into the rock and see how it bounces back (siesmic) or shoot EM into the rock and see how it reacts.

Like this:


The data that is returned is just that: Data. It's not a picture of what's under the rock. It's just a stream of data that says what happened to the EM or sound once it entered the rock and returned. So they have to interpret it. The software looks at the results and tries to answer the question "what kind of land formation could produce this data". The data could imply multiple kinds of land formations, some oil bearing, and some not.

Here's an example of a 3d seismic survey and a possible interpretation.



Back when I worked at TGS, most of this work was done on several-feet-long printouts of seismic running up and down our office hallways, where geologists pored over them with colored pencils in hand. If you think IT guys are the high paid golden boys in most companies, we had nothing on these guys. They were making 4 or 5 times what I was at the time, and most of them had 5 large screen monitors side-by-side on their desks long before it was fashionable (or affordable) to have multiple monitors on one computer.
In a way, seismic is a lot easier. It kind of does return a direct image of the subsurface, but in terms of reflections from (basically) the interfaces between different rock types. You can invert it for the speed at which vibrations move through rock, which depends on rock type. However, for oil exploration, it's a lot more common to create a direct image from reflections. You're limited in resolution by the frequency of the vibrations. A lot of seismic computing is involved in sharpening and cleaning up the image. What you end up with is immediately recognizable to even a layperson as a picture of the structure of the earth. EM doesn't afford you that luxury. There's not much you can take from the direct image (there is one), because everything overlaps into a big blob. You have to invert. And even when you invert it, it's still fairly blurry.

So why use it? Because inverted EM tells you fairly precisely what the fluid content of the rock is. Seismic gives you a clear picture of the rock structure (folds, faults, layers), but not type or fluid content. For years, that was enough. We know oil migrates upward and gets caught under faults, or under the rounded top of layers. But there's also a lot of false positives. EM can tell saltwater (water at depth tends to be briny) from oil. The former conducts electricity well (toaster in bathtub), the other is an insulator (oil is used as an insulator in power line transformers or capacitors). So put the two together, and you stand a much better chance of not drilling a $200M dry hole.[DOUBLEPOST=1444232249,1444231864][/DOUBLEPOST]
I was referring to the folks who do their reviewing of the collected data in VR 3D rooms.

--Patrick
They tried to push this at the university I taught at. They built this $25M 3d vis center called LITE with the Field of Dreams business philosophy. It didn't work. The problem is that it has no real niche. You already have the best hardware for turning pseudo-3D images into 3D between your ears, and the UI inconveniences (despite what hollywood and tony stark want us to believe) far outstrip the minor added value from true 3D reflection.

That said, walking through data on an omnidirectional treadmill is fun.
 

GasBandit

Staff member
Oh, so he's just reversing polarity on the deflector dish. Now I understand.
And so we BOUNCE a graviton particle beam off the main deflector dish! That's the way we do things lad, we're makin' things up as we wish.
The Klingons and the Romulans hold no threat for us, for when we find we're in a bind we just make some shit up!
 

fade

Staff member
Fade and Tinwhistler are talking about way smarter shit than I will ever understand in my life.
Nah, you'd get it if I had the room to explain it better. I keep meaning to move my old geophysiwiki (my attempt at plain-English geophysics, physics, and math) to Wikia. I can put this there, too.
 
My professor made me cry. Not in the way you might think. After getting the email I talked about in Terrik's thread, I wrote him back. I explained that I got my dates mixed up, about the move and only having our internet turned on yesterday afternoon (all true). I was braced for a "tough noogies kid read the syllabus more closely next time" and there goes 25% of my grade.

He wrote me this morning and gave me an extension until tomorrow at midnight. I'll have points taken off for being late, but I have a chance of getting better than a C in the class (and that would only be if I was perfect on all other assignments and the two exams). I'm all verklempt.
 
Didn't mention that one of our cats had bolted on Monday afternoon. Hadn't seen any trace of her all week. But the other times she had bolted, she would turn up at the door during hours when I was at work this week. So about 20 minutes ago, I go downstairs for coffee, and there she is at the door. Of course when she has gotten out, she doesn't want to get brought back in. It took a couple of tries to get a hold of her, but now she's back safe and sound.
 
Top