How to decide between 2 fields of research(Astrophysics vs Brain Modelling)? (1 Viewer)

Users Who Are Viewing This Thread (Users: 0, Guests: 1)

Hi,
I get my grad school decision in about 2 weeks.:cry:
I'm having a problem deciding between 2 fields that I would love to do a research project in. Both involve Computational/Simulations Science [Computer Science(computability and 3D engine development) & Mathematics].

The 1st is ALife/Brain Modelling/Adaptive Learning(Neural Networks,Evolutionary Techniques).
The 2nd is Astrophysics/Planetary/Geophysics(multiscale science)

The question I have is how would you choose between the two? What questions would you ask if you were in my place?

I am passionate about both and began studying them at the undergrad level before I got lost. And i can see myself in either field. I have been told that you should only focus on one project which has led to this indecision.
Also is a student allowed to pursue 2 Phd's(1 after the other)? Is there any point?

Thank you for helping me, in advance
Jack
 

berkeman

Mentor
54,306
4,714
I'd choose the neural network path, but that's just my personal preference. There are some great potential advances in the NN field, IMO. In your other choice, would you be focusing on practical things like Earth weather modelling and earthquake modelling, etc.? Or would you be wanting to focus more on the astrophysics aspect?
 
hi berkeman thanks for replying. I was thinking of designing code to handle both aspects actually. My main interest is creating generic code for N-body simulations of various levels of science but the project itself would be focused on any of those mentioned above(i guess it would depend on the consulting supervisor).

So you think there's greater potential in NNs and Adaptive Learning. I will have to take that into great consideration.
Thank you again.
 

berkeman

Mentor
54,306
4,714
What hardware are you targeting to run you code on? Will it rely on massive parallelism? For the kind of simulations that you are talking about, it would seem like a good idea early on to decide what physical platforms are best, and use that to help you plan your code architecture.
 
berkeman: I'm hoping to start off with one computer...for simplicity(still learning about Nbody)... but once I'm in the 2nd year(depending on how it is taught in grad school) i'd like to go into high computing/parallelism.
Do you know of any references(books/websites) i can take alook at to understand the difference in coding architecture that you mentioned above(going from a single PC to a supercomputer)
 
Last edited:
784
1
My choice would be neural networks, but that may be because I recently read Complexity, by Waldrop. :smile: I would second what berkeman had to say about this field, though.
 
I officially got accepted so does anyone else have advice? I got till sept to decide i think.
 

0rthodontist

Science Advisor
1,230
0
neurocomp2003 said:
Hi,
I get my grad school decision in about 2 weeks.:cry:
I'm having a problem deciding between 2 fields that I would love to do a research project in. Both involve Computational/Simulations Science [Computer Science(computability and 3D engine development) & Mathematics].

The 1st is ALife/Brain Modelling/Adaptive Learning(Neural Networks,Evolutionary Techniques).
The 2nd is Astrophysics/Planetary/Geophysics(multiscale science)
Neither of those involves computability theory. You mean algorithms and numerical techniques?

Personally I would also prefer neural networks and AI. The industrial age automated and made reproducible the work of manual labor--the computer age will automate and make reproducible the work of thought.
 
Last edited:
computability theory isn't used directly in them but its about studying whats possible and whats not(or whats efficient and whats not), algorithms and numerical techniques IMO is part of the foundations of 3D-engine development.
 
Last edited:

0rthodontist

Science Advisor
1,230
0
Computability theory is just about what's possible or what's not, not about efficiency. Pretty much anything you'd be doing in those fields is going to be possible. I think computability is applied more in fields like compiler design, where it helps to know that no matter what you do you can't write a perfect tool to detect unreachable code.
 
Last edited:

The Physics Forums Way

We Value Quality
• Topics based on mainstream science
• Proper English grammar and spelling
We Value Civility
• Positive and compassionate attitudes
• Patience while debating
We Value Productivity
• Disciplined to remain on-topic
• Recognition of own weaknesses
• Solo and co-op problem solving
Top