/sci/ - Science & Math

What's life like doing a PhD? Going to pursue a PhD in astrophysics so I'd like to know.

Are you sad that you are going to miss out on 300k starting?

For a three-dimensional system of equations, and given some initial conditions, prove that smooth solutions to the navier-stokes equation always exist, or that if they do exist they have bounded kinetic energy.

>>3225356
Bumping for answer, the exact same situation here.

>>3225356
>What's life like doing a PhD?
Um, really depends where you're at and how your advisor is. Some people I know have very structured routines with weekly lab meetings with their research group/advisor, while a lot of people (like me) are very self-directed and able to delve into whatever catches our fancy.

Socially it's... well, my hard partying days are long behind, so i enjoy the semi-adult life.

>Going to pursue a PhD in astrophysics so I'd like to know.
Cool, I know some guys in the area.

>>3225368

Not OP but it will depend on the system that you use to solve it. Some are conditionally stable whilst others unconditionally so.

>>3225379
>thinks he knows answer to a millennium prize problem

>>3225381

I never said I knew a solution, but you need to be a bit more specific in your requests.

Please give a layman's explanation for why a wing forced to move in a forward direction at sufficient speed experiences an upward force. Not Bernoulli.

>>3225372
What programming language / languages do you use?
Do you develop own systems to solve your problems or do you mostly use already existing ones?

What did you have for dinner last night?

>>3225368
meh. If it were ever to be solved, I fully expect the answer to be that unique solutions always exist.

>>3225387
>more specific

why?

the problem is smoothness and existence for ALL initial conditions

what's your favourite pr0n?

>>3225392
I'm very adept in Fortran (luls), C, C++, CUDA, MPI, OpenMP, Matlab... I generally prefer to write my own solver software, but depending on circumstances I'll use what's provided. All the post- and pre- processing goes on with industry standard packages though.

>>3225393
>What did you have for dinner last night?
chicken tacos

>>3225398
I'm bothered by the framing of the question. They explicitly use the incompressible version of NS, which strikes me a bit odd. I'm sure some pure mathematician could inform me of the reasoning behind it. In the hyperbolic case, at least when it's inviscid, it's trivial to prove that smooth and unique solutions do not always exist.

>>3225406
what do you do in fluid mechanics?
just solve PDE's all day?

>>3225421
in an abstract sense, yes. but i'm also very much on the applied side, where we have to analyze and design specific geometries.

on that end, it's like being an experimentalist with a wind tunnel, except we're not in the dark ages, and my wind tunnel is software running on computer clusters.

>>3225406
Do you use Fortran just because there´s so much already written solutions to many problems lying around - or is it really that much better at numerical problems?

What level of maths is required?

Is it fun?

Do you have a physics or engineering background?

How does computational fluid mechanics compare to normal fluid mechanics? I have CFM as a module option for my masters next year and am looking at what I might do.

>>3225451

Hes a rocket scientist, hes got all of them backgrounds
-

We were being taught Fortran too, but I dropped that shit to get ahead on math and physics so they dont kick me out next year. Will I have trouble teaching myself?

>>3225466
nobody uses fortran anymore

you'll only need it if you're modifying or debugging old software.

>>3225436
Using/knowing fortran is strictly due to legacy code. It is useful to know (although considerably easier than most any more modern language), but I'd never start up a new project with it.

>>3225451
>What level of maths is required?
Well, there's the standard multivariate calculus and differential equations just to pose the problem of navier stokes. Actually understanding the ins and outs of the various methods requires knowledge of fourier methods, calculus of variations, linear algebra, complex analysis, more advanced diffeq type stuff than undergrads see, and lots of numerical methods.

>Is it fun?
Well, yes. Fun and challenging, and we produce some of the prettiest pictures in nature.

>Do you have a physics or engineering background?
aerospace engineering.

>>3225500

do you do more than just make pretty pictures. is CFM just procrastinating on computers but with a research budget to keep you happy?

>fluid mechanics

Say I wanted to make a build a missile. Think a model rocket but with a with control surfaces on the fins. The control surfaces will be moved by servos hooked to a microcontroller. Is there some software I can use to simulate aerodynamic forces? Free or available by torrent is a plus.

>>3225506
there is lots of software available, but i seriously doubt you could get anything out of it. there's some canned rocketsim program (that might actually be the name) that freshmen use. understanding the forces and actively controlling them with moving surfaces ain't exactly trivial.

>>3225508
At present, it's superior to experimentalism in many ways, and with the obviously increasing computing power available, that only becomes more true.

it has supplanted wind tunnels as the primary driver of aerodynamic design in all industries. ask the 787.

Answer this OP >>3225401

>>3225570
not /sci/ related!

>>3225651

I guess personal Qs go?

How far in are you? How's it going, etc? Where are you going to work and do research after the PhD?

>>3225514

I googled rocket sim and got this:
http://www.apogeerockets.com/rocksim.asp
Is that what you're talking about?

>understanding the forces and actively controlling them with moving surfaces ain't exactly trivial.

I knew it wouldn't be trivial but I took an EE controls class. I wanted to start with a pretty stable rocket to make the controls side easier. I figured if I had some attitude and altitude sensors I could use feedback from the sensors to make corrections. Maybe I could launch it out of a pvc pipe and maintain level flight. I want to simulate to get get a feel for how much force is made when I move a fin. Is there something I'm missing that will make this really hard?

>>3225651
Trying to see how porn preference correlates with profession isn't science?

>>3225659
>I guess personal Qs go?
anon'ly, sure.
>How far in are you? How's it going, etc? Where are you going to work and do research after the PhD?
did 2 years of MS, 1.5 years of PhD. it's going well.

i really don't know where i'll work/research after phd. there's a lot of options in the aerospace field. professorships are hard to come by, but there's tons of good paying, interesting industry jobs.

HEY OP

which FEM program do you use? im working with ANSYS ADPL right now, goddamn it's a piece of fucking shit.

also what were you studying in undergrad? engineering of some kind?

>>3225695
>Trying to see how porn preference correlates with profession isn't science?
dammit, you got me there. let's just say nothing illegal, and nothing that i would be overly ashamed to mention drunkenly amongst friends.

>>3225682
yeah, rocksim, i forget how detailed it is though.

if you want a first order estimate, it's very easy to determine a general idea of the load on a flat plate of a given area at a given angle at a given speed. 2*pi*(angle of attack in radians) will give you a lift coefficient for a flat plate, multiply by 1/2 rho*V^2*fin_area and you've got a ballpark estimate for the lift, hence the load. that's the layman bit, hopefully nobody wants to bitch and show how smart they are.

How'd you lead into the Phd?
Like what degrees and majors have you done?

>>3225711
>>3225711
>>3225711
>>3225711

>>3225733
i answered all of those, brosef. and yes, ansys ain't great, especially if you don't know how to use it. it suffers from badly dated design. and the not-so-unfortunate need for you to know the underpinnings of the methods in order to use it.

>>3225349

What do you find to be the most fitting programming language for general scientific computing?

What percentage of your source is functional as opposed to procedural or OOp?

What is a good introduction to parallel/distributed computing?

What math subfields are useful in your fields?

>>3225742
>>3225742
>Fortran (luls), C, C++, CUDA, MPI, OpenMP,Matlab

none of these are modelling programs... are you saying you don't use any modelling programs?

How old are you?

What institution do you belong to?

What are you going to do after you finish your Phd?

>>3225651

>not sci related
>>3225406
>What did you have for dinner last night?
chicken tacos

Definitely /sci/ related.

My apologies for having disturbed you, master.

>>3225742
>>3225742
also
>implying I don't know the underpinnings of the methods or how to use it

>>3225712

I love how any curiosity can be phrased as scientific inquiry.

>flat plate
Does that mean it has to be rectangular or the estimation is off? Also something I've always wondered about is the shape of fins. Is there an advantage for triangular fins or are model rocket hobbyists just using them because they look cool? Is there an advantage for moving the whole fin vs just part of it? Picture is what I'm thinking is a triangular fin with only part of it moving in the back. It's also my face if I ever get this to work.

>>3225754
>are you saying you don't use any modelling programs?

i use commercial mesh generation programs for preprocessing, and tecplot for post processing. all my solvers are in-house codes, but things like ansys/fluent can do equivalent things.

>>3225797

I've used NUMECA a fair bit. What's your opinion of that suite?

>>3225745
>What do you find to be the most fitting programming language for general scientific computing?
"general" is too broad. c++ for large scale and parallel stuff where performance is of utmost priority. old fortran codes also fit this.

>What percentage of your source is functional as opposed to procedural or OOp?
i'd say it's all OOP design, but i think functional programming kinda fits in there. no matter what, at the end of the day you have a procedural basis for all numerical methods.
>What is a good introduction to parallel/distributed computing?
llnl.gov has tutorials on openMP, MPI, and general parallel computing. those are the very best, at least on the numerical side.

>What math subfields are useful in your fields?
covered above.

>>3225807
never used it. for research, it's not really okay to use closed-source software where you don't know everything going into the methods, IMO.

>>3225797
>>3225797
that's pretty impressive. what numerical methods do you use for cfd? I don't know much about fluid dynamics but it seems fairly interesting. does the navier stokes equation underpin pretty much everything you do?

and what's the reason you don't use the commercial packages? licenses too expensive? also can you post some pictures of your work, pretty postprocessing plots etc.

What is your fetish, OP?

How social/extroverted do you have to be in order to be succesful?

What is 6/2(1+2) = ?

what was your undergrad gpa?

how did you achieve it? study habit tips plz

>>3225849
my ugrad gpa was actually quite poor, but i managed to impress the right people. i am probably the worst "studier" ever, but a lot of that makes me fairly good at particular research tasks.

>>3225842
>that's pretty impressive. what numerical methods do you use for cfd?
time integrators tend to use implicit or explicit runge-kutta methods from basic ODEs, spatial discretization can use fourier/spectral methods, least-squares interpolation, or simple numerical differentiation, numerical integration.... on the solution of linear systems, things like conjugate gradient methods, simple gauss seidel methods, etc.

>does the navier stokes equation underpin pretty much everything you do?
yup. conservation of mass momentum and energy.

>and what's the reason you don't use the commercial packages?
well if you're going to publish research, you need to be able to say *exactly* what the methods used are in great detail. commercial, closed source solvers generally make that impossible.

>also can you post some pictures of your work, pretty postprocessing plots etc.
sorry, it'd violate anon. there's billions of pretty pictures similar to what i do. browse tecplot's site or something.

>>3225881

so you got into a masters program with a low gpa? im confused, explain your tricks to me and how you impressed the right ppl sir

why waste time in such a stupid field when you could apply your computational knowledge to so many much more beneficial fields?

>>3225938
>so you got into a masters program with a low gpa? im confused, explain your tricks to me and how you impressed the right ppl sir
professors teach undergrad classes. they also advise and fund graduate students at the school, in addition to having recommendations that carry weight with others.

if you do really well in a professor's class in an area you're interested in doing research in, then there's no black magic there. i was offered research work without the prof ever knowing my gpa, because it was of no concern. then you have a foot in the door.

>>3225960
a good point, but what's awesome about my field is that we attack what are objectively the nastiest PDEs in existence with massive computational power. because of the myriad difficulties involved, it enables people in our field to attack other problems like computational biology, plasma physics simulation, finance, etc.

i do plan to branch out from fluid mechanics more, but what field do you have in mind?

>>3225981
A.I., neuro interfacing, etc.

>>3225981
how do you know you aren't wrong. like weather forecasts are sometimes wrong.

did you ever cause a billion dollar airplane to crash?

>>3225981

go into quant finance, make dat 300k

>>3225349


Have you ever tried to simulate urine?

>>3225993
those are pretty far outside of my capabilities, on the math/physics end. i do work on optimization, which one could call a branch of AI... admittedly i'm interested.

>>3226004
>how do you know you aren't wrong. like weather forecasts are sometimes wrong.
there is a lot of boring (to the layman) validation and verification that goes on, but it still takes experience and knowledge to read the output of a computer program and make a valid judgment as to it's applicability.

basically, we know when we simulate certain situations that it's "right within 1%", "right within 5%", or "fuck i dunno, it's alright, but it needs further study." it's the nature of the beast.

What is your favourite maxwell equation and why?

>>3226024
blood, yes, water, yes. i expect the fluid properties of urine are probably pretty identical to water.

>>3225972
This give me hope OP.

Thanks!

>>3226032
faraday's law, obviously

>>3226045
keep in mind this doesn't give you the right to be stupid or lazy (one or both is probably the reason for a low gpa in the past). you just have to find something you have a knack for and enjoy, or are bad at but enjoy so much you bust ass at it and eventually do it very well. and then also be sociable enough to get the prof to like you.

>>3225852

not op but it's 9 and now gtfo

>>3226050
Wrong, the nonexistence of magnetic charge is the best

Fluid mechanics is so depressingly dull. How do i revise for my exam in it next week without killing myself?

>>3226164
fuck, you got me this time

Do you have to choose any speciality when you are a bout to finish your degree there in the US? As in, engines, aricraft/missiles, airports, or space science. I am still in my second year and undecided

>>3226256
for a typical undergrad, no not really. there are a handful of more specific "tech electives" at most universities, but it's not terribly important, unless you're after grad school.

I'm starting out at a community college, hoping to transfer to a four-year university to get some type of mathematics degree.
However, I'm accumulating credits slowly because I'm 23 and I have to make money and do adult real-life things instead of focusing solely on education.
The only things I have going for me are a passion for learning and a wad of school funds.

In your experience, do you have to be a young hotshot to get somewhere in the field? Is there any hope for me or will I end up teaching middle-school algebra?

>>3226463
>In your experience, do you have to be a young hotshot to get somewhere in the field? Is there any hope for me or will I end up teaching middle-school algebra?

I also did introductory classes at a community college, and didn't get my BS till 8 years after high school.

What do you mean by "get somewhere"? I know tons of people that started at your age and older and did perfectly well, and landed good industry jobs after undergrad. Some were guys that did military stints before college, some just older guys going back to school.

i would like to study differential geometry later, down the road. What are the prerequisites to understand it?

>>3226494
differential calculus and some linear analysis, i'd imagine. i don't really deal with it explicitly.

OP
Others will flame me but i'm very interested. How much is your IQ? And your colleagues average IQ?

>>3227078
good picture.

>Others will flame me but i'm very interested. How much is your IQ?
honestly... high. i've only ever done the silly online iq things, and not for maybe 8 years. i've drank a fucking lot of booze since then. i generally scored between 145-165 back then, but i expect i've boozed myself back to normalcy.

>And your colleagues average IQ?
i have absolutely no way of knowing. obviously other people working towards a phd in this field are outwardly somewhat bright.

i think it's an odd metric, and one that can do harm. i was absolutely an underachiever as a teen, but everyone knew i was bright. but i developed no work ethic. then i get to college, and while i'm still quite bright by comparison, i interact with professors that, for the first time, really really humble me. i realize quited explicitly that i'm not the smartest guy around. as i've gone deeper into the study, i realize that no matter how high an iq, you have to put in a lot of work to succeed.

Fascinating thread, OP. High school senior here about to graduate, going to study physics next year. Thanks for sharing some of your experiences.

>>3227286
Great, good luck. A coworker/colleague/officemate of mine did his undergrad in physics.

One thing I love about the applied physical sciences area is that people from a lot of different backgrounds come back together to work on the same problems. Engineers, physicists, and mathematicians all publish in the same area, and we're academically better for the diversity.

Why no answer to 3225391 ? I only wrote "no Bernoulli" because the traditional soft science answer strikes me as no answer at all. Any explanation you want to give is better than none.

>>3227320
haha, sorry, figured wikipedia had my back on that one. i'm sorry to say we don't really get trained in throwing out layman explanations. let me ramble.

i'm guessing from your statement that you're aware that the equal transit time is a fallacy.
so for a semi-layman explanation, i'll refer you to the kelvin circulation theorem:
http://en.wikipedia.org/wiki/Circulation_(fluid_dynamics)
http://en.wikipedia.org/wiki/Kelvin_circulation_theorem

important here is that it's the "circulation" mathematical idea that gives you lift. that's why a spinning drum/baseball/circle creates lift.

in an incompressible flow, you have constant density by definition, and so physically you can think that if there's more stuff in the way on the top of a wing (like it's camber/thickness) it's gonna have to travel faster to the back to maintain constant density. bernoulli's principle is certainly correct in saying that the faster moving fluid is going to have lower pressure, so you get lift.

>>3227320
>>3225391

The viscosity of air causes it to "stick" very slightly to the airfoil. Think of the front end of the airfoil as "grabbing" the air and the back end as "slinging" it downward. Since the air is given some downward momentum, the wing gets some upward momentum.

>>3227357
this is an okay hyper-layman description.

what's really interesting about this idea is that since it has to go faster over the top, you have a circulation in one direction... but following the wake, you have a circulation in the opposite direction. and you can determine lift of the actual body by the circulation in it's wake.

>>3227370
Although, I dunno how "okay" it is, since it's ignoring a lot of other things, but it's hard not to. you've implicitly pretended that the wake is the same as the free stream, and ignored what happens in compressible/supersonic flows....

>>3225349
>procrastinating PhD candidate in computational fluid mechanics

Do you make F1 cars go fast?

>>3227391
Yes. Poor bastards have their track time cut down so they have to throw stacks of cash at CFD.

>>3227374
So what you're saying is that I'm ignoring changes in the energy of the flow? Doesn't conservation of momentum hold regardless? And how else would you describe the transfer of momentum other than through redirection of the flow via viscosity? Honest questions, I'm about to start grad school for aerospace but I'm more of a dynamics guy and it's been a while since I've dealt much with flows.

op, can you answer these questions?

>>3225760

i'm not the one who originally asked, but i would also like to know the answer to them. i am looking forward into my PhD one day. i do research in building aerodynamics. yeah, there is such a thing.

>>3227357
you're momentum explanation is correct... but you're "sticky" explanation of air isn't. There is no intermolecular force strong enough (inb4 gravity or van der valls) to cause air to want to "stick together".

viscosity is only useful in creating a starting vortex that insures the kutta condition is matched. Once the kutta condition holds, circulation and Bernoulli principle explain lift really well without the need for viscosity. Drag is another story however...

>>3227489
well, whenever you have viscous effects, you also have a loss of energy by friction, same as for any solid. "conservation of momentum" holds so long as you account for those losses. hence the use of conservation of energy.

>>3227585
i don't like this either, because even without the kutta condition matched, you can and do still generate lift.

>>3227511
I think I already did, as well as I can. I'm in my late 20s, I go to a very good university whose name I won't say because it'll destroy any anonymity, and I don't yet know what I'll do post-PhD. The options are postdoc, direct professorship, applied research work at a private aerospace company, and mcdonalds. I'm fairly open to the first 3 and it depends on what I find most interesting at the time.

>>3227603
I thought those losses wouldn't interfere with conservation of momentum? If a box skids to a stop on the ground, doesn't the ground receive an equal and opposite impulse to that required to stop the box, regardless of whether the friction caused heat to be generated?

>>3227660
the important term here in navier stokes is the diffusive (viscous) therm.

when you consider a box sliding to a stop on the ground in terms of conservation of momentum, you don't look at it in terms of the impulse imparted to the ground, right? at the beginning, you have a block sliding on a still ground, and the end, nothing is moving. the block stopped, and the ground isn't moving. so your system had momentum initially, when the block was moving, but later on, your system has no momentum, cause everything is stationary, right?

so momentum isn't conserved, so you have to consider conservation of energy. the movement of that block was converted to heat energy by the viscosity/friction of the movement.

>>3227684

Actually if you are looking from an inertial frame, aren't the block and ground moving at the same infinitesimal speed at the end? The kinetic energy of the block gets converted to heat, sure, but the collision, even if inelastic, should result in the same momentum of the box+ground system at the end as at the beginning.

>>3227684
viscosity is merely the molecular diffusion of momentum. In a flow with a velocity gradient perpendicular to the flow direction, viscosity is caused by faster particles diffusing towards the slower area, and slower particles diffusing towards the faster area. by definition momentum is ALWAYS conserved. the only way you can add/remove momentum from a fluid is at boundary conditions. whatever the boundary is must absorb the momentum it adds/removes from the fluid. viscosity doesn't eliminate momentum, but rather dissipates the momentum to the entire body of fluid.

>>3227603
this question about lift got asked because a few days ago there was a thread about this. trolls brought up that in superfluid He, no lift is generated by airfoils. this is because it is pure potential flow with no circulation. viscosity is only useful in describing lift to the point where it makes the flow follow (or nearly follow) the kutta condition. once this is matched, inviscid calculations get pretty damn close to the correct amount of lift.

bump