I fear this may not be the appropriate forum for this post, but I don't know where else to go and since every bit of info I can get is critical to making this very important life decision, I'm going to anyways. Very soon I will have to decide whether to continue on with pure math with the intent of getting a phd, or to instead opt for an applied math masters with a concentration in electrical engineering and dsp. I really like math, and I think I would enjoy applying mathematical analysis to solve real world problems a l� signal processing. I have done quite a bit of lurking on this forum and it seems like some of you know quite a bit of advanced math in the areas of complex and fourier analysis, and even branching out into analytic number theory from time to time (way cool). But I generally can't figure out which posts come from people in industry, and which come from people in academia. I like physics and engineering as well and I think I would enjoy the other aspects of dsp engineering: programming, writing reports, working with others, troubleshooting etc. But I also like to be intellectual challenged and creatively apply math, so is this the right job for me? Am I deluding myself about the life of a dsp engineer? Would someone with such a math heavy degree even be considered of value in this industry? Thanks for your time.
Masters in Applied Math w/ concentration in EE & DSP
Started by ●January 8, 2012
Reply by ●January 8, 20122012-01-08
On 8 Jan, 09:27, Eric Proffitt <espr...@gmail.com> wrote:> I fear this may not be the appropriate forum for this post, but I > don't know where else to go and since every bit of info I can get is > critical to making this very important life decision, I'm going to > anyways. > > Very soon I will have to decide whether to continue on with pure math > with the intent of getting a phd, or to instead opt for an applied > math masters with a concentration in electrical engineering and dsp.Are those the two options?> I really like math, and I think I would enjoy applying mathematical > analysis to solve real world problems a l� signal processing.Should somebody who seriously considers a PhD in maths go into DSP... That's like a contender for a F1 driver position seriously considering becoming a bus driver: Both make their living by driving motor vehicles. Yes, there is some maths involved in DSP. No, it's not at all neither challenging, interesting or even fun. If you get into DSP you will be at the mercy of dillettantes who consider it a skill to know how to play a CD on a home stereo system, and amateurs who do know how to use a mobile phone: Both systems rely on DSP, and being a user of such systems is widely considered enough competence to get jobs in managment of DSP'ish projects. Get into DSP if you ar a masochist with Kafkaish leanings; otherwise, stay far away. As for the PhD in maths - that's up to you to decide: Spend 3-5 years to obtain a diploma to hang up on the wall, or spend that time getting a position where you actually will be able to make a living? Then there is the 3rd option: If you *really* like maths, and you *really* like applied work, consider state space methods for time series analysis. The hard work has already been done (check out the writings by Durbin and Koopman); lots of work remains on using these kinds of techniques to solve real problems. I know that's the academic line I would attempt to pursue, if I were able to rewind the calendar by the necessary 25 or so years. But then, if I were able to rewind like that, I might far more likely drop academia alltogether...> I have > done quite a bit of lurking on this forum and it seems like some of > you know quite a bit of advanced math in the areas of complex and > fourier analysis, and even branching out into analytic number theory > from time to time (way cool).Fourier analysis is trivial. The reason it generates heated debate here, is that the majority of people here - me included - have little or no formal training in math. Don't let *our* math discussions decieve you into pursuing a PhD degree!! Number theory - yes, it's used in coding theory. But again, it seems (I don't have the detailed knowledge to really comment) to be plain vanilla engineering maths, that was established centuries ago.> �But I generally can't figure out which > posts come from people in industry, and which come from people in > academia.Doesn't matter. Pursue DSP at most as an hobby.> I like physics and engineering as well and I think I would enjoy the > other aspects of dsp engineering: programming, writing reports, > working with others, troubleshooting etc. �But I also like to be > intellectual challenged and creatively apply math, so is this the > right job for me?No.>�Am I deluding myself about the life of a dsp > engineer?Yes.>�Would someone with such a math heavy degree even be > considered of value in this industry?Eh... having the PhD degree in DSP with more math than was usual for DSP degrees at the time (but nowhere near what a maths degree), and having had a number of 'token' jobs (managment with little or no DSP competence proudly pointing at me and proclaiming "We have a PhD on staff!" but where I have no real saying in what to do or how), I don't really know what to answer: - Yes, I can see that people like you would be of significant value if assigned the right tasks and responsibilites. - The chances of somebody like you ever actually *being* assigned those those tasks and responsibilities are less than for you winning the lottery [*]. I will not pretend to be able to tell you what to do. But I can confidently tell you what *not* to do: Stay away from a carreer in DSP. Pursue DSP if you want, but don't let it become more than a hobby. Rune [*] In the lottery, somebody actually wins.
Reply by ●January 8, 20122012-01-08
On 08/01/2012 08:27, Eric Proffitt wrote:> I fear this may not be the appropriate forum for this post, but I > don't know where else to go and since every bit of info I can get is > critical to making this very important life decision, I'm going to > anyways. > > Very soon I will have to decide whether to continue on with pure math > with the intent of getting a phd, or to instead opt for an applied > math masters with a concentration in electrical engineering and dsp. > I really like math, and I think I would enjoy applying mathematical > analysis to solve real world problems a l� signal processing. I have > done quite a bit of lurking on this forum and it seems like some of > you know quite a bit of advanced math in the areas of complex and > fourier analysis, and even branching out into analytic number theory > from time to time (way cool). But I generally can't figure out which > posts come from people in industry, and which come from people in > academia. > > I like physics and engineering as well and I think I would enjoy the > other aspects of dsp engineering: programming, writing reports, > working with others, troubleshooting etc. But I also like to be > intellectual challenged and creatively apply math, so is this the > right job for me? Am I deluding myself about the life of a dsp > engineer? Would someone with such a math heavy degree even be > considered of value in this industry? Thanks for your time.Seems to me you are looking towards what used to be called "mathematical science" and maybe still is in places, but is more usually called Computer Science. The people in academia I have got to know and occasionally work with think of themselves as mathematicians (one has a background in astrophysics), but are what would be called computer scientists these days. With the prevailing mood of "technology transfer" directed from above, chances are you would have as many opportunities to work with industry as you would want, while being based in an academic research department. Not so much writing reports as research papers and funding applications. Plenty of scope for working on dsp, for example in HPC fields (massively parallel computation, etc). And if physics is a high interest: they probably still need people with heavy math skills plus computing plus EE plus physics and dsp to work on stuff at CERN, on the LHC Grid, or on the proposed follow-ups to the LHC. Huge data analysis and simulation tasks, among other things - in effect dsp applied to the largest digital camera on the planet. The alternative would be to be head-hunted by an investment bank to work on algorithms for option pricing et al (and, inevitably, applying the FFT to stock price movements) - people still have faith that could actually work, despite all the evidence to the contrary. The rule for getting into dsp seems to be, if you like the application area. So, you do audio dsp because you like audio, or comms dsp because you like comms. But not so much because you like math! Richard Dobson
Reply by ●January 8, 20122012-01-08
Eric Proffitt wrote:> I fear this may not be the appropriate forum for this post, but I > don't know where else to go and since every bit of info I can get is > critical to making this very important life decision, I'm going to > anyways. > > Very soon I will have to decide whether to continue on with pure math > with the intent of getting a phd, or to instead opt for an applied > math masters with a concentration in electrical engineering and dsp. > I really like math, and I think I would enjoy applying mathematical > analysis to solve real world problems a l� signal processing. I have > done quite a bit of lurking on this forum and it seems like some of > you know quite a bit of advanced math in the areas of complex and > fourier analysis, and even branching out into analytic number theory > from time to time (way cool). But I generally can't figure out which > posts come from people in industry, and which come from people in > academia. >(ObDisclosure: I'm a BS Computer Science who does self-directed continuing education, and doesn't post here much. I am doing DSP as an educational thing only, and it has nearly no application in my day job beyond understanding some things I've run into.) I am not 100% sure that there is that much opportunity as a a PhD in applied math in DSP. Of the people I've known who had a PhD in applied math, they were: - a hedge fund guy - a CS professor - a systems guy at a defense contractor. *One* other was a DSP guy. In addition, there are several folks who have masters' ( EE, math, the like ) that *have* done DSP but are not *doing* DSP.> I like physics and engineering as well and I think I would enjoy the > other aspects of dsp engineering: programming, writing reports, > working with others, troubleshooting etc. But I also like to be > intellectual challenged and creatively apply math, so is this the > right job for me? Am I deluding myself about the life of a dsp > engineer? Would someone with such a math heavy degree even be > considered of value in this industry? Thanks for your time.It's important to plan, but realize that "no plan survives first contact with the enemy" - Moltke. What is important is to start networking now so you have people to ask about jobs after you finish. Find out what fora your professors and fellow students use for this. For reasons I won't go into, there's a widening schism between what academia does and what industry does. -- Les Cargill
Reply by ●January 8, 20122012-01-08
On Jan 8, 3:18�am, Richard Dobson <richarddob...@blueyonder.co.uk> wrote:> On 08/01/2012 08:27, Eric Proffitt wrote:> Seems to me you are looking towards what used to be called "mathematical > science" and maybe still is in places, but is more usually called > Computer Science. The people in academia I have got to know and > occasionally work with think of themselves as mathematicians (one has a > background in astrophysics), but are what would be called computer > scientists these days. With the prevailing mood of "technology transfer" > directed from above, chances are you would have as many opportunities to > work with industry as you would want, while being based in an academic > research department. Not so much writing reports as research papers and > funding applications. Plenty of scope for working on dsp, for example in > HPC fields (massively parallel computation, etc). And if physics is a > high interest: they probably still need people with heavy math skills > plus computing plus EE plus physics and dsp to work on stuff at CERN, on > the LHC Grid, or on the proposed follow-ups to the LHC. Huge data > analysis and simulation tasks, among other things - in effect dsp > applied to the largest digital camera on the planet. > > The alternative would be to be head-hunted by an investment bank to work > on algorithms for option pricing et al (and, inevitably, applying the > FFT to stock price movements) - people still have faith that could > actually work, despite all the evidence to the contrary. The rule for > getting into dsp seems to be, if you like the application area. So, you > do audio dsp because you like audio, or comms dsp because you like > comms. But not so much because you like math! > > Richard DobsonI guess I should clarify, if I get a job in academia it would not still be necessary to work with industry, although I wouldn't necessarily be opposed to it. There is actually another masters program farther away that offers a masters in computational linguistics, I wouldn't be entirely prepared for such a program with my background primarily in math, but natural language processing and comp. linguistics seem like pretty cool fields. Do you know anything about those areas of computer science and what its like to work on say google translate and the like?
Reply by ●January 8, 20122012-01-08
On 8 Jan, 21:09, Eric Proffitt <espr...@gmail.com> wrote:> �There is actually another masters > program farther away that offers a masters in computational > linguistics, I wouldn't be entirely prepared for such a program with > my background primarily in math, but natural language processing and > comp. linguistics seem like pretty cool fields. �Do you know anything > about those areas of computer science and what its like to work on say > google translate and the like?That's an about turn! - Lingusitics has little if anything to do with maths - Computational linguistics has a *long* way to go before it comes anywhere near close to 'applied'. Rune
Reply by ●January 8, 20122012-01-08
On Sun, 08 Jan 2012 01:39:26 -0800, Rune Allnor wrote:> On 8 Jan, 09:27, Eric Proffitt <espr...@gmail.com> wrote: >> I fear this may not be the appropriate forum for this post, but I don't >> know where else to go and since every bit of info I can get is critical >> to making this very important life decision, I'm going to anyways. >> >> Very soon I will have to decide whether to continue on with pure math >> with the intent of getting a phd, or to instead opt for an applied math >> masters with a concentration in electrical engineering and dsp. > > Are those the two options? > >> I really like math, and I think I would enjoy applying mathematical >> analysis to solve real world problems a là signal processing. > > Should somebody who seriously considers a PhD in maths go into DSP... > > That's like a contender for a F1 driver position seriously considering > becoming a bus driver: Both make their living by driving motor vehicles. > > Yes, there is some maths involved in DSP. No, it's not at all neither > challenging, interesting or even fun. If you get into DSP you will be at > the mercy of dillettantes who consider it a skill to know how to play a > CD on a home stereo system, and amateurs who do know how to use a mobile > phone: Both systems rely on DSP, and being a user of such systems is > widely considered enough competence to get jobs in managment of DSP'ish > projects. > > Get into DSP if you ar a masochist with Kafkaish leanings; otherwise, > stay far away. > > As for the PhD in maths - that's up to you to decide: Spend 3-5 years to > obtain a diploma to hang up on the wall, or spend that time getting a > position where you actually will be able to make a living? > > Then there is the 3rd option: If you *really* like maths, and you > *really* like applied work, consider state space methods for time series > analysis. The hard work has already been done (check out the writings by > Durbin and Koopman); lots of work remains on using these kinds of > techniques to solve real problems. > > I know that's the academic line I would attempt to pursue, if I were > able to rewind the calendar by the necessary 25 or so years. But then, > if I were able to rewind like that, I might far more likely drop > academia alltogether... > >> I have >> done quite a bit of lurking on this forum and it seems like some of you >> know quite a bit of advanced math in the areas of complex and fourier >> analysis, and even branching out into analytic number theory from time >> to time (way cool). > > Fourier analysis is trivial. The reason it generates heated debate here, > is that the majority of people here - me included - have little or no > formal training in math. > > Don't let *our* math discussions decieve you into pursuing a PhD > degree!! > > Number theory - yes, it's used in coding theory. But again, it seems (I > don't have the detailed knowledge to really comment) to be plain vanilla > engineering maths, that was established centuries ago. > >> But I generally can't figure out which >> posts come from people in industry, and which come from people in >> academia. > > Doesn't matter. Pursue DSP at most as an hobby. > >> I like physics and engineering as well and I think I would enjoy the >> other aspects of dsp engineering: programming, writing reports, working >> with others, troubleshooting etc. But I also like to be intellectual >> challenged and creatively apply math, so is this the right job for me? > > No. > >> Am I deluding myself about the life of a dsp >> engineer? > > Yes. > >> Would someone with such a math heavy degree even be >> considered of value in this industry? > > Eh... having the PhD degree in DSP with more math than was usual for DSP > degrees at the time (but nowhere near what a maths degree), and having > had a number of 'token' jobs (managment with little or no DSP competence > proudly pointing at me and proclaiming "We have a PhD on staff!" but > where I have no real saying in what to do or how), I don't really know > what to answer: > > - Yes, I can see that people like you would be of > significant value if assigned the right tasks and responsibilites. > - The chances of somebody like you ever actually *being* > assigned those those tasks and responsibilities are less than for you > winning the lottery [*]. > > I will not pretend to be able to tell you what to do. But I can > confidently tell you what *not* to do: Stay away from a carreer in DSP. > Pursue DSP if you want, but don't let it become more than a hobby. > > Rune > > [*] In the lottery, somebody actually wins.Rune, I'm not sure that what you say reflects working in Norway, or if you're just getting bitter as you age. But while there's a wide streak of that here in the US, it ain't that way through and through. Big corporations can have departments with that attitude, but smaller companies can't afford it. When you _get_ a smaller company with that attitude they tend to run out of money and die; when you get a big corporation department with that attitude, well, they can only hide for so long before they get audited and chopped off as deadwood. Granted, some times the "so long" is decades, but eventually the saw appears and corporate pruning happens. If you get hooked up with the right group, there certainly are opportunities to couple up your mental horsepower with the product-design road, and create some real and lasting work. If I thought it was practical, I'd suggest you get a green card and come over here for a while. If you seriously look you can probably get something on an H1-B1 (AKA "indentured servitude for naive furreners"), and maybe even get a green card, but you my understanding is that you have to be pretty determined. (If you really want, let me know and I'll hook you up with an engineer I know. She's Romanian-born and younger than the two of us combined, both of which make a difference, but she's here and she's working). -- My liberal friends think I'm a conservative kook. My conservative friends think I'm a liberal kook. Why am I not happy that they have found common ground? Tim Wescott, Communications, Control, Circuits & Software http://www.wescottdesign.com
Reply by ●January 8, 20122012-01-08
On Sun, 08 Jan 2012 00:27:31 -0800, Eric Proffitt wrote:> I fear this may not be the appropriate forum for this post, but I don't > know where else to go and since every bit of info I can get is critical > to making this very important life decision, I'm going to anyways. > > Very soon I will have to decide whether to continue on with pure math > with the intent of getting a phd, or to instead opt for an applied math > masters with a concentration in electrical engineering and dsp. I really > like math, and I think I would enjoy applying mathematical analysis to > solve real world problems a là signal processing. I have done quite a > bit of lurking on this forum and it seems like some of you know quite a > bit of advanced math in the areas of complex and fourier analysis, and > even branching out into analytic number theory from time to time (way > cool). But I generally can't figure out which posts come from people in > industry, and which come from people in academia. > > I like physics and engineering as well and I think I would enjoy the > other aspects of dsp engineering: programming, writing reports, working > with others, troubleshooting etc. But I also like to be intellectual > challenged and creatively apply math, so is this the right job for me? > Am I deluding myself about the life of a dsp engineer? Would someone > with such a math heavy degree even be considered of value in this > industry? Thanks for your time.Having an unconventional degree like that will make your first few jobs harder to get, but if you can show that you're useful it won't matter after five years or so. If you're really career focused I'd suggest that instead of getting your higher degree in applied math with a concentration in EE, that you look into what it'll take if you bite the bullet and just get a master's (or PhD) in EE with an emphasis on DSP. It'll probably mean some remedial classes in EE, but depending on your school it may be less than you think. That "real EE degree" will give you a lot more 'street cred' when it comes time to find a job. Keep in mind that if you're going to get a job somewhere doing _just_ PhD- level DSP work, it'll either be as a principal scientist at some startup that does something really startling with DSP (and has the associated huge risks and huge rewards), or it'll be as a member of the technology R&D staff at some HUGE company that does really high-tech stuff (i.e., Qualcomm or similar, developing the coding for the G5 cell phone stuff). You can probably do either of those with a PhD in math, particularly if you can choose an area of research from which some useful technology can be developed. (In fact, if you think your balls are really big, and you focus on the right research topic, you and your work can be the core of a startup -- it depends on how good you are at making contacts with, and impressing, the money people, and how quickly you can make your work look like it's actually going to generate some $$ on the open market). -- My liberal friends think I'm a conservative kook. My conservative friends think I'm a liberal kook. Why am I not happy that they have found common ground? Tim Wescott, Communications, Control, Circuits & Software http://www.wescottdesign.com
Reply by ●January 8, 20122012-01-08
You'll find mathematicians in quite many jobs that don't have a lot to do with math in the first place, but where a "certified bright person" is required. The challenge I'm seeing now more than ever is that just about anything is - forgive me the overused buzzword - multidisciplinary. As a Ph.D. you'll most likely become an expert in a rather narrow field, and maybe it's quite unlikely you'll use any of it in post-academic work. From an academic's point of view, most of the industrial work must be trivially simple. Generally speaking, industry doesn't like "hard" problems - it implies high risk, and risk needs to be managed very carefully. What I mean is for example, I can't afford to get stuck for three weeks on finding an elegant, closed-form solution that may or may not exist. Instead, we've got lots of CPUs in the basement - brute force isn't a particularly stylish tool, but it gets the job done reliably.
Reply by ●January 8, 20122012-01-08
On Sun, 08 Jan 2012 14:54:14 -0600, mnentwig wrote:> You'll find mathematicians in quite many jobs that don't have a lot to > do with math in the first place, but where a "certified bright person" > is required. > > The challenge I'm seeing now more than ever is that just about anything > is - forgive me the overused buzzword - multidisciplinary. As a Ph.D. > you'll most likely become an expert in a rather narrow field, and maybe > it's quite unlikely you'll use any of it in post-academic work.This is partially true -- many of the guys that I know with PhD's are basically hired because they're obviously smart enough to get the job done, but obviously dumb enough to take it on in the first place. Sort of like the criteria for recruiting a kid to the Marines or Special Forces, but involving math instead of guns. On the other hand, I can count at least three people (one was a prof when I was in school, one was a co-worker, another was a vendor and second- degree acquaintance through an old prof) who consistently make (or made) money in the area that was core to their area of research when they were doing their PhD, and quite a few more with engineering PhD's who are working in closely related areas (i.e. a mechanical engineering PhD working on mechanical problems, an optics PhD working on optical systems, a control systems PhD who works on inertial navigation systems, etc.).> From an academic's point of view, most of the industrial work must be > trivially simple. Generally speaking, industry doesn't like "hard" > problems - it implies high risk, and risk needs to be managed very > carefully. What I mean is for example, I can't afford to get stuck for > three weeks on finding an elegant, closed-form solution that may or may > not exist. Instead, we've got lots of CPUs in the basement - brute force > isn't a particularly stylish tool, but it gets the job done reliably.Conceptually simple, perhaps, but not trivially so. And there's fewer and fewer problems that can be solved with closed-form solutions these days -- a lot of math, and even more engineering, is getting done "experimentally". -- My liberal friends think I'm a conservative kook. My conservative friends think I'm a liberal kook. Why am I not happy that they have found common ground? Tim Wescott, Communications, Control, Circuits & Software http://www.wescottdesign.com
