"Prestige" of University where gained PhD

[thomasjohnson]

Hi

I have just graduated with a first class masters degree in mathematics from Imperial College London, and would like to study for a PhD in General Relativity. However, I also want to have the preparations in place to become a quant, if I would like to get out of research for some reason. My question is this; will people looking to hire quants with PhD's be too bothered about the name of the university you gained it from? I only ask because the university with active research in the area I would like to go into are not as recognized as Oxford or Cambridge. In mathematical circles the quality of your thesis is more important than the name, but only a select group of people will know what the paper is on about!

I appreciate any help on this matter, Tom

 

[Daniel Duffy]

GR???

Not exactly computational or am I missing something?

 

[thomasjohnson]

if you choose correctly, there is a lot of numerical methods involved eg. modelling gravitational waves emitted from binary stars. However, I was lead to believe that having computation as part of your PhD wasn't absolutely necessary as you can learn this in your own time?

 

[Daniel Duffy]

if you choose correctly, there is a lot of numerical methods involved eg. modelling gravitational waves emitted from binary stars. However, I was lead to believe that having computation as part of your PhD wasn't absolutely necessary as you can learn this in your own time?

IMO
This is a huge roundabout way to learn numerics. And GR is no use outside shooting stars.

20th century mathematics is a bit unimaginative. The real stuff took place in the 18th and 19th centuries.

You can't learn computation in 'spare time'.

My 2 cents.

On the other hand I believe Phelim Boyle has a PhD in GR from TCD Dublin.

 

[thomasjohnson]

thanks for your response, but I'm not going into GR because I want to learn numerics. Also I didnt define spare time very well, I essentially meant "not part of my PhD". I want to do a GR PhD because I think it is interesting, not to become a quant. I just want to make sure I am prepared.

before I reply to your other comment, what is your knowledge and background in mathematics?

also, GR is used in sat nav, if you want one real world use outside "shooting stars"

 

[Barny]

Here's the truth - you are obscenely unlikely to get a faculty job in GR or theoretical physics.

Therefore, your options are to not do a PhD, or to do a PhD for fun and get another job at the end of it.

Assuming you really want to do the latter, and enjoying your research and your life for 3-4 years is a priority, you really want to find the best supervisor you can - regardless of where he is based. If on the other hand, you want to maximise your chances of getting a job outside of academia, then you need to do a PhD at the biggest brand name school. The trade off is as simple as that. Choose what you want more - fun or prestige. You can be really lucky and have fun at a prestigious uni, but that is highly unlikely. And given that harsh fact, I would rather not do a PhD than suffer through a PhD at a prestigious university with a supervisor and research topic I hated.

So if I were your case and doing a PhD as a hobby (since you are not going to get a faculty job at the end), then I would forget about brand-names and actually enjoy my hobby.

So to reiterate - you need a good supervisor at a good uni, unless you treat it as a pure hobby, in which case a good supervisor is all you need. Otherwise, don't do a PhD.

 

[Daniel Duffy]

thanks for your response, but I'm not going into GR because I want to learn numerics. Also I didnt define spare time very well, I essentially meant "not part of my PhD". I want to do a GR PhD because I think it is interesting, not to become a quant. I just want to make sure I am prepared.

before I reply to your other comment, what is your knowledge and background in mathematics?

also, GR is used in sat nav, if you want one real world use outside "shooting stars"

I have B.A. (Mod), MSc and PhD all in pure and applied/numerical mathematics. (Trinity College, Dublin)

I want to do a GR PhD because I think it is interesting, not to become a quant.
Do you want to become a 'quant'? BTW quants spend > 70% of their time programming, yes?

I know nothing about 'sat nav' but a guess is that a lot of it is C/assembler programming Kepler's equations etc. Is Newton/Lagrange/Hamilton good enough?

Industry and academia are worlds apart.

 

[thomasjohnson]

Barney - thanks, this is kind of what I have been thinking. It's a "shame" that my favorite subject is one in which the bubble has burst recently, but I knew if I did any other topic/job I'd constantly be regretting not at least getting it out of my system. May I ask if you have been through something similar like me, and if so, is this in the UK or US? Also, do you know if having spent 4 years getting a PhD will be detrimental to go for jobs that only require a bachelors, e.g. actuary, or will it have no influence?

Daniel - kudos on the PhD. I can't comment on the world of numerics, but I know that for example the world of geometry has gone pretty insane in the past 20 years (e.g. Donaldson at my uni works on 4 manifolds) and the proofs of the positive mass theorem and penrose inequality in GR.

Also from what I have seen here in the UK the roles of quants are separated into developers (more programming) and mathematics (testing/coming up with the models and strategy's) which still involves coding but not as heavy. but maybe I have been mislead? Further, my uni does an MSc in math finance, and I believe its seen as one of the best in the UK that only contains two half course on coding in C++, the rest is spent of theory i.e stochastic analysis, portfolio theory etc., and as part of it you spent your summer doing a project switching between uni and a hedge fund or wherever they sort you out, which is different to alot of the MFE's I have heard about.

and you are correct that Newtons laws are used to approximate most space systems, but one thing Newtons fails to predict completely is the notion of time dilation, which is quite important for sat nav as the satellites in orbit above the earth experience time slower that us on earth. I think its only something like 34 micro seconds but it actually has quite an effect when determining the position.

 

[Daniel Duffy]

Also from what I have seen here in the UK the roles of quants are separated into developers (more programming) and mathematics (testing/coming up with the models and strategy's) which still involves coding but not as heavy. but maybe I have been mislead?

Probably the distinction is not so large these days.

Further, my uni does an MSc in math finance, and I believe its seen as one of the best in the UK that only contains two half course on coding in C++,

At least they _d0_ do C++. Some don't at all.

 

[Barny]

thomas - there is no distinction anymore. There is no such thing as a pure quant, who sits around building models. If they exist, they are rarer than faculty positions in GR, so you shouldn't aim to do that either.

Doing an MFE is generally not a good idea (in the UK). Your undergraduate degree is good enough to get into any banks grad programme as it is. If you're looking for quant specific skills, then an MFE won't cut it. You need to become a good programmer. Spending 4 years of your PhD writing Python and C on a unix system is a good start, but I don't know how much of that you'll get to do in a GR PhD which will likely require many hours of sitting down with pen + paper.

I will reiterate that doing a PhD at a prestigious university with a bad supervisor is pretty much your worst possible outcome. All other combinations are OK (except for bad-bad, of course).

 

[bigbadwolf]

Also, do you know if having spent 4 years getting a PhD will be detrimental to go for jobs that only require a bachelors, e.g. actuary, or will it have no influence?

Neither beneficial nor detrimental -- just irrelevant. An acquaintance of mine completed a B.A. and part III in math from Cambridge, then completed a PhD in geometric topology from Liverpool. He just never cited his PhD when working in the City.

Daniel - kudos on the PhD. I can't comment on the world of numerics, but I know that for example the world of geometry has gone pretty insane in the past 20 years (e.g. Donaldson at my uni works on 4 manifolds) and the proofs of the positive mass theorem and penrose inequality in GR.

The world of Donaldson and Witten is a world apart from that of quant and coding. But the ferocity and calibre of competition is such that it's virtually impossible to make a niche for oneself in that rarefied world unless one has monumental talent. The skill set for quant is not so exacting but it's different and still requires time to develop.

Also from what I have seen here in the UK the roles of quants are separated into developers (more programming) and mathematics (testing/coming up with the models and strategy's) which still involves coding but not as heavy. but maybe I have been misled? Further, my uni does an MSc in math finance, and I believe its seen as one of the best in the UK that only contains two half course on coding in C++, the rest is spent of theory i.e stochastic analysis, portfolio theory etc., and as part of it you spent your summer doing a project switching between uni and a hedge fund or wherever they sort you out, which is different to a lot of the MFE's I have heard about.

Not enough for real quant jobs. Those theory-heavy jobs have mostly evaporated and today's quant is a glorified coder, a coder+, where the "+" is the financial, mathematical and statistical acumen he brings -- but the base is still solid coding skills.

 

[Daniel Duffy]

The world of Donaldson and Witten is a world apart from that of quant and coding. But the ferocity and calibre of competition is such that it's virtually impossible to make a niche for oneself in that rarefied world unless one has monumental talent. The skill set for quant is not so exacting but it's different and still requires time to develop.

IMO
In academia you have a solution(can be the pet field/topic in the department) for all problens. Quants also have problems but any solution if good if it is on time and it works.

 

[Daniel Duffy]

Also from what I have seen here in the UK the roles of quants are separated into developers (more programming) and mathematics (testing/coming up with the models and strategy's) which still involves coding but not as heavy. but maybe I have been mislead? Further, my uni does an MSc in math finance, and I believe its seen as one of the best in the UK that only contains two half course on coding in C++, the rest is spent of theory i.e stochastic analysis, portfolio theory etc., and as part of it you spent your summer doing a project switching between uni and a hedge fund or wherever they sort you out, which is different to alot of the MFE's I have heard about.

In general it takes [5, 10] years to become a good programmer (and maths background is neither necessary nor sufficient to become one).

 

[thomasjohnson]

yeah geometry is insane. My supervisor for my masters was supervised for his PhD by Richard Schoen, and you could just tell he operated on another level. nice guy though. That's why I thought GR was such a beautiful theory though, the blend of geometry and physics was wonderful.

Okay thank you all three for the advice. I've done coding in C, and I enjoyed it, but I am much more of a theory man. So do banks contract out "blue skies research" or something? For example, I was speaking to a guy from G Research, and coding wasn't the be all and end all for them (they were very interested in my friend who is doing a stochastic analysis PhD, with no programming). Would companies like that be more of a theory role, or is it literally completely dead? To that end, is a math finance PhD no more useful to be a quant, unless it contains large amounts of coding? (Just out of curiousity).

It's a shame that I have missed the mathematical quant bubble, but I take heart from what you have said. It seems to me that supposing I do a PhD in the area I like (black holes) at worst I come out 3 years later in just a bit better position (I (hopefully) would have matured a bit more and done more networking!) with my mathematical passion somewhat more satiated, and if I am not qualified enough\don't want to be quant I can just go down other routes that I would have done anyway. Cons would be 3 years of wasted earning potential but I'm not fussed with that. My original concern with that though was that companies might not like the fact I'm 26 before I start an actual job but this is unfounded according to BBW am I right in thinking?

thanks again for the comments. It's been quite reassuring if the above is correct as I feel somewhat lost atm.

 

[thomasjohnson]

Your undergraduate degree is good enough to get into any banks grad programme as it is

would be if I didn't keep failing all those stupid nugatory tests they make you do!

 

[Barny]

yeah geometry is insane. My supervisor for my masters was supervised for his PhD by Richard Schoen, and you could just tell he operated on another level. nice guy though. That's why I thought GR was such a beautiful theory though, the blend of geometry and physics was wonderful.

Okay thank you all three for the advice. I've done coding in C, and I enjoyed it, but I am much more of a theory man. So do banks contract out "blue skies research" or something? For example, I was speaking to a guy from G Research, and coding wasn't the be all and end all for them (they were very interested in my friend who is doing a stochastic analysis PhD, with no programming). Would companies like that be more of a theory role, or is it literally completely dead? To that end, is a math finance PhD no more useful to be a quant, unless it contains large amounts of coding? (Just out of curiousity).

It's a shame that I have missed the mathematical quant bubble, but I take heart from what you have said. It seems to me that supposing I do a PhD in the area I like (black holes) at worst I come out 3 years later in just a bit better position (I (hopefully) would have matured a bit more and done more networking!) with my mathematical passion somewhat more satiated, and if I am not qualified enough\don't want to be quant I can just go down other routes that I would have done anyway. Cons would be 3 years of wasted earning potential but I'm not fussed with that. My original concern with that though was that companies might not like the fact I'm 26 before I start an actual job but this is unfounded according to BBW am I right in thinking?

thanks again for the comments. It's been quite reassuring if the above is correct as I feel somewhat lost atm.

Blue skies research happens in academia. If real quants who work in banks like something that academics do (very rare), then they might try and implement the model. If you have a PhD in stochastic analysis, you might have a chance doing something more theoretical, for someone. The focus on the real world is on practicality. It is simply not worth spending a long time sitting down with pen+paper when your models are already doing the job. Don't forget that your job in a bank is to make money, not to enjoy yourself by playing around with equations.

The hay-day of theoretical quant work was in the 80's. It's the same for every area in which quantitative techniques first get hold - in fact, the best advice I can give to anyone in quantitative fields is to go into a NEW area. That is why high-energy physics and theoretical physics are such appalling PhD choices - they have been done to death, the only people publishing important papers are professors with 30+ years of experience. A new PhD student can make hardly any impact.

26 is very young in the quant field. If you do a PhD in Germany or France or Italy you graduate when you are 30. Most quants in the UK are not from the UK.

I also fear you are underestimating the challenge of becoming a good programmer. Writing a toy program in C is one thing, writing, reading and maintaining a code based of 50k lines with hundreds of classes, functions, header files, external libraries and packages and what not is another beast altogether. That is the kind of skill you need.

 

[Barny]

would be if I didn't keep failing all those stupid nugatory tests they make you do!

You have to learn to jump through the stupid hoops even with a PhD. Better to start sooner rather than later.

 

[Barny]

In general it takes [5, 10] years to become a good programmer (and maths background is neither necessary nor sufficient to become one).

I know people who have been programming for 5 years are still awful. Do not confuse experience with expertise.

 

[Daniel Duffy]

I know people who have been programming for 5 years are still awful. Do not confuse experience with expertise.[

Then you are not screening properly. After 5 years you should be reasonably competent.

It's probably related to the Pareto 20/80 rule.

 

[Barny]

Then you are not screening properly. After 5 years you should be reasonably competent.

It's probably related to the Pareto 20/80 rule.

I'm not screening anybody, just observing. Competency and knowing some syntax are two different things. There is no guarantee that anyone is competent in something just because they've spent time doing it. To become competent you need to push yourself, extend your boundaries, incorporate new ideas and techniques into your coding. Many people stop learning once they know how to write a function.