Search This Blog

Tuesday, May 12, 2015

The quality of faculty

 I have been to numerous selection committees, and I am quite aware of the quality of faculty selection in our tier 2 institutions. As soon as you go beyond IITs, BITS, IIITH, IIITD, and a handful of other institutions, invariably the selection committee meeting ends with the Director making a request to the committee to recommend at least one candidate, any one who was marginally better than the others, so that the next semester's courses can be taught. In any of these committees, I always look at my role as providing my honest views about the candidate, and making sure that there are no biased selections. I don't look at my role as quality control (beyond giving my honest feedback). The primary stake holder in that hiring is the institute and not me, and hence they have to do quality control and not me. So, invariably, I will agree to recommend that one candidate, who could not speak a word about his thesis that he had just submitted, and who could not answer simple questions about the course that he had recently taught, and yet in my opinion, was the best amongst all those who were interviewed. Participating in such committee meetings can be very depressing as they expose this myth about India being the largest producer of scientific manpower. These candidates can hardly be called "scientific manpower" and yet they represent the best that our scientific manpower production factories have produced. What are we doing to our next generation, forcing them to study from these types of faculty members. And this is one reason why I always ask people to seriously consider studying abroad, if they can afford it.

A few weeks ago, I was in a selection committee to recruit faculty for a new institution. The new institution (which shall remain nameless) has this agenda of excellence that they would not want to recruit just about anyone. So they had shortlisted a set of people whose CVs looked pretty good, better than the most candidates that I have met in tier two institutions. Most of them were existing faculty members in our Tier II institutes, which included IIITs, NITs, IIMs, and a few select state colleges and universities with excellent reputation. (IIMs are certainly Tier 1 institutions, but only when it comes to management related programs.) If you were to select faculty purely on the basis of their CV and pedigree, it would be extremely tough to take a call. And I was really excited about a new place being able to attract such CVs.

All the candidates were told to bring in prints of a few papers of theirs, preferably those papers which were recent and in which they had a significant contribution. One member of the committee had a very sound strategy for the interview. He would ask the candidate which paper is the strongest work. He would then ask a simple question about something in that paper, and the result was often hilarious (actually, very sad, given that these represented the second tier of institutions in the country).

The hottest area of research amongst those was "optimization." Can you explain what is a generic problem of optimization (not in your specific context). Sorry, I don't understand the question. OK, you have this equation in your paper. Can you give us an insight into this equation. Sorry, no. I just copied this equation from Matlab help files. Do you think it is ok to copy from Matlab help files without understanding. What's wrong, everyone does it. But then this appears to be a central equation in your paper. Why are you using this one and not anything else. I don't remember. But this is your recent paper, with you as the first author. I am the first author, because I am the supervisor. How do you avoid convergence to a local maxima, and guarantee that you will find the global maxima. I don't understand the question. Do you know what a local maxima is. Yes, the points that are close to global maxima. Thank you very much, you can leave now.

There were a few senior professors as well amongst the candidates. There was a faculty member who said that he is very strong in Operating Systems. When we asked to narrow down the area further, he said memory management. OK. Why do we need virtual memory. Virtual memory is needed to allow a program of larger size than the physical memory of the computer. Really. But it means that if the physical memory in a computer is larger than the addressable memory, then we don't need virtual memory. Yes, we don't need virtual memory. If we can somehow enforce that all programs will be smaller than the RAM in the computer, we don't need virtual memory. Yes, we don't need virtual memory in such a situation. How would you ensure that two programs who are using the same address space will not conflict with each other's memory. Sir, I have been teaching OS every year for the last 20 years. My students have got jobs in top CS companies, and you are telling me that I have been teaching wrongly all these years. Yes, my dear professor, you have been teaching wrongly for all these 20 years.

I suspected this, but did not want to believe this. In the last 20 years, I would have asked this question on virtual memory to at least 100 potential MTech/PhD students, and till date, NOT EVEN ONE (other than those who have studied at an IIT) has answered it correctly. The sequence of exchange that I had with this senior professor of operating systems in one of the good institutes of the country, was exactly the same I have had with students from such colleges. And yet, I always believed that they were taught correctly, but they didn't pay attention. Or they are learning from poor quality text=books. I could not believe that they learnt wrongly because their professors did not know.

I can go on and on, but the summary is that most of these faculty members with great CVs did not know even the basics of what they were teaching, or what they were doing research in. They did not know that copying content from elsewhere was plagiarism. They did not know that they should understand what they write in a paper. It was almost as if the papers were generated with some automatic paper generating software.

And what is worst is that when we investigated further with these faculty members who all had substantial number of journal papers, it turned out that many of those papers were in paid journals with no peer review. And yet, they would defend the practice by saying that everyone else did it too.

Since most of these faculty members were young and recent PhDs from similar institutes (IIITs, NITs, and good state colleges), it also puts a question mark on the kind of PhDs being produced in the country.

And yet in a country where the quality is determined primarily by the amount of money their graduates can get in the market, these institutions are rated very high by the society. So there is no hope of any improvement either. If they are all doing very well, where is the pressure to change. The only positive of these institutions is that they have successfully created a culture of self-learning (necessity is the mother of invention), which will ensure that their alumni keep learning on their own throughout their careers.

But is this enough for the "Make in India" to succeed. Are we really on our way to harness our demographic dividend.


Varun Aggarwala said...

Dear Prof. Sanghi,
Thank you for writing this article. I still cannot believe that the "senior professor" whose expertise was in virtual memory could not answer that basic question.
I was wondering if you really think that the faculty at "tier 1" institutes like IIX etc. is also upto the mark. Of course most of them are brilliant and can hold their own in any scientific conversation, but is their scientific contribution comparable to the other top 100 universities in the world. I am not talking about the number of papers but I am specifically focusing on atleast one major discovery or impactful piece of work.
Do you think that a significant majority of assistant or full professors at IIX have made an impact in their field of research? Also you raise a brilliant point about most of the candidates who happen to be recent PhDs from these institutes. Is there no quality control mechanism at these places? How do they award PhDs if the student does not show enough evidence of independent scientific productivity and can hold their own in thesis defense?
PS: I really love reading your hard hitting articles. Please do not consider the lack of comments as lack of interest :)

Dheeraj Sanghi said...

@Varun, I don't think the quality of faculty at IIXs is comparable to the quality of faculty in top 100 universities in the world. There are brilliant individuals, no doubt, but as a whole, there is a problem. However, let me also add that if good people are not getting attracted to Tier 1 and Tier 2 places in the country, or good people after joining lose focus and stop being productive, one has to seriously look at regulatory framework for the higher education in India as well as other issues such as sufficient investments in facilities, administrative leadership, and so on. Most of the Directors who request that at least one faculty be hired to teach courses would have done absolutely nothing to attract good applications.

Varun Aggarwala said...

Prof. Sanghi,
Thanks for the reply.
I do hope that one day "ache din" or even "ache sitare" will come for the Indian academia!
PS: I have made a transition from engineering to

S said...

@Varun Aggarwala: You ask "Do you think that a significant majority of assistant or full professors at IIX have made an impact in their field of research?"

I am not sure about all disciplines, but certainly in specific areas, such as Theoretical Computer Science, at least a couple of the IIXs would easily rank among the top 20-30 places in the world using any reasonable metric of "impact".

Disclaimer: I am not currently affiliated with any of the IIXs.

iFunda said...

One important subject that I feel doesn't get enough attention in CS schools is Debugging. There is a wealth of knowledge one can discover if one just learns how to debug user mode and kernel mode code.

Multi-threading/Synchronization, Security, I/O, Memory management-Kernel address space vs user address space, Optimization, Hooking, Exploits, Context switch, Registers, 32-bit vs 64-bit, ... One who knows how to debug knows how to learn.

I would love to see IITs introduce full credit Debugging course. Windows/Unix & C/C++. Not only will the students learn a ton but the teachers will also rectify their short comings when they see things in action. There is no way they can bluff their way to glory.

Great article, Sir! Please keep up the good work.

Varun Aggarwala said...

@iFunda: Do you seriously think that IITs should become trade schools and offer debugging classes? While learning a trade and being good at it is of utmost importance, I think both the faculty and the students will be better off learning about the concepts of programming (typing, functional programming etc.) with real applications, rather than just hacking their way to "glory".

Do not get me wrong. A good assignment and a project teaches/clarifies a concept via real experience. However I think great minds at IIT should not just focus on getting real experience without broadening their horizons.

Also I forgot to complete the last sentence in my previous comment. I just wanted to highlight that I have made a transition from engineering to biomedical sciences and I think that India is attracting a lot of excellent critical mass in that area; and I wish that the same happened in other fields too.

Varun Aggarwala said...
This comment has been removed by the author.
jayanti said...

Thank you sir for writing this nice piece. Being a candidate for faculty positions in some of these 'Tier 2' institutes I completely agree with what you have written. In one of these institutions I was asked for showing my post-doc certificate ! When I said I do not have any but can produce reference letters of professors (who are well know figures) with whom I worked during my post-doc if I am given 24 hours they were quite upset and put the remark 'no research experience' on my form (I have more than 5 year post doc experience). Next thing they asked me was whether I have written any 'lab manual' I said no so they put another negative remarks on my form. There was no extra weight given to candidates with prestigious awards like DST-INSPIRE and DST-FAST fast track. When I faced the interview committee the first thing they asked me was to show the hard copies of my Ph.D thesis and research papers which I did not have. Committee was quite disappointed and told me then what basis they should evaluate me since I have nothing ! I was really shocked. How much difference it would had made if I would had produced hard copies ? Is it possible to read and judge Ph.D thesis and research papers in 10 minute by just looking at those ? Why in 2015 some of these 'Tier 2' institutions still do not have Internet connections (?) or independence means of judging the quality of research of a candidate ? Anyway after the interview of 'external' candidates was over 'internal' candidates were interviewed and most of who were carrying a bundle of 3-4 kg papers with them ! During my interview one member asked me if photon is massless then how it gets deflected by mass in gravitational lensing ? My answer was that in Einstein's theory of General relativity every massive objects 'curves' space-time around it and every other object in its surrounding can 'feels' the space-time curvature irrespective to what mass it has. I think the members had not background in this area so he kept repeating his question. My point is that why some of these Tier 2 institutions cannot shortlist candidate based on their research areas and invite experts accordingly.

Dheeraj Sanghi said...

@Jayanti, I am surprised by your experience. The faculty shortage is so huge (at least in engineering. may be not so much in your discipline) that institutes can not afford to be so rigid and unfriendly. At least in places that I have been, it is common for applicants to send links to online papers in advance, and I have never heard any candidate not being given credit for experience just because there is no written document. May be, only some type of institutes call me for selection committee, others don't for the "fear" of my putting the conditions on this blog (even if the institutes and individuals remain nameless).

Sanjay Goel said...

Thanks Prof. Sanghi for this very interesting and as usual bold article. You are absolutely right. I too feel that "Make in India" will reduce to "Manufacture in India" and not get converted into "Conceive, design and develop in India" unless something significantly changes on this core front.

Your article reminds me of one of my own article written few years ago on the same theme -

However, it is also interesting to recall that usually the PhDs in many of the so called second tier Indian institutes have their PhD examiners coming from IIT. It will make a very useful study as to what fraction of the PhD theses at these institutes, are returned by the examiner (usually from an IIT) for significant improvement. My feeling is that this fraction it is very small.

As long as PhD education remains a part time 3-4 year endeavour for working faculty (often with many family responsibilities too), there is no hope for any significant improvement. But surely the voices like yours must continue to alert the society.

Finally, I hope that you also agree not all IITs are doing an excellent job wrt PhD quality in all their departments and IIT tag is also being misused by some.

-Sanjay Goel

Dheeraj Sanghi said...

Dear Sanjay, Nice to hear from you. I fully agree with your points. Even in IITs, not every PhD is good. I think the reason for IIT faculty not rejecting PhD theses of other places is multifold. I don't want to determine your quality. I will write a report which points out many holes in the thesis, but I would not write that the thesis be rejected. If, after reading my report, you still go ahead and award the PhD without any significant changes to the thesis, I don't feel responsible for it. Indeed, there have been times when I have written very hard hitting reviews, including in some cases pointing out that the thesis is plagiarized (and at least in those cases recommending that the thesis be rejected), and yet I see later on the PhD being awarded. So I have stopped accepting PhD theses from Tier 2 institutions unless I know the supervisor and I know that he will care for my comments. In most cases, PhD review is an organized fraud where we will suggest names of our friends who would do a quick review and always send a wishy washy report.

Imtiaz Mohammad said...

@Varun Attaching a debugger to validate and/or reverse engineer the OS, networking, compilers, data structures & programming language concepts is NOT a trade. It is priceless :-)

Fine. Let's call it a trade. We do labs/projects in almost every course at IITs. You can call it a Debugging/Reverse engineering lab.

shoda M said...

Dear Prof. Sanghi
Thanks a lot for writing your bold experiences. I fully agree with you. It is not easy to change the Indian academic scenario unless we think it ourselves repeatedly. I request you to write some other experiences in future also.

मुकुंद माधव मिश्र said...

Very nice article Prof Sanghi. I was wondering where the actual problem lies. Being too young for this task, I have never been to a selection committee but I interview Undergraduate students at different college on regular basis. No wonder that the performance of the students is very much similar to what you have experienced in selection committees. They have written a computer program in Mathematica but they do not understand the commands. A student is writing a code to solve a differential equation using Runge Kutta method but when asked, s/he does not know what a numerical solution means and how it was different from the analytic one. The most surprising fact is that at the end of the day, I am forced by my colleagues to give marks to such students and pass them. At many nearby colleges, I am blacklisted and they wont prefer me as an external examiner for their students as I do not hesitate marking poor students with a zero. The conclusion from my side goes like this. We are culturally poor. We do not know what learning of a subject means. A very deformed meanings of the words Education, Teaching, Learning, Evaluating, Research, Paper publication etc is kind of running our blood. This all is a part of culture, this all is prevailing in our society.

Dheeraj Sanghi said...

@Mukund, my explanation is simpler, may be too simplistic. But it boils down to regulation, expectations and leadership. Higher education policy is based on four pillars - access, equity, cost and quality. Obviously, some of these pillars are in conflict with each other, and hence leadership at the national level is required to decide the policy carefully. In India, we have a policy to ensure access, equity and low cost. But no focus on quality (though we provide lip service to it, there is no policy initiative for this). And we have been exceptionally successful in meeting the policy goals. You can see an engineering college everywhere providing low cost degrees. Since quality was never a focus of our policy, we should be happy that we have been successful in meeting and indeed exceeding our goals.

Sanjay Goel said...

If most IIT faculty collectively decide that they will not recommend the acceptance of low quality PhD work at any university, this country will start seeing a ripple effect in improvement of PhD quality all over the country, at least in those disciplines in which IIT as a brand is considered to be the Mecca of India. Few IIT professors abstaining from the process does not help, when many others become silent spectators (or even careless gatekeepers) by not writing hard hitting reviews, when so deserved.


sk said...

Dear Prof Sanjay,

Is it the responsibilities of IIT Professors only to improve the system? Are they have any power to take action on other institutes? Is it not the responsibilities of all institutions? Why it has become the habit in India to just find problems in IITs or IIT Professors? They are neither having more power nor getting more infrastructure or more perks. We always talk about the quality of output from IITs, why not also to talk about quality of output from NITs, State universities, private universities. We have to improve at all levels.


Vinayak Naik said...

The state of medical education in India is also similar.

Varun Aggarwala said...

Nature has a brilliant profile of Prof. K VijayRaghavan in it's current issue.
I am sure that with excellent "administrators" like him, Indian science will definitely improve. I thought I should share this article because the comment section was becoming a bit too negative (guilty as charged !).

S said...

@ifunda, Imtiaz Mohammad:

This reminds me of a very important thing the then HoD of an IIX CS department said when some students were trying to convince him that the first algorithms/programming course should be in a language like C rather than a comparatively modern language like Java, because that was what the companies that came for placements "wanted". He said, "Our goal is not to teach students programming languages. Our goal is to equip them to pick up any language in less than a week if they need to."

Nitesh Bhandari said...

Same is true for school education as well. Over all very poor condition. Almost 90% government teacher positions are sold, and then these teachers are absent in class. CTET was a good step, but stopped at exposing the rot. B.Ed. is again practically a free degree, only used to delay workforce into teaching. Some private schools are doing good but all the rest are just enjoying the money or stuck due to CBSE's diktats. CBSE+NCERT ensures that they have the "results" (NTSE/JEE). Over all a pathetic board, not really serving any segment of society. Students end up in tuition classes or coaching classes, cramming to prove they are great by getting A+ in cramming .....

And this raw material goes into our colleges. After this schooling, the nation is left with around 3k academically inclined students (not marks, institute, "career", money inclined ones). Most of these end up in IITs and the rest of the institutes wanting to be good get the leftovers due to big mistakes of JEE type exams. These institutes spend at least 2 years undoing the damage done by schooling and are able to salvage a handful of students. Most are already waiting for the placement day the day they step in. Some fixed kids are also get "real" the day the 4th year bell rings.

School is where the problem is seeded into the kids. Fixing colleges is much harder than fixing schools due to the competence, cost and infrastructure requirement. Who's listening though?

tut m said...

Dear Dheeraj, you have very eloquently exposed the rot which has set in our technical education (outside IIT system). I always suspected this as I have worked with and interviewed many scores of such 'engineers' and found their theoretical and analytical foundation dreadfully weak. Outside IIT system , only other engineering college students who were generally competent were from BITS,Pilani (I am talking of B.Techs only). The disease of 'passing the exam' by cramming has led to this.
However, I'd like to add a note of caution here. In recent past , while interviewing students from IIT, I have noticed a distinct deterioration in IIT students also as far their understanding of fundamentals go. Whether it is a 'coaching' effect or 'software' effect , I have no idea. Newer IITs are not progressing as well as was expected and in fact, in some ways even lag behind BITS. (My comparison is only based on the parameters of understanding of the subject and ability to communicate ) .

Dheeraj Sanghi said...

@tut m, I also feel that IITs are focusing less on teaching and more on research now a days, and the quality of teaching is suffering (which is not necessary for improvement of research). However, there is a difference. The quality of learning is going down, not because they are being taught wrongly, but because the faculty is not interested in giving assignments, giving feedback during the course, giving easy grades so that students don't have an incentive to study at a deeper level, etc.

Ankur Bhatnagar said...

very bold article..I am glad to know that some 'rational' people are part of selection committees.
For me what is important is how to improve? What makes us and our university faculties so pathetic in comparison with other some foreign B grade universities?

Ashish Sureka said...

Dear Prof. Dheeraj,

I really like your reading your blogs – very insightful and informative.

Please see my recent blog/thought on a paradox in our education system (particularly Indian context) - I can’t paste the whole content of my blog here as a comment (will be too lengthy for a comment) but just to summarize my main point – we all agree that there is a severe scarcity of good quality education and teaching in India but at the same time there are some exceptions and islands of excellence [outside the IIT system] (for example, IIIT Delhi which you yourself rate quite highly). While there is a good percentage of students making the best from the resources available and learning [not only academically successful but I must say very good attitude also] – why do we still see a lot of academic dishonesty, cheating, plagiarism, disrespectful students (attacking teachers), disinterested/disengaged, disciplinary action and so on in places where there is good teaching and infrastructure?

While there is certainly a problem of scarcity [analogy: malnutrition], I have seen a problem of plenty also [analogy: obesity and food wastage also] = A Paradox.

Thanks, Ashish

smart said...

Dear Prof. Sanghi,
I did my B.Tech from NIFFT and them M.Tech from IITX in 2006. I am now working for a firm in US.

It is interesting to note that you are passing blanket comments on both ends. After reading your post i get an impression that IITs (so called Tier 1 institutes) are great and beyond that every thing in India is basically crap, unless you know them personally.

In my experience, I found 4/10 faculty in IITX absolutely crap. One of them could not properly answer some basic queries about Fe-C diagram, a very basic phase diagram in Metallurgy). Thanks to my teachers in NIFFT (Prof. P. T. Pushp, Prof. S. N. Prasad and Prof. Sohan Lal in particular), I could do very well. Prof.Pushp use to explain solidification and formation of microstrucutre of Cast Irons as if we were watching movie in the class. Their teaching standard was way better than IITX. They use to complain about lack of research infrastructure and were mainly involved in helping foundry and forge industry. They were very respected people. I feel very bad when somebody from an IVORY TOWER start passing judgement.

To be very frank, IITs carry minimum pay load (value addition Vs Tax payer money spent). In fact it is the alumini from these 2nd Tier and 3rd Tier (so called) institutions, with much less funding and often neglected, who are working in Indian labs and contributing ( think Agni, Mangalyan, etc.). IITs alumni are mainly adding value to US economy.

Prof., IITs have been disproportionately funded by tax payers of this country for more than 5 decades and only now other are getting some chance.

Dheeraj Sanghi said...

@smart, NIFFT is the premier institute in foundry and forging. So, it is not a surprise that you found the quality good. How would you compare NIFFT with which ever is the 10th rank institute in your field.

Your sensitivity towards criticism is proof enough for what I have written. Nowhere have I said that all faculty members in IITs are good, and nowhere I have written that all faculty members in non-IITs are bad. If you were a regular follower of my blog, you would have noticed how much I have criticized IITs even on quality issues.

Ajit R. Jadhav said...

Dear Dheeraj,

Noticed this post only today. I come from the extreme end of the other side; currently, I am having interviews at what would be the tier 3 institutions according to your scheme. (I try to avoid the tier 4 ones, even if I am jobless, in loans, and desperately need a job.)

Somewhat paradoxically I both agree and disagree with you, and ditto with "jayanti", and in fact ditto also with "smart" to a certain extent.

I agree with you as far as the quality of candidates goes. (The *other* candidates, obviously!... How could it be *me*? LOL!)

I agree with jayanti because at COEP, they did ask me where my PhD thesis was. When I told them that a soft copy was in my laptop and that I could give them a copy right away, they had only contorted faces as a reply. The Director had asked in angry tone, "Don't you know, you have to bring your PhD thesis to your interview? Aren't you proud of it?" Why this disbelief that I might not actually have defended a printed thesis? What the heck were they going to get by going through a 2kg heavy PhD thesis when they had not gone through my papers (which were enclosed with the application, and with complete with dramatic statements in application that this was a result conceptually first in 200 years, etc.?)

There are candidates, and there are candidates. There are interviewers, and there are interviewers.

I disagree with you and jayanti. At COEP, they didn't bother to let me explain even one of my results for a continuous stretch of 45 seconds. They would simply keep on interrupting me at the rate of some 3 people simultaneously loudly jumping in and raising absolutely incompetent/irrelevant questions about the question I was answering. (If in doubt, contact Prof. Gaitonde of IIT Bombay, who was in the experts committee. He and a junior faculty from IIT Bombay (I forgot his name, but he was one with a VJTI BE and a foreign PhD I guess) were the only ones who allowed time for me to think of, and give, my answers.) For instance, if I said in an apologetic style that "well, this result of mine is *not* really as impressive as it sounds, because this is a problem where the harmonic analysis applies and so the boundary data really determine the solution within the domain uniquely, and so, it's really just a simple initial thing to try more complicated things later, nothing profound," they wouldn't get it. Nothing. Nothing of importance. From the look on their faces, I can tell you, most of them didn't get what well-posed-ness of a problem means (when I mentioned it once).

In places other than COEP, the matter of the papers I wrote never, ever, comes up. (If in doubt, ask Prof. Gaitonde again. He again was in a committee which I faced last week.) They always remain minorly confused and majorly aggressive (if not offended) about the fact that I propose to be a Professor of Mechanical Engineering even if I had had my BE and MTech in Metallurgy.

[continued in part 2/4 of my comment]

Ajit R. Jadhav said...

One last point. I have learnt through hard experience that what a senior IIT professor thinks is the One Really True Fundamental Answer need not actually be the case, but it is Not Smart to point out the limits of his knowledge in a job interview. As I said, Prof. Gaitonde was there on two interviews, one at COEP a few years ago, and now, this past week at a private college in Mumbai. Both times he expected me to know what the fundamental principle behind FEM was.

At COEP, an incompetent HoD had raised the question: Give me a definition of FEM, and Prof. Gaitonde had joined the discussion a bit, though in his usual polite manner and with a perfect civic sense, only later. But it was clear that he remained in agreement about the validity of the question with the questioner, not me, what I was almost haphazardly trying to get at.

The first time, at COEP, here is how I had begun narrating how FEM was a hodge-podge technique whose bases are still not sufficiently clear. At that time, I had tried to explain in brief what FEM was, beginning with: "you divide the domain into a finite number of finite volumes that are mutually exclusive and collectively exhaustive." There was this Interruption. "What is the definition?" I: "Then, you assume a certain form for the primary unknown, say, a polynomial." Interruption (by more than one interviewer, now condensed here): "Why do you talk so much? Why can't you answer a simple question? You are applying for a Professor's position, can't you understand what a definition means? Can you give me just a definition of FEM? What is FEM---in one line?"

This time round (last week), Prof. Gaitonde asked something similar: "Can you name the fundamental principle that lies at the base of FEM?"

dirrrrrr..... The HDD of my mind thrashed a lot, right on the fly. I rapidly went through many things, including what I really wanted to say, trying to estimate what his response would be, estimating the further sub-trees we would be spanning, the options without interruptions that I would be left with, etc. He had grown just a bit quizzical by this time. I knew I had to play the same game; the disk stopped thrashing.

"As of today (with an emphasis on today), it would be the method of weighted residuals." He was happy.

To those who know FEM: To the best of my knowledge, the question cannot be answered as posed.

So, once it was raised in the stated form, here is what I silently went through: (i) the *classical* variational principle over a *single* element as the *only* fully rigorously established base; (ii) the sub-domain collocation as the defining characteristic of FEM, as accepted by pioneering and eminent professors like Prof. Taylor at Berkely (they rightly distinguish the Ritz method from the earlier Rayleigh method); (iii) but with sub-domain collocation not yet being on a fully well-understood theoretical basis (else, e.g., I wouldn't be able to *claim* a first in 200+ years for my result denying instantaneous action at a distance in diffusion); (iv) the method of weighted residuals (MWR) as just an analog to the classical variational principle in those instance in which its efficacy is redundant, viz. the cases in which a self-adjoint operator (or a functional form) can at all be defined; and, most crucially, (v) not a single soul (including mine) knowing as of today the basis behind the basis, viz. why MWR should still work when a functional form is not available (as happens in Navier-Stokes, or any equation having, what *I* (but not others) call, a differential nonlinearity). You see, MWR has never been rigorously established for those problems where a functional form is not available; it's just a conjecture for such problems.

[continued in part 4/4]

Ajit R. Jadhav said...

But it provides a *framework*---complete with all the unproved and even unknown voids---with which *some* thing can be built, even if the fundamentality of the principle isn't guarunteed. And so, the actual practice involves putting up smaller principles to deal with those voids of inconsistency such as nonlinearities etc. No one knows the scopes of these smaller principles. A multi-billion dollar successful industry indicates that MWR indeed is a good framework to to have. I too marvel at it. But I am careful to distinguish a conjecture from a theorem, and completeness from incompleteness---others are not.

In short, there is no One Fundamental Principle that is also rigorously reliable. You must say in the least: MWR + qualifications for non-self-adjoint problems + crucial ad-hoc measures (that aren't generally reliable but simply work in some situations, and you have to develop a sense of where they work). No text-book mentions this part.

But, I knew that Prof. Seshu of IIT Bombay (my former classmate at IITM) has written a neat text-book, that the syllabus at Uni. of Mumbai is patterned after his book (in part because it's inexpensive, & in major part because it comes from an IIT professor), that it begins with MWR (but not the CoV) as the first chapter (thereby distoring the proper hierarchy right at the fundamental level), that Seshu treats it this way not on his own but only because that's the current general agreement among FEM researchers world-wide, and therefore, that I should cite MWR as The Fundamental Principle NOW---IIT professors would be happy with my answer, NOW.

So, I said: "As of today, MWR."

Turns out, I was being smart---my HDD thrashing must have really been at a top-notch speed, because I could answer the question within the time-frame of the expected pause at the interview.

Yeah, I am getting better---at least at my interview techniques. But wish me luck. I still don't have a job in hand.


L said...

You are so right about research in colleges and state universities. Even the most basic requirement like checking reproducibilty of experimental results is unheard of. Many "international" journals publish if you pay Rs.2500-3000. There is "peer review" by a panel that gets paid for 'Okaying' a paper.

here is another link

L said...

You are so right about research in colleges and state universities. Even the most basic requirement like checking reproducibilty of experimental results is unheard of. Many "international" journals publish if you pay Rs.2500-3000. There is "peer review" by a panel that gets paid for 'Okaying' a paper.

here is another link

Dheeraj Sanghi said...

@Ajit, I feel a bit uncomfortable when length of the comments exceed the length of the blog post itself. This discourages readers from reading all comments and add more comments and affect the discussion on this forum. I would prefer if you really have so much to say than write it somewhere else (your own blog, for example), and post a link as a comment, with perhaps a brief summary here.

Ajit R. Jadhav said...

Dear Dheeraj,

Thanks; I will keep that in mind. It in any case is much more respectful of me than outright moderating my comments out, as has happened elsewhere. The reason I didn't want to post it at my blog immediately is that I am undergoing interview process, and while I don't care about the consequences of the contents, I want them---the interviewers---to focus on what I *can* teach. So, my current top-most post will remain there for some time. Please allow me some time, and I will sure move this material to my blog, too.

However, the absence of your comment on the content of what I wrote, even if directly crossing some of your implied positions, also is a bit interesting to me by itself.

Anyway, before closing this reply, just one question, which actually comes from the material I on-the-fly deleted this morning while posting my reply here; I now quote verbatim; see if you wish to address it:

"A question to you, re. optimization: `How do you really avoid getting trapped into any of the several local minima, and guarantee that you will reach the global minimum?' What is the answer expected here? That you exhaust the entire configuration space by enumeration i.e. brute force?"

I will check back this thread once tomorrow. My material should be at my blog around mid-July.



Unknown said...

I mostly agree to what Prof. Sanghi said. I passed out from a NIT 25 yrs back. I remember that three lecturers of my college used a 12 yr old classnote from BHU to teach us! We were expected to reproduce the class notes in verbatim ... including comma, semicolon and fullstop. Our topper used to prepare by completely mugging and re-writing the answers at least 5-6 times. Similar approach was also needed during the viva-voce.

And yes, the lecturer's knowledge could be questioned only at one's own peril.

The system asks for mostly learning by rotes. We get such 'Scholars' becuase our education system produces it.

Kundan Negi said...

Sir please give the correct answer too for virtual memory question. You see Sir, this is the problem with teachers in India. Most of them don't have correct knowledge and those who have it don't share it. Please don't feel bad but by reading your statement I felt you were non different than many viva teachers who ask very fantastic questions but never give answer for that and this way the cycle of ignorance keeps on going. But Sir real change will come when someone will break this cycle by giving the correct answer. So I request you to give correct answer to need for virtual memory question asked by you.

Dheeraj Sanghi said...

@Kundan, it is so easy to find the answer in any OS book, or just google it. I just googled it and the very top answer explains why we need virtual memory. As of this instance, the top answer is:

So instead of blaming others for not helping, please help yourself. That would take you far in your career. Spoon feeding is only for 1-year olds.

Mister Twistoff said...

Mr. Dheeraj, the lowering of quality has started from the school level.Teachers and boards heap marks on students to keep out of trouble. Teachers are not respected like they were since knowledge and thinking ability have taken back-seat. The emphasis is on looking-good superficiality- smart people know how to get through exams and land plum jobs. Indian companies will start feeling the heat soon on the heels of poor quality in higher institutions. I wonder why there are so many mistakes in the official JEE answers. This is happening in many places in the world. People are not ready to invest in hard thinking and work, since everything is done by a s/w package. These people will slip into writing programs for aircraft and spacecraft with disastrous consequences. It is difficult to find a honest, knowledgeable doctor or civil engineer. We are rotting from top to bottom.

Please write about this problem with your examples and data to the mainstream papers. An honest blog about this issue is so deeply buried in the internet.