Friday, February 13, 2009

Theory vs. Experiment

In an interesting editorial, Professor Balaram wonders about the differences between engineering and science, especially in research and goes on to examine the divide between theory and experiment. He says

In institutions like my own, the Indian Institute of Science, where research is a prime focus, the distinctions between scientists and engineers are indeed blurred. Are there two distinct species of researchers who can be identified, whose characteristics mark them out as decidedly different? I might venture to suggest that a contemporary classification may separate theorists from those who do experiments. Even in the engineering departments which once boasted of large halls full of clanking machinery, the workshops are largely still and silent. The mechanics are largely gone, glassblowers nearly extinct and draughtsmen swept into memory by advancing technology. The lathes stand forlorn and unattended and large structures one used for experiments serve solely to remind us of an era long gone. ...The terms computational chemistry and biology describe an increasing tribe of researchers far removed from the pain, excitement and thrill of experiment.

Three years back, I was asked to investigate the number of thesis (and publications) from engineering departments that had an experimental component. When I reported back with a number of 17%, the administration was surprised because the number was higher that what they had expected (which was 10%).

Three reasons are often given as to why experimental research in engineering is scarce. The first is that it is too difficult, the second is that engineering research everywhere is theoretical and the third is that people switch. I will analyze all the three arguments and present my own argument.

The first argument is that the number of faculty and students who do experimental research in engineering is so few in India because of the enormous difficulties encountered in setting up and maintaining an experimental research laboratory. Even a few people who return from US after doing experimental research for their doctorates find out it is simply too time consuming and difficult to do experimental research in India, take the easy way out and switch to computations.
However, the science faculty in IISc predominantly do experimental research (nearly 70%). After all, the inefficient financial and administrative staff encountered by both the science and engineering faculty are the same. Thus the lack of experimental research in engineering can not be just because of the physical and mental difficulties encountered in doing experimental research in India.

The second reason is that engineering research everywhere is predominantly theoretical. I tried to investigate whether engineering research is also theoretical in US. However, when I did an analysis for the top 20 universities (based on their publications), I found that this was not so. However, I did find that a disproportionately large number of Indians worked in theoretical research groups in USA.

The other argument that I am told (half-jokingly) is that many engineering faculty are good in math that they can switch to computations much more easily than science (especially chemistry and biology) faculty. Therefore, engineering faculty switch from experimental work to simulations while science faculty do not. However, I do not think so. I think it has largely to do with the training of the faculty during their graduate school and people do not switch fields entirely and be productive in that field.

I think the problem of not having good faculty in engineering who do experimental research is the following. What has happened now is that professors in engineering who do theory and simulations control the funding as well as recruitment of new faculty. Naturally, they are able to judge their own ilk better and often unwilling to take the risk of recruiting faculty who do experimental research. The start up funding required for a faculty who does simulations is restricted to a server costing less than Rs. 1 million while it can run into several millions for a faculty who does experimental research. Therefore, the argument is that even if a mistake is made in recruiting a faculty who does no research after joining, the cost of failure is not large!

I am reminded of Tesla's statement, "Today's scientists have substituted mathematics for experiments, and they wander off through equation after equation, and eventually build a structure which has no relation to reality."

I think this problem is going to get aggravated in the future. Last week, I had visited a "reputed" engineering college. Their mechanical engineering students had not seen a lathe because workshop was an optional subject in their curriculum ! We have started a new IIT and admitted students but we have not identified even the city (let alone the land or building) in which it is going to start. We have started several more IITs and IISERs which function from a small building. Faculty for all these institutions had to be recruited in a hurry because we admitted students before we recruited faculty. All these institutions preferred to recruit faculty who do computations because there was no way that faculty who do experimental research can be recruited into institutions that have only one small building. Ultimately, these faculty who do theoretical research will recruit other faculty and the problem is going to even more acute.

Therefore, I think if some one is really keen to enhancing experimental research in engineering, it has to be done in established institutions like IISc and the old IITs. Of course, they are several ways to do this but it can be only accomplished if one is interested in doing this.


Anonymous said...

I agree with the observation. I think we need to build at least a few experimental labs in certain specific areas that are of common interest to the faculty.

Anonymous said...

i truly agree with u. If somebody doing research at IISc..he/she should be truly non-experimental. I am a recent drop out from IISc and one of the reason to drop out is the same ;)

Giri@iisc said...

I am very sorry that you dropped out of IISc. It should not have happened. You should be able to switch advisors.

Anonymous said...

I am completely in agreement with you. It is reasonable to support the experiments with computational facilities, but complete dependence on thoery is not going to help the scientist/society. There are areas like weather prediction, where there is almost no scope for experimentation and one has to rely mostly on computations. Unless we have such problems, it is always good to have experimental facilities atleast to prove the simulation results. I regularly watch scientific programs on NGC/Discovery, where students from MIT, Stanford, etc do great equipment design and manufacture on their own(right from welding to preparing electronic circuit) very easily, which is very much required in research and such skills are missing in India. To materialize these in India, there has to be good focus on experiments (atleast to understand theory) right from schooling and has to be continued without any break........

I feel that there has to be balance between theory and experimentation, and any extreme, I feel is not going to help in efficient research.

Giri@iisc said...

"I feel that there has to be balance between theory and experimentation, and any extreme, I feel is not going to help in efficient research."

Could not have said it better !

MITS GROUP said...
This comment has been removed by a blog administrator.
Anonymous said...

The other issue is that many engineering faculty in IISc and IITs do consultancy that research takes a backseat. This is one reason why engineering faculty publish so little in India