0. The earlier posts in this series may be found here [^] and here [^].
In this post, I am going to talk about two stunning papers.
1. The First Paper:
Before coming to the paper itself, let me discuss its background a bit.
The paper has a topical relevance to the current news; its subject matter in a way refers to those most unfortunate accidents on the Mumbai-Pune Expressway.
In the recent times, there has been a spate of one particular kind of accidents on the Expressway: a tire of a car (or of a tempo or even a container truck) suddenly bursts open, the car gets out of control, crosses over the median and lands in a lane on the opposite side of the traffic, and ends up having a head-on collision.
People have been discussing the matters, building awareness campaigns, trying to analyze causes. For instance, things like: over-speeding as the cause of the tires bursting open; the absence of a tall enough separating wall on either sides; and human factors such as loss of concentration, drunk-driving, etc.
It’s a fact that on the Expressway most every car exceeds the speed limit (of 80 km/hr). You can see many amateur YouTube videos of passenger buses and ordinary passenger cars registering 120–140 km/hr as a routine. BTW, while watching these videos, notice that in India we drive on the left-hand side, and so, the fastest traffic is (usually!) to be found in the rightmost lane.
When a car unfortunately develops a burst tire and lands up on the other side of the road, not only is it the fastest moving car in the forward direction but also the cars in whose path it lands up after crossing over the median, also happen to be the fastest moving ones on the opposite side. Thus, collisions can possibly occur at, say, 100 + 120 = 220 km/hr (even after discounting the reduction in speed during the crossover, say, from 140 to 100 km/hr), making the situation far more grave. Contrast this situation with that of a car moving at 80 km/hr in the left-most lane. Even if the driver loses control and the car veers on the left-hand side, it still collides on to the protective railing (having quite a bit of capacity to absorb the impact energy) at only, say, 60–70 km/hr. The difference of speeds in a head-on collision is, thus, three times over. Little wonder the percentage of fatalities in this kind of accidents has been so high.
Then, there are issues like the lights of the cars on the opposite side shining directly in your eyes, because there are no shades, no dividing walls that are high enough.
Though I am not at all an expert here, another issue I am toying with is the development of a peculiarly wavy kind of surface due to the combined effects of rolling and vibrations, especially as caused by heavy weight container trucks travelling on a road. A similar effect can be seen also on the railway tracks, where a wavy profile gets developed. However, in the case of the rails, due to the homogeneity of the material, it becomes a somewhat simpler problem to analyse within a mechanics framework. Even then, let me tell you, it’s a very complex problem to analyse. IIT Kanpur, for instance, has been conducing research sponsored by railways on this problem for many years, and one could say that they are still in a somewhat early modelling stage. Now, when it comes to a steel-reinforced cement-concrete road, it becomes, from the applied mechanics viewpoint, a much more complex problem to model. When the road is made of concrete, I am guessing, the peaks of the wavy profile could perhaps be a bit a too sharp, thereby introducing enhanced high frequency components in the car-road interactions, vibrations and what effectively are small-scale but high-speed impacts. Now, if the tires are worn out and are conventional (i.e. if they carry tubes and are not of the tubeless variety), and if a driver has been over-speeding for a long time thereby overheating them, especially on this kind of a more sharply wavy surface, there could be increased chances of the tube of the tire bursting open. (The tubeless tires have a tendency to more gracefully handle an event of a similar kind; the running out of the air or nitrogen is less intense and not so catastrophic.)
So, people have been talking of things like that, and, as you can see, I too have joined them, wondering aloud about what possibly could be the causes of such accidents.
Against this backdrop, here is a paper I found in a recent Web search:
N. P. Dharmadhikari, D. C. Meshram, S. D. Kulkarni, S. M. Hambarde, A. P. Rao, S. S. Pimplikar, A. G. Kharat, and P. T. Patil (2010) “Geopathic stress: a study to understand its nature using Light Interference Technique,” Current Science, vol. 98, no. 5, pp. 695–697
I must say that the paper is simply stunning.
Check it for yourself, and if you think otherwise, do drop me a line. I am sure, though, that you, too, would find it simply stunning. Especially noticeable is the level of rigor in both experimentation and the kind of data that have been gathered therein. Do have a look.
I gather that “Current Science” is “the [sic] leading interdisciplinary science journal from India […] started in 1932 by the then stalwarts of Indian science such as CV Raman, Birbal Sahni, Meghnad Saha, Martin Foster and S.S. Bhatnagar.” I also gather that it is indexed in “Web of Science, Current Contents, Geobase, Chemical Abstracts, IndMed and Scopus” [^].
BTW, the fact that the journal is indexed by so many services, esp. Scopus, goes to satisfy the requirements clarified by Professor Dr. T. P. Singh (Director, Symbiosis Institute of Technology). Recently, in an informal chat, he had mentioned to me that he cannot hire me because I do not have two journal papers already published, and so, I cannot become a PhD guide at Symbiosis from day 1. He then added that it is now his requirement to hire only those people as Full Professors who have two papers published in Scopus/similar indexed journals already, so that they can become PhD guides right from day 1 (his emphasis) and then, he continued in the same conversation, that during the last semester (when I too had applied at Symbiosis) he had hired two people as Full Professors but without PhD degrees so they cannot act as PhD guides from day 1 (here, the emphasis is mine). BTW, at the time that he hired those two people, I, too, had applied for the same post (having PhD already in hand), but was not called for interview at all. … Yes, sometimes Prof. Singh somehow has a charming way of reminding me of the late Jaspal Bhatti, an engineer.
Anyway, coming back to this paper, please note, engineering college professors from JSPM, MIT and Sinhagad have participated in it. There are as many as 7 authors to this paper. (Isn’t this fact stunning by itself?) They all, I suppose, have been approved by the UGC committees for recruitment of professors. In fact, this research began with a PhD granted by the University of Pune, in the first place.
BTW, if you are interested, here is another article which carries more or less the same level of rigor but it is definitely much more readable: [^].
2. The Second Paper:
Another paper, too, has a bit of a topical importance, at least as far as this blog is concerned, in the light of the recent discussion on the Fourier theory and all.
The paper in question is this:
G. K. Kharate, V. H. Patil and N. L. Bhale (2007) “Selection of mother wavelet for image compression on basis of nature of image,” Journal of Multimedia, vol. 2, no. 6, pp. 44–51
This paper, too, is stunning.
Why do I find it stunning?
In this paper, the authors experimentally study the extent to which different types of images can be compressed, using wavelets transforms. In particular, the extent of compression that is possible for a natural image (a woman’s portrait) is compared and contrasted with certain artificial images consisting of simple regular patterns such as horizontal/vertical/oblique lines. The conclusions reached by the authors consist of the following:
“The results demonstrate that for line-based image percentage of zeros is more for db1 as compared to other wavelets and more energy is retained. It shows that the loss of information is less hence the quality is better. Also for natural image percentage of zeros is more for db4 as compared to other wavelets that indicates more compression and the loss of information is less.
If the import of the conclusion is not immediately clear, here are a few passages from the main text of the paper that may help you in ascertaining the nature of the paper:
“It is observed that for periodic functions, Fourier analysis is ideal. However, wavelet transforms are not restricted to only the periodic function, but for any function, provided it is admissible.”
I was left wondering: How about the computational complexity of FFT vis-a-vis wavelets—particularly because the paper deals with compression efficiency? Reading further on, I then saw this:
“We suggest the minimum two steps for the wavelet decomposition. In first step, decompose the image by using any mother wavelet. On the basis of energy content of approximation and detail coefficients, the image is [first] classified as line image or natural image. Then select the proper mother wavelet for further decompositions as mentioned above.” [Emphasis in bold is mine.]
How about entropy? Well, this aspect is not as clearly indicated.
The Journal of Multimedia is, of course, an international journal published from the USA; it is indexed in many services, including Scopus [^].
Anyway, the first author is the current Dean of the Faculty of Engineering at the University of Pune. I mean to say: he was first awarded a doctorate in engineering by the University of Pune (under the guidance of A. A. Ghatol); he was selected by a UGC committee for a Full Professor’s position; he was also democratically elected by the professors coming from all the engineering colleges affiliated to the university of Pune; he now sits on the UGC interview committees that decide whether someone (e.g. me) is to be at all interviewed or not.
Do note that it’s not his the then PhD student (i.e. V. H. Patil), but Kharate himself who is the first author of this paper. Not so stunning. Not any more.
But, yes, the paper, overall, is stunning. Its conclusion simply has a way to force you to do a double take. Its very appearance is against what you expect intuitively—you never expect a paper of this kind to appear in a typical journal covering topics such as multimedia, image or signal processing, or wavelets analysis.
As you might know, the wavelets transform is similar to the Fourier transform though it has a categorically less computational cost involved with it as compared to FFT. Also, take a moment to think about what kind of connections could at all be possible between computational complexity, information entropy and the compression efficiency.
Now, reconsider the cited text and the conclusion of this paper.
Hopefully, now you can see why I say that you must read this journal paper, too.
* * * * * * * * * * * * * * *
BTW, I have finally been rejected in the interviewing process at the JPBTI Phaneesh Murthy’s company, iGATE. The reason communicated to me (over the phone) by Asad Kadri this late evening had to do with the fact that I don’t have documentary evidence for having actually done jobs at three places: (i) Mukand (in 1983), and (ii) Thermax (in 1983–84); at both these places I was a trainee engineer; and then, (iii) QueriSoft (in 1995–96 i.e. about 15+ years ago), a company which is closed by now. BTW, I had arranged for a former colleague to informally verify the fact of my having worked with Querisoft, and anyway did have a xerox copy of the appointment letter for it. However, iGATE apparently thinks that their end-client, Rio Tinto, would find it unacceptable that all of my employment cannot be verified. In any case, they didn’t reply back to that email written by that past colleague of mine (who later on worked as a director in a large American MNC, and has now taken voluntary retirement from the IT field).
On my part, I find both Rio Tinto and iGATE to be very funny companies—tinted, as if they were, with a very dogged and out-of-the-world sort of a sense of humour. (They took three months, two interviews with the VP, three interviews with HR, and many telephonic calls and emails exchanges, in order to reach the conclusion that they cannot appoint me because I do not carry experience certificates for companies where I worked two decades ago and more.)
So, my job search continues, and something on the academia side might materialize—something that allows me at least, say, survival. I will keep you posted.
… But, don’t forget only to see those two papers, na!
* * * * * * * * * * * * * * *
A Song I Like:
[For the above-mentioned reasons, the inclusion this time round of this section is, once again, purely on the whim.]
(Hindi) “machalti hui hawaa main chham chham…” (“hamare sang sang chalein Ganga ki leheren”)
Singers: Kishore Kumar [an unlikely choice, esp. given early ’60s], and Lata Mangeshkar
Lyrics: Majrooh Sultanpuri [?]