The MGPT was the most eagerly awaited event for me on the academic calendar. The theory end quizzes were more of a formality that needed to be gotten out of the path, while the projects and MGPT were the real challenges. Last time around, I had almost done it. Having been misled by an sample input-output sequence that served as a red-herring, I lost my way in the chase. Even after having implemented the solution in just over twenty minutes, I failed to identify the cause of the phantom single X, the X mark of death, as far as an MGPT is concerned. I was totally devastated and vowed that I would do things differently the next time around, little knowing that it would prove to be equally disastrous.

The first of the two problems was an implementation of Prim’s algorithm in graphs while the second was a simple enough problem of string manipulation. At first glance, I thought that the graph related problem was beyond me, and that I would only tackle the second problem. However, fate had different plans in store for me and I turned the page over, to take a second look at the problem, without realizing that I was making a fatal choice.

After the second look, the problem seemed solvable and a skeleton of a solution even formed right there, in my mind, and it happened. Instead of first trying to solve the string related problem which was not only easier, but also one that I had conquered earlier (in its C avataar), I took up the first one first. My mind led me on a suicidal course, telling itself that I could solve both. Bad mistake. In hindsight, both the problems were eminently solvable, but the sequence was wrong. I ought to have tried the second one first, as the five students who cleared the MGPT by solving the same problem did.

An MGPT, I painfully realized, a couple of hours later, is very much like a Formula One Race. An F1 car has around 80,000 moving components within and if the pre-race assembly is 99.9% perfect instead of a perfect 100, the car can develop 80 different problems on the day of the race. An MGPT is no different. In fact, what sank my campaign was a simple component. I needed the value of the number of edges in the given graph. The method I adopted in the lab, on the fly, could not do it, and the tension did me in. Henceforth, I’m going to pay attention to every single bolt that goes into the car, ie every single line of code that goes into my program. The next time around, my accuracy will not be 99.9. I will push myself to achieve the three digit figure.

Hoping for a better show in my next date with Parikshak. And yes, I will remember the lessons learned.