New algorithm revolutionizes the formula “x / log (x)” so far accepted by computer scientists and can pave the way to new discoveries. The author of the discovery himself is “in shock” with his formula, which calls into question everything we knew about space and time in computing.
“It’s a kind of abnanch in my view of the world,” he begins by explaining Ryan Williamsresearcher at Massachusetts Institute of Technology. ”I’m still shocked by the fact that this does not even exist.”
Time and memory space are the two main restrictions on computing, explains the.
Some problems require a lot of memory, and some a lot of time (or both). The time in computing is The number of steps a computer takes to perform a certain task, and the space is the Number of memory positions that the same task requires.
If a task requires x steps, at worst, when the computer needs to access your memory for each step, you will need x memory positions – any computing that takes x steps can be done with x/log x of memory. Or not?
A new one shows that this ratio can be drastically reduced – to the square root of x / log x. Instead of 100 or 50 memory slots, a 100 -step problem could be reduced to 15 slots.
The researcher who discovered this formula, Ryan Williams, is “in shock” with the discovery, that calls into question everything we knew about space and time in computing.
It took months to convince me that it wasn’t fake, ”he says. “It is still very difficult for me to understand what is happening. I can go through all the steps, the proof, and to verify that each step is correct and that it is true. But in the end, I still question me. ”
The expert then created a model that can represent any computational problem and then applied the new Tree Evaluation Algorithmshowing that it could drastically reduce the amount of memory required. The algorithm involves mathematical tricks and “magic cancellations” that ultimately provide valid answers.
“It’s very exciting,” comments Ian Mertz of Charles University in the Czech Republic, one of the investigators who developed the new algorithm. “It is undoubtedly a Contra-intuitive resultAlthough I say that our tree assessment algorithms were already quite counterinitive. ”
“Now that we know that efficient time in terms of time can be efficient in terms of space, we can Search for solutions That are quite good for both time and space at the same time, and this is useful in a real sense, ”he concludes.
