About us

Simulation

Published on Oct.29 ,2014 : The Denki shinbun(The Electric Daily News)
Shojiro Matsuura
Chairman of JANSI

As the computer and its operating technology advance, future predictions by computer simulation are performed in many areas. For example, global warming simulation, which was actively performed for a period of time in the past and slowed down subsequently, is being reactivated. Diverse economic trend predictions seem to be performed more and more actively.

What we usually get wonder about the results of these predictions is how we should understand their accuracy, i.e. uncertainty. When we predict global warming, we basically use the computation method and input data obtained from the past climate observation data. Thus, a certain level of certainty may be guaranteed. However, the impact of recent and future human civilization activities and that of practically unpredictable crustal movements and major meteor strikes cannot be developed into input data. If that is ever possible, it will merely render the ambiguity of the results extremely significant. We cannot think it possible at all to use such data.

As for economic trend prediction, past information, including statistical data, makes predictions considerably accurate. However, they may involve the difficulty that prediction methods must extensively consider diverse personal and social characteristics and their fluctuations.

Nonetheless, the development of computational technology and specialists’ efforts have connected various simulation programs one another. In reference to these, social decisions of considerable importance are being made as a practical matter. What we are always concerned about such decisions is whether the certainty or uncertainty of simulation results on which the decisions are based is appropriate. How should we determine it?

In the area of nuclear energy in which I am involved, it is one of the most important challenges to determine “to what extent the level of nuclear use can be lowered” or conversely “to what extent nuclear use is essential.” What is questioned here is the accuracy of simulation of Japan’s future energy demand and energy resource necessary to meet the demand.

With regard to the future prediction of energy demand, a large number of simulation results have already been published. The question is to what extent we can trust them. Although there may be a sort of common-sense estimation among experts, we outsiders do not have a clue. It is inevitable that future prediction involves some uncertainties. There should be some kind of empirical attempt leading to a normal common-sense decision.

For instance, how about computing the process with the current future prediction method, by which we can go back to a point of time in the past and come back to the present? That is to simulate the process starting from 1973 in which we were hit by the oil crises up to the present time. This will allow us to understand the details over that period almost exactly. It is natural that it was impossible to input accurate data leading to the present in 1973. Computational conditions would also change rapidly. Therefore, it is natural that the wrong result will be computed if the basic data and prediction conditions available in 1973 are used. If we make changes as major condition alterations occur, however, we should be able to predict the present status in the end.

During this period of time, we can grasp the reasons for input data and condition changes considerably accurately. We already have the information about what happened under what circumstances and what we should input or modify. We should be able to understand what difference would result ultimately if we do not make any modifications. If this kind of attempt is made as to whether or not to select nuclear use, we may be able to understand more accurately what degree of adjustment is necessary against what sort of changes for future prediction. I hope some experts will make such an attempt.


end