| DICE2007is a fairly large model. GAMS Program that Nordhaus uses consists of 1200 equations and 1300 variables. GAMS software is very expensive and therefore it show us high performance. Although I do not know exactly, I saw remarks that GAMS can solve differentiation by symbolically. It is wonderful. Thus GAMS can solve this large model at once. I cannot make a similar program to GAMS. I am the amateur of this field. So it is very difficult to construct a program that can process those many equations and variables. When I made up my mind to make a program that can executable DICE2007 simulation, I have to devise the structure of the model depending on its special form. If I would not do so, new program cannot solve the model even if it uses 24 hours to execute. GRAS does not use so many variables as GAMS program uses. In order to see how I squeeze the number of variables, let us see the following diagram. It shows the interdependence among variables of DICE2007 model. ![]() GAMS program uses all those variables and add corresponding constraint equations. GRAS uses only series of gross products Y(t) and series of reduction rates MIU(t). The other variables are directly derived in every steps. |