Probleemstelling voor ChatGPT:
I want to make a GeoDMS script using array operations to choose for a set of buildings i:=1...10M a random combination of k:=1...4 alternatives for each of j:=1...10 measures each with unconditional likelihood of P_ijk >= 0.
Each alternative provides an amount of discretized integer energy saving of S_ijk in 0...MaxS. You may assume that S_ijk >= S_ijl if k < l.
A combination of alternatives k_j is feasible for building i if SUM S_ijk_j >= MaxS. Assume MaxS is something near 50.
How can I effectively pseudo-random draw k for each pair of i,j with probability proportional to PROD j:=1...10 P_ijk_j for feasible k_j and zero probability for all infeasible combinations.
Please provide GeoDMS code and assume that I am able to use iterations in the generation of the GeoDMS syntax.
Apply Dynamic Programming on the j-partial sum of S_ijk and trace back among the alternatives to select a feasible combination for each building to avoid having to consider all combinations of measures.
If we use Float32 for likelihoods and probabilities and UInt8 for S_ijk, how much RAM would be required to run this code?
Probleemstelling voor ChatGPT:
I want to make a GeoDMS script using array operations to choose for a set of buildings i:=1...10M a random combination of k:=1...4 alternatives for each of j:=1...10 measures each with unconditional likelihood of P_ijk >= 0.
Each alternative provides an amount of discretized integer energy saving of S_ijk in 0...MaxS. You may assume that S_ijk >= S_ijl if k < l.
A combination of alternatives k_j is feasible for building i if SUM S_ijk_j >= MaxS. Assume MaxS is something near 50.
How can I effectively pseudo-random draw k for each pair of i,j with probability proportional to PROD j:=1...10 P_ijk_j for feasible k_j and zero probability for all infeasible combinations.
Please provide GeoDMS code and assume that I am able to use iterations in the generation of the GeoDMS syntax.
Apply Dynamic Programming on the j-partial sum of S_ijk and trace back among the alternatives to select a feasible combination for each building to avoid having to consider all combinations of measures.
If we use Float32 for likelihoods and probabilities and UInt8 for S_ijk, how much RAM would be required to run this code?