Fishing quotas are unpleasant but efficient to control the productivity of a fishing site. A popular model has a stochastic differential equation for the biomass on which a stochastic dynamic programming or a Hamilton-Jacobi-Bellman algorithm can be used to find the stochastic control–the fishing quota. We compare the solutions obtained by dynamic programming against those obtained with a neural network which preserves the Markov property of the solution. The method is extended to a multi species model and shows that the Neural Network is usable in high dimensions.
Citation: |
[1] |
Agardy, M. T., Advances in marine conservation: The role of marine protected areas, Trend. Ecol. Evol., 1994, 9(7): 267−270. doi: 10.1016/0169-5347(94)90297-6.![]() ![]() |
[2] |
Allen, P. M. and McGlade, J. M., Modelling complex human systems: A fisheries example, European Journal of Operational Research, 1987, 30(2): 147−167. doi: 10.1016/0377-2217(87)90092-0.![]() ![]() |
[3] |
Auger, P. and Pironneau, O., Parameter identification by statistical learning of a stochastic dynamical system modelling a fishery with price variation, Comptes Rendus Mathematique, 2020, 358(3): 245−253. doi: 10.5802/crmath.2.![]() ![]() |
[4] |
Bachouch, A., Huré, C., Langrené, N. and Pham, H., Deep neural networks algorithms for stochastic control problems on finite horizon: Numerical applications, Methodol. Comput. Appl. Probab., 2022, 24: 143−178. doi: 10.1007/s11009-019-09767-9.![]() ![]() |
[5] |
Bick, A., Quadratic-Variation-Based dynamic strategies, Management Sciences, 1995, 41(4): 722−732. doi: 10.1287/mnsc.41.4.722.![]() ![]() |
[6] |
Brochier, T., Auger, P., Thiao, D., Bah, A., Ly, S., Nguyen-Huu, T. and Brehmer, P., Can overexploited fisheries recover by self-organization? Reallocation of the fishing effort as an emergent form of governance, Marine Policy, 2018, 95: 46−56. ![]() |
[7] |
Carmona, R. and Laurière, M., Convergence analysis of machine learning algorithms for the numerical solution of mean field control and games: II–The finite horizon case, To appear in Annals of Applied Probability, arXiv: 1908.01613, 2019.
![]() |
[8] |
Danielsson, A., Efficiency of catch and effort quotas in the presence of risk, Journal of Environmental Economics and Management, 2002, 43(1): 20−33. doi: 10.1006/jeem.2000.1168.![]() ![]() |
[9] |
Gatto, M., The evolutionary optimality of oscillatory and chaotic dynamics in simple population models, Theoretical Population Biology, 1993, 43(3): 310−336. doi: 10.1006/tpbi.1993.1013.![]() ![]() |
[10] |
Gobet, E. and Munos, R., Sensitivity analysis using Itô–Malliavin calculus and martingales, and application to stochastic optimal control, SIAM Journal on control and optimization, 2005, 43(5): 1676−1713. ![]() |
[11] |
Haddon, M., Modelling and Quantitative Methods in Fisheries, CRC Press, New York, 2011.
![]() |
[12] |
Han, J. and E, W., Deep learning approximation for stochastic control problems, Deep Reinforcement Learning Workshop, NIPS, 2016.
![]() |
[13] |
Hecht, F., New development in FreeFem++, J. Numer. Math., 2012, 20(3–4): 251−265. ![]() |
[14] |
Katsukawa, T., Numerical investigation of the optimal control rule for decision-making in fisheries management, Fisheries Science, 2004, 70(1): 123−131. doi: 10.1111/j.1444-2906.2003.00780.x.![]() ![]() |
[15] |
King, M., Ecology and Ecosystem in Fisheries Biology, Assessment and Management, Blackwell Publishing, 1995.
![]() |
[16] |
Laurière, M. and Pironneau, O., Dynamic programming for mean-field type control, J. Optim. Theory Appl., 2016, 169(3): 902−924. doi: 10.1007/s10957-015-0785-x.![]() ![]() |
[17] |
Le Bris, C. and Lions, P. L., Existence and uniqueness of solutions to Fokker-Planck type equations with irregular coefficients, Comm. Partial Differential Equations, 2008, 33(7): 1272−1317. doi: 10.1080/03605300801970952.![]() ![]() |
[18] |
Moussaoui, A. and Auger, P., A bioeconomic model of a fishery with saturated catch and variable price: Stabilizing effect of marine reserves on fishery dynamics, Ecological Complexity, 2021, 45: 100906. doi: 10.1016/j.ecocom.2020.100906.![]() ![]() |
[19] |
Moussaoui, A., Bensenane, M., Auger, P. and Bah, A., On the optimal size and number of reserves in a multi-site fishery model, Journal of Biological Systems, 2015, 23(1): 31−47. doi: 10.1142/S0218339015500023.![]() ![]() |
[20] |
Pagès, G., Numerical Probability: An Introduction with Applications to Finance, Springer, Cham, 2018.
![]() |
[21] |
Pagès, G. and Pironneau, O., Protection of a Fishing Site with Optimal Quotas: Dynamic Programming versus Supervised Learning, to appear, Encyclopedia.
![]() |
[22] |
Pagès, G. and Printems, J., Optimal quadratic quantization for numerics: The Gaussian case, Monte Carlo Methods and Appl., 2003, 9(2): 135−165. doi: 10.1515/156939603322663321.![]() ![]() |
[23] |
Pimm, S. L., The dynamic s of flows of matter and energy, In: McGlade, J.(ed), Advanced Ecological Theory: Principles and Applications, Chapter 6, Blackwell Science, 1999.
![]() |
[24] |
Volterra, V., Variations and fluctuations of the number of individuals in animal species living together. In: Chapman, R. N. (ed.), Animal Ecology, McGraw–Hill, 1931.
![]() |
[25] |
Weitzman, M. L., Landing fees vs harvest quotas with uncertain fish stocks, Journal of Environmental Economics and Management, 2002, 43(2): 325−338. doi: 10.1006/jeem.2000.1181.![]() ![]() |
Solution
Values of the average cost versus
Simulation of the fishing model with a quota function computed by SDP and
Simulation of the fishing model with a quota function computed by SDP and
SDP solution:
HJB solution: value function
Solution
Values of the cost functions for the 3 space
Simulation of the fishing model with a quota function computed by HJB and
Simulation of the fishing model with a quota function computed by HJB and
HJB:
HJB:
HJB:
HJB:
HJB:
HJB:
HJB:
HJB:
HJB:
HJB: Optimal biomass and quota function when X1(0) = 0.7 and X1(0) = 1.3
HJB: Optimal biomass and quota function when X2(0) = 0.7 and X2(0) = 1.3
HJB: Optimal biomass and quota function when X3(0) = 0.7 and X3(0) = 1.3
Dynamic feedback control,
Single species: Values of the cost versus
Simulation for a single fish species computed by the Markovian Neural Network and
Simulation for a single fish species computed by the Markovian Neural Network and
Optimal biomass and quota function computed with the two layers NN when X0=0.7 and X0=1.3
Optimal biomass and quota function computed with the two layers NN when X0=0.7 and X0=1.3
Optimal biomass and quota function computed with the two layers NN when
Values of the average cost versus
5 species:
5 species:
5 species:
5 species:
5 species:
5 species:
5 species:
5 species:
5 species:
5 species: Optimal biomass and quota function computed with the single layer NN when
5 species: Optimal biomass and quota function computed with the single layer NN when
5 species: Optimal biomass and quota function computed with the single layer NN when
3 species:
3 species:
3 species:
3 species:
3 species:
3 species:
3 species:
3 species:
3 species:
3 species: Optimal biomass and quota function computed with the single layer NN when
3 species: Optimal biomass and quota function computed with the single layer NN when
3 species: Optimal biomass and quota function computed with the single layer NN when