Automatic simulation generation

1 . A computer-implemented method comprising:

obtaining supply chain data of a supply chain;

generating a supply chain network graph that represents relationships between locations indicated by the supply chain data;

determining classifications of the locations indicated by the supply chain data;

determining agent rule models based on the supply chain data; and

generating a supply chain simulation based on the supply chain network graph, the classifications of the locations, and the agent rules, wherein generating the supply chain simulation comprises:

receiving a natural language user input specifying one or more constraints of the simulation;

providing the natural language input as input to a trained transformer network that is configured to generate constraint representations from natural language inputs, wherein the transformer network has a tuning layer that generates constraint representations from internal representations;

generating the simulation according to the supply chain network graph, the classifications of the locations, and the agent rule models, and the one or more constraints; and,

receiving, as output from the transformer network, one or more constraint representations for the natural language input.

2 . The method of claim 1 , wherein the user input specifies one or more parameters that were not reflected in the supply chain data, and further comprising:

using the existing supply chain data to estimate the one or more parameters that were not reflected in the supply chain data.

3 . The method of claim 1 , wherein the natural language input specifies one or more aspects of a supply chain that were not reflected in the supply chain data.

4 . The method of claim 3 , wherein the natural language input specifies an additional factory, warehouse, distribution center, or retail store that was not reflected in the supply chain data.

5 . The method of claim 3 , wherein the natural language input specifies a range representing a number of possible factories, warehouses, distribution centers, or retail stores.

6 . The method of claim 3 , further comprising repeatedly adjusting one or more parameters of the simulation and repeating the evaluation process for each different combination of parameters of the simulation; and

generating an output identifying a combination of parameters having a highest scoring performance metric.

7 . The method of claim 1 , further comprising:

performing an evaluation process on the generated simulation and computing one or more performance metrics for the generated simulation.

8 . The method of claim 7 , wherein the evaluation process and the generation of one or more performance metrics is performed for a plurality of generated simulations.

9 . The method of claim 1 , further comprising:

searching through a plurality of possible supply chain simulations for an optimization of one or more performance metrics.

10 . The method of claim 9 , wherein the one or more performance metrics includes a cost, time, inventory, fragility, or carbon cost.

11 . A system comprising:

one or more computers and one or more storage devices storing instructions that are operable, when executed by the one or more computers, to cause the one or more computers to perform operations comprising:

obtaining supply chain data of a supply chain;

generating a supply chain network graph that represents relationships between locations indicated by the supply chain data;

determining classifications of the locations indicated by the supply chain data;

determining agent rule models based on the supply chain data; and

generating a supply chain simulation based on the supply chain network graph, the classifications of the locations, and the agent rules; wherein generating the supply chain simulation comprises:

receiving a natural language user input specifying one or more constraints of the simulation;

providing the natural language input as input to a trained transformer network that is configured to generate constraint representations from natural language inputs, wherein the transformer network has a tuning layer that generates constraint representations from internal representations;

generating the simulation according to the supply chain network graph, the classifications of the locations, and the agent rule models, and the one or more constraints; and,

receiving, as output from the transformer network, one or more constraint representations for the natural language input.

12 . The system of claim 11 , wherein the user input specifies one or more parameters that were not reflected in the supply chain data, and further comprising:

using the existing supply chain data to estimate the one or more parameters that were not reflected in the supply chain data.

13 . The system of claim 11 , wherein the natural language input specifies one or more aspects of a supply chain that were not reflected in the supply chain data.

14 . One or more non-transitory computer storage media encoded with computer program instructions that when executed by one or more computers cause the one or more computers to perform operations comprising:

obtaining supply chain data of a supply chain;

generating a supply chain network graph that represents relationships between locations indicated by the supply chain data;

determining classifications of the locations indicated by the supply chain data;

determining agent rule models based on the supply chain data; and

generating a supply chain simulation based on the supply chain network graph, the classifications of the locations, and the agent rules; wherein generating the supply chain simulation comprises:

receiving a natural language user input specifying one or more constraints of the simulation;

providing the natural language input as input to a trained transformer network that is configured to generate constraint representations from natural language inputs, wherein the transformer network has a tuning layer that generates constraint representations from internal representations;

generating the simulation according to the supply chain network graph, the classifications of the locations, and the agent rule models, and the one or more constraints; and,

receiving, as output from the transformer network, one or more constraint representations for the natural language input.