# US 2026/0187310 A1 · Simulation of an electronic block

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2026/0187310 A1
- Document ID: US20260187310A1
- Status: Published application
- Kind: A1
- Application number: 18/761,144
- Filed: 2023-04-17
- Published: 2026-07-02
- Earliest claimed: 2022-04-18
- Applicants: SRI International
- Inventors: John Byrnes; Mike Piacentino; David Zhang; Indhumathi Kandaswamy; Gooitzen van der Wal
- Web page: https://patentgrep.com/patent/US20260187310A1

## CPC

- G06F 30/27: using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
- G06N 3/04: Architecture, e.g. interconnection topology
- G06F 30/32: Circuit design at the digital level (reconfigurable circuits G06F30/34)
- G06F 30/3323: using formal methods, e.g. equivalence checking or property checking
- G06N 3/084: Backpropagation, e.g. using gradient descent
- G06N 3/0895: Weakly supervised learning, e.g. semi-supervised or self-supervised learning
- G06F 2111/10: Numerical modelling
- G06F 2117/08: HW-SW co-design, e.g. HW-SW partitioning
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

A model generator receives inputs of formal descriptions of an expected behavior of an electronic block. The model generator generates one or more models. Each generated model has a deep learning architecture to simulate an electronic block based upon the formal descriptions of an expected behavior of the electronic block. The model generator can then train each model with the deep learning architecture to simulate the electronic block using the formal descriptions of an expected behavior of that electronic block as logical reasoning to constrain machine learning of that model from training data during training of that model to simulate the electronic block to be consistent with the logical reasoning.

## Claims

### Claim 1

An apparatus, comprising:
a model generator with one or more inputs configured to receive inputs of formal descriptions of an expected behavior of an electronic block, where the model generator is configured to generate one or more models, where each generated model has a deep learning architecture to simulate an electronic block based upon the formal descriptions of an expected behavior of the electronic block, where the electronic block consists of at least one of i) an electronic component, ii) an electronic circuit, iii) an electronic subsystem, iv) a system on a chip, and v) an integrated circuit chip, where any portion of the model generator and generated models that are implemented with software, then the software is stored on one or more non-transitory machine readable mediums and are to be executed by one or more processors.

### Claim 2 (depends on 1)

The apparatus of claim 1, where the deep learning architecture created in the model is a neural network constructed and trained with training data to simulate the electronic block.

### Claim 3 (depends on 1)

The apparatus of claim 1, where the electronic block to be simulated is a representation of an individual electronic circuit within a system that contains a device under test to be tested and verified in an electronic design automation testing environment.

### Claim 4 (depends on 1)

The apparatus of claim 1, where the model generator is configured to construct the deep learning architecture to simulate the electronic block based on the formal descriptions of the expected behavior rather than needing a formal specification on a full architecture making up the electronic block, which is configured to cause a verification of a device under test to occur in less time than the verification of the device under test would take with the full architecture making up the electronic block.

### Claim 5 (depends on 1)

The apparatus of claim 1, where the model generator is further configured to train the model with the deep learning architecture to simulate the electronic block using the formal descriptions of an expected behavior of that electronic block as logical reasoning to constrain machine learning of that model from training data during training of that model to simulate the electronic block to be consistent with the logical reasoning.

### Claim 6 (depends on 5)

The apparatus of claim 5, where the model generator is configured to generate the one or more models, where the models have been trained to simulate corresponding specific electronic blocks to cooperate with an electronic design automation tool.

### Claim 7 (depends on 5)

The apparatus of claim 5, where the model generator is configured to cooperate with an electronic design automation tool to test and verify a device under test in a system containing the device under test and the one or more models that have been trained to simulate the corresponding specific electronic blocks.

### Claim 8 (depends on 1)

The apparatus of claim 1, where the device under test is simulated at RTL and the one or more generated models that each have the deep learning architecture to simulate the electronic block are simulated at a level of abstraction above RTL.

### Claim 9 (depends on 1)

The apparatus of claim 1, where the model generator further comprises:
a theory module configured to receive as input the formal descriptions of the expected behavior of that electronic block expressed in first order logic elements and then encode and send the encoded formal descriptions of the expected behavior of that electronic block to a model building module in the model generator to use machine learning to discover distributed vector representations of a meaning associated with information in the formal descriptions of the expected behavior and generate the deep learning architecture.

### Claim 10 (depends on 5)

The apparatus of claim 5, where the model generator is configured to train the deep learning architecture to a threshold accuracy that agrees with the formal descriptions of the expected behavior of that electronic block with a mixture of labeled training data examples and unlabeled training data examples, where an amount of unlabeled training data examples during the training to simulate the specific electronic block is at least twice the amount of labeled training data examples needed to train the deep learning architecture.

### Claim 11 (depends on 5)

The apparatus of claim 5, where the constructed deep learning architecture is configured to learn from a combination of labeled training data along with unlabeled training data during the training.

### Claim 12 (depends on 5)

The apparatus of claim 5, where the model generator is configured to backpropagate loss based on failure to satisfy the formal descriptions of the expected behavior during the training of the deep learning architecture, which allows the deep learning architecture to generalize beyond a set of test vectors that may be available.

### Claim 13

A non-transitory machine-readable medium, which stores further instructions in the executable format by the one or more processors to cause operations as follows, comprising:
using a model generator with one or more inputs to receive inputs of formal descriptions of an expected behavior of an electronic block, where the model generator is configured to generate one or more models, where each generated model has a deep learning architecture to simulate an electronic block based upon the formal descriptions of an expected behavior of the electronic block, where the electronic block consists of at least one of i) an electronic component, ii) an electronic circuit, iii) an electronic subsystem, iv) a system on a chip, and v) an integrated circuit chip.

### Claim 14 (depends on 13)

The non-transitory machine-readable medium of claim 13, where the deep learning architecture created in the model is a neural network constructed and trained with training data to simulate the electronic block.

### Claim 15 (depends on 13)

The non-transitory machine-readable medium of claim 13, where the electronic block to be simulated is a representation of an individual electronic circuit within a system that contains a device under test to be tested and verified in an electronic design automation testing environment.

### Claim 16 (depends on 13)

The non-transitory machine-readable medium of claim 13, further comprising:
using the model generator to construct the deep learning architecture to simulate the electronic block based on the formal descriptions of the expected behavior rather than needing a formal specification on a full architecture making up the electronic block, which causes a verification of a device under test to occur in less time than the verification of the device under test would take with the full architecture making up the electronic block.

### Claim 17 (depends on 13)

The non-transitory machine-readable medium of claim 13, further comprising:
using the model generator to train the model with the deep learning architecture to simulate the electronic block using the formal descriptions of an expected behavior of that electronic block as logical reasoning to constrain machine learning of that model from training data during training of that model to simulate the electronic block to be consistent with the logical reasoning.

### Claim 18 (depends on 17)

The non-transitory machine-readable medium of claim 17, further comprising:
using the model generator to generate the one or more models, where the models have been trained to simulate corresponding specific electronic blocks to cooperate with an electronic design automation tool.

### Claim 19 (depends on 17)

The non-transitory machine-readable medium of claim 17, further comprising:
using the model generator to cooperate with an electronic design automation tool to test and verify a device under test in a system containing the device under test and the one or more models that have been trained to simulate the corresponding specific electronic blocks, where the device under test is simulated at RTL and the one or more generated models that each have the deep learning architecture to simulate the electronic block are simulated at a level of abstraction above RTL.

### Claim 20

A method for simulating one or more electronic blocks, comprising:
using a model generator with one or more inputs to receive inputs of formal descriptions of an expected behavior of an electronic block, where the model generator is configured to generate one or more models, where each generated model has a deep learning architecture to simulate a first electronic block based upon formal descriptions of the expected behavior of the first electronic block, where the first electronic block consists of at least one of i) an electronic component, ii) an electronic circuit, iii) an electronic subsystem, iv) a system on a chip, and v) an integrated circuit chip.
