# US 10,481,744 B2 · Computerized methods and computer apparatus for determining the values of physical quantities in problems governed by Poisson or Helmholtz equations, and method for fabricating an electronic component

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

- Number: US 10,481,744 B2
- Document ID: US10481744B2
- Status: Granted
- Kind: B2
- Application number: 15/927,452
- Filed: 2018-03-21
- Published: 2018-09-27
- Granted: 2019-11-19
- Earliest claimed: 2017-03-21
- Assignees: Fieldscale
- Applicants: Fieldscale
- Inventors: Georgios Bouzianas; Nikolaos Zikos; Nikolaos Sismanis
- Web page: https://patentgrep.com/patent/US10481744B2

## CPC

- G06F 3/044: by capacitive means
- G03G 15/00: Apparatus for electrographic processes using a charge pattern (G03G16/00, G03G17/00 take precedence)
- G06F 17/12: Simultaneous equations {, e.g. systems of linear equations}
- G06F 17/16: Matrix or vector computation {, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization (matrix transposition G06F7/78)}
- G06F 2203/04103: Manufacturing, i.e. details related to manufacturing processes specially suited for touch sensitive devices
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G03G: ELECTROGRAPHY; ELECTROPHOTOGRAPHY; MAGNETOGRAPHY

## Abstract

A computer system and method are proposed for solving physical problems governed by Poisson or Helmholtz equations. Values of at least one unknown physical quantity at points are determined using known values of at least one second physical quantity. The points are clustered into a hierarchical skeleton structure using a clustering algorithm based on distances between the points. Interactions between distant clusters are approximated, to give a system of equations which can be solved more efficiently.

## Claims

### Claim 1

A method of fabricating an electronic component, the method comprising:
(a) forming a numerical model of physical structure of the electronic component in a three-dimensional space, the numerical model comprising, for each of a plurality of discretization points in the three-dimensional space, a known value of an electrical potential of the electronic component at the discretization point, an unknown electrical charge distribution at the discretization points being related to the electrical potential of the discretization points, by a Poisson equation expressed via an interaction matrix A having elements which correspond to respective pairs of the discretization points,
(b) repeatedly applying a clustering algorithm, to classify the discretization points into a hierarchical tree structure having a plurality L+1 of levels, labelled by an integer index l=0, . . . ,L,
where level l=0 of the tree structure is all the discretization points, level l=1 of the tree structure is composed of non-overlapping clusters of the discretization points, and
for l=2, . . . ,L, level 1 of the tree structure is composed of, for each cluster of the (l−1)-th level, a respective plurality of non-overlapping clusters of the discretization points of that cluster of the (l−1)-th level,
the clustering algorithm defining the clusters by minimizing distances between discretization points of each cluster to a respective center of that cluster, and maximizing distances between points of each cluster to the centers of other clusters;
(c) using the hierarchical tree structure to form a compressed interaction matrix which is expressed recursively according to the levels;
(d) determining the electrical charge distribution of the discretization points from the values of the electrical potential at the discretization points and the compressed interaction matrix;
(e) determining whether the determined electrical charge distribution meets a criterion; and
(f) determining that the electronic component is to be fabricated according to the numerical model when the electrical charge distribution meets the criterion, or that the electronic component is not to be fabricated according to the numerical model when the electrical charge distribution does not meet the criterion.

### Claim 2 (depends on 1)

The method of claim 1 in which the electronic component is a capacitive touch sensor.

### Claim 3 (depends on 1)

The method of claim 1 in which the step (d) of determining the electrical charge distribution is performed by:
forming an inverted matrix from the compressed interaction matrix; and
multiplying the inverted matrix by a vector of values of the electrical potential at the discretization points.

### Claim 4 (depends on 1)

The method of claim 1 in which the step (d) of determining the electrical charge distribution is performed iteratively by repeatedly:
multiplying the compressed interaction matrix by a vector formed of preliminary estimates of the electrical charge distribution at each of the discretization points,
forming residual values from the result of matrix multiplication and the values of the electrical potential at the discretization points; and
improving the preliminary estimates of the first physical quantity using the residual values.

### Claim 5 (depends on 1)

The method of claim 1 comprising a step, after step (c) and prior to step (d), of determining whether the compressed interaction matrix meets a criterion, and according to whether the determination is positive, performing said step (d) by a process comprising inverting the compressed interaction matrix, or determining the electrical charge distribution iteratively by repeatedly improving successive preliminary estimates of the electrical charge distribution at the discretization points using the result of a matrix multiplication of the compressed interaction matrix and the preliminary estimate of the electrical charge distribution at the discretization points.

### Claim 6 (depends on 1)

The method of claim 1 in which the clustering is the algorithm of k-means.

### Claim 7 (depends on 1)

The method of claim 1 in which, at each level, the compressed interaction matrix approximates the physical interaction between pairs of discretization points which do not meet a first proximity criterion, the first proximity criterion being based on the cluster in which a first discretization point of the pair of discretization points lies.

### Claim 8 (depends on 7)

The method of claim 7 in which the first proximity criterion is that the second point of the pair of discretization points is within a proxy surface of the cluster in which the first of the discretization points lies.

### Claim 9 (depends on 7)

The method of claim 7 including:
for each cluster defining a respective list of neighboring clusters of the same level which meet a second proximity criterion;
for each cluster, using the respective list of neighboring clusters to identify discretization points of the neighboring clusters meeting the first proximity criterion.