Main Content

decsg

Decompose constructive solid 2-D geometry into minimal regions

Description

example

dl = decsg(gd,sf,ns) decomposes the geometry description matrix gd into the geometry matrix dl and returns the minimal regions that satisfy the set formula sf. The name-space matrix ns is a text matrix that relates the columns in gd to variable names in sf.

Typically, you draw a geometry in the PDE Modeler app, then export it to the MATLAB® Command Window by selecting Export Geometry Description, Set Formula, Labels from the Draw menu in the app. The resulting geometry description matrix gd represents the CSG model. decsg analyzes the model and constructs a set of disjointed minimal regions bounded by boundary segments and border segments. This set of minimal regions constitutes the decomposed geometry and allows other Partial Differential Equation Toolbox™ functions to work with the geometry.

Alternatively, you can use the decsg function when creating a geometry without using the app. See 2-D Geometry Creation at Command Line for details.

To return all minimal regions (sf corresponds to the union of all shapes in gd), use the shorter syntax dl = decsg(gd).

example

[dl,bt] = decsg(___) returns a Boolean table (matrix) that relates the original shapes to the minimal regions. A column in bt corresponds to the column with the same index in gd. A row in bt corresponds to the index of a minimal region. You can use bt to remove boundaries between subdomains.

Examples

collapse all

Create a 2-D geometry in the PDE Modeler app, then export it to the MATLAB workspace and decompose it to minimal regions by using decsg.

Start the PDE Modeler app and draw a unit circle and a unit square.

pdecirc(0,0,1)
pderect([0 1 0 1])

Enter C1-SQ1 in the Set formula field.

Export the geometry description matrix, set formula, and name-space matrix to the MATLAB workspace by selecting the Export Geometry Description option from the Draw menu.

Decompose the exported geometry into minimal regions. The result is one minimal region with five edge segments: three circle edge segments and two line edge segments.

dl = decsg(gd,sf,ns)
dl =
     2.0000   2.0000    1.0000    1.0000    1.0000
          0        0   -1.0000    0.0000    0.0000
     1.0000        0    0.0000    1.0000   -1.0000
          0   1.0000   -0.0000   -1.0000    1.0000
          0        0   -1.0000         0   -0.0000
          0        0    1.0000    1.0000    1.0000
     1.0000   1.0000         0         0         0
          0        0         0         0         0
          0        0         0         0         0
          0        0    1.0000    1.0000    1.0000

View the geometry. Display the edge labels and the face labels.

pdegplot(dl,"EdgeLabels","on","FaceLabels","on")
axis equal

Geometry consisting of a unit circle with the first quadrant removed

For comparison, decompose the same geometry without specifying the set formula sf and the name-space matrix ns. This syntax returns the union of all shapes in the geometry gd.

dl_all = decsg(gd)
dl_all =
    2.0000    2.0000    2.0000    2.0000    1.0000    1.0000    1.0000    1.0000
         0    1.0000    1.0000         0   -1.0000    0.0000    1.0000    0.0000
    1.0000    1.0000         0         0    0.0000    1.0000    0.0000   -1.0000
         0         0    1.0000    1.0000   -0.0000   -1.0000         0    1.0000
         0    1.0000    1.0000         0   -1.0000         0    1.0000   -0.0000
    3.0000    2.0000    2.0000    3.0000    1.0000    1.0000    3.0000    1.0000
    1.0000         0         0    1.0000         0         0    2.0000         0
         0         0         0         0         0         0         0         0
         0         0         0         0         0         0         0         0
         0         0         0         0    1.0000    1.0000    1.0000    1.0000

View the resulting geometry.

pdegplot(dl_all,"EdgeLabels","on","FaceLabels","on")
axis equal

Geometry consisting of a unit circle and a unit square. The geometry has three faces. Face 1 is the part of the circle without the first quadrant. Face 2 is the part of the square located outside the unit circle. Face 3 is the first quadrant of the circle.

Start the PDE Modeler app and draw a unit circle and a unit square.

pdecirc(0,0,1) 
pderect([0 1 0 1])

Enter C1+SQ1 in the Set formula field.

Export the Geometry Description matrix, set formula, and Name Space matrix to the MATLAB workspace by selecting the Export Geometry Description option from the Draw menu.

Decompose the exported geometry into minimal regions. Because the geometry is a union of all regions, C1+SQ1, you can omit the arguments specifying the set formula and name-space matrix when using decsg.

[dl,bt] = decsg(gd)
dl =
    2.0000    2.0000    2.0000    2.0000    1.0000    1.0000    1.0000    1.0000
         0    1.0000    1.0000         0   -1.0000    0.0000    1.0000    0.0000
    1.0000    1.0000         0         0    0.0000    1.0000    0.0000   -1.0000
         0         0    1.0000    1.0000   -0.0000   -1.0000         0    1.0000
         0    1.0000    1.0000         0   -1.0000         0    1.0000   -0.0000
    3.0000    2.0000    2.0000    3.0000    1.0000    1.0000    3.0000    1.0000
    1.0000         0         0    1.0000         0         0    2.0000         0
         0         0         0         0         0         0         0         0
         0         0         0         0         0         0         0         0
         0         0         0         0    1.0000    1.0000    1.0000    1.0000

bt =
     1     0
     0     1
     1     1

View the geometry. Display the edge labels and the face labels.

pdegplot(dl,"EdgeLabels","on","FaceLabels","on")
axis equal

Geometry consisting of a unit circle and a unit square. The geometry has three faces. Face 1 is the part of the circle without the first quadrant. Face 2 is the part of the square located outside the unit circle. Face 3 is the first quadrant of the circle.

Remove the subdomain boundaries by using the csgdel function.

[dl2,bt2] = csgdel(dl,bt);

View the resulting geometry.

figure
pdegplot(dl2,"EdgeLabels","on","FaceLabels","on")
axis equal

Geometry consisting of a unit circle and a unit square fused together into one face

Input Arguments

collapse all

Geometry description matrix, specified as a matrix of double-precision numbers. The number of columns corresponds to the number of shapes used to construct the geometry. Each column in the geometry description matrix corresponds to a shape in the CSG model. The model supports four types of shapes:

  • For a circle, the first row contains 1. The second and third rows contain the x- and y-coordinates of the center. The fourth row contains the radius of the circle. The radius must be a positive value.

  • For a polygon, the first row contains 2. The second row contains n, which is the number of line segments in the boundary of the polygon. The next n rows contain the x-coordinates of the starting points of the edges, and the n rows after that contain the y-coordinates of the starting points of the edges.

  • For a rectangle, the first row contains 3, and the second row contains 4. The next four rows contain the x-coordinates of the starting points of the edges, and the four rows after that contain the y-coordinates of the starting points of the edges.

  • For an ellipse, the first row contains 4. The second and third rows contain the x- and y-coordinates of the center. The fourth and fifth rows contain positive values, specifying the lengths of the semiaxes of the ellipse. The sixth row contains the rotational angle of the ellipse, measured in radians.

All shapes in a geometry description matrix have the same number of rows. Elements of the matrix that are not required to encode information about the corresponding shapes must be set to zero.

When you export geometry from the PDE Modeler app by selecting Export Geometry Description, Set Formula, Labels from the Draw menu in the app, you can use any variable name for the exported geometry description matrix in the MATLAB workspace. The default name is gd.

Data Types: double

Set formula, specified as a string scalar, character vector, or numeric vector that can be converted to a valid character vector. Set formula includes the names of shapes, such as C1, SQ2, E3, and the operators +, *, and - corresponding to the set operations union, intersection, and set difference, respectively. The precedence order, from the highest to the lowest, is as follows: -, *, +. You can control the precedence by using parentheses.

When you export geometry from the PDE Modeler app by selecting Export Geometry Description, Set Formula, Labels from the Draw menu in the app, you can use any variable name for the formula in the MATLAB workspace. The default name is sf.

Example: '(SQ1+C1)-C2'

Data Types: char | string | double

Name-space matrix, specified as one of these values:

  • String vector where each entry is a shape name

  • Character matrix where each column is a shape name

  • Cell array of character vectors where each vector is a shape name

  • Nonsparse numeric array equivalent to a character array where each column is a shape name

The number of shape names specified in ns corresponds to the number of shapes specified in gd. For string vectors and cell arrays, the number of shapes equals the number of elements. For character or numeric matrices, the number of shapes equals the number of columns.

When you export geometry from the PDE Modeler app by selecting Export Geometry Description, Set Formula, Labels from the Draw menu in the app, you can use any variable name for the name-space matrix in the MATLAB workspace. The default name is ns. The PDE Modeler app exports name-space matrices as nonsparse numeric arrays.

Data Types: string | cell | char | double

Output Arguments

collapse all

Decomposed geometry matrix, returned as a matrix of double-precision numbers. It contains a representation of the decomposed geometry in terms of disjointed minimal regions constructed by the decsg algorithm. Each edge segment of the decomposed geometry corresponds to a column in dl. Edge segments between minimal regions are border segments. Outer boundaries are boundary segments. In each column, the second and third rows contain the starting and ending x-coordinates. The fourth and fifth rows contain the corresponding y-coordinates. The sixth and seventh rows contain left and right minimal region labels with respect to the direction induced by the start and end points. Circle and ellipse segments are always defined counterclockwise. There are three types of possible edge segments:

  • For circle edge segments, the first row is 1. The eighth and ninth rows contain the coordinates of the center of the circle. The 10th row contains the radius.

  • For line edge segments, the first row is 2.

  • For ellipse edge segments, the first row is 4. The eighth and ninth rows contain the coordinates of the center of the ellipse. The 10th and 11th rows contain the lengths of the semiaxes of the ellipse. The 12th row contains the rotational angle of the ellipse.

The matrix uses the minimum number of rows necessary to encode all of the information. For example, if only line segments appear in the decomposed geometry, then the matrix has seven rows. But if there are also circle segments, then the matrix has ten rows, and the last three rows of the columns corresponding to line segments are filled with zeros.

RowCircleLineEllipse
1124
2Starting x coordinateStarting x coordinateStarting x coordinate
3Ending x coordinateEnding x coordinateEnding x coordinate
4Starting y coordinateStarting y coordinateStarting y coordinate
5Ending y coordinateEnding y coordinateEnding y coordinate
6Region label to left of segment, with direction induced by start and end points (0 is exterior label)Region label to left of segment, with direction induced by start and end points (0 is exterior label)Region label to left of segment, with direction induced by start and end points (0 is exterior label)
7Region label to right of segment, with direction induced by start and end points (0 is exterior label)Region label to right of segment, with direction induced by start and end points (0 is exterior label)Region label to right of segment, with direction induced by start and end points (0 is exterior label)
8x coordinate of circle center0x coordinate of ellipse center
9y coordinate of circle center0y coordinate of ellipse center
10Radius0Length of first semiaxis
1100Length of second semiaxis
1200Angle in radians between x axis and first semiaxis

Data Types: double

Boolean table relating the original shapes to the minimal regions, returned as a matrix of 1s and 0s.

Data Types: double

Limitations

  • In rare cases decsg can error or create an invalid geometry because of the limitations of its algorithm. Such issues can occur when two or more edges of a geometry partially overlap, almost coincide, or are almost tangent.

Tips

  • All shapes must be unique. When checking uniqueness, ellipses with equal semiaxes ignore rotation, and polygons ignore the order of their points.

  • All radii and semiaxis lengths must be positive.

  • Polygons must not be self-intersecting or self-overlapping.

  • Each polygon must be defined with a set of distinct points. Individual polygons cannot reuse points, but different polygons can share points.

  • decsg errors if it cannot evaluate the set formula sf.

Version History

Introduced before R2006a

expand all