Main Content

pdeCoefficientsToDouble

Convert symbolic PDE coefficients to double format

Since R2021a

Description

example

coeffs = pdeCoefficientsToDouble(symCoeffs) converts the symbolic objects of the structure symCoeffs to double-precision numbers or function handles. The output is a structure coeffs that can then be used to define the coefficients of a PDE model by calling specifyCoefficients in Partial Differential Equation Toolbox™.

The structure coeffs contains the coefficients m, d, c, a, and f for the PDE system with the form

m2ut2+dut·(cu)+au=f

that can be solved in Partial Differential Equation Toolbox. For more information, see Equations You Can Solve Using PDE Toolbox (Partial Differential Equation Toolbox).

example

coeffs = pdeCoefficientsToDouble(symCoeffs,u) converts the symbolic objects of the structure symCoeffs to double-precision numbers or function handles for the coefficients of a PDE system with the dependent variables u.

Examples

collapse all

Create a symbolic PDE that represents the Laplacian of the variable u(x,y).

syms u(x,y) f
pdeeq = laplacian(u,[x y]) == f
pdeeq(x, y) = 

2x2 u(x,y)+2y2 u(x,y)=f

Extract the coefficients of the PDE as symbolic expressions and display their values.

symCoeffs = pdeCoefficients(pdeeq,u,'Symbolic',true);
structfun(@disp,symCoeffs)
0
0

(-100-1)

f
0

pdeCoefficients converts the symbolic PDE into a scalar PDE equation of the form

m2ut2+dut-(cu)+au=f,

and extract the coefficients a, c, m, d, and f into the structure symCoeffs.

Choose a value for f. Since symCoeffs are symbolic objects, use pdeCoefficientsToDouble to convert the coefficients to double data type. The coefficients with double data type are valid inputs for the specifyCoefficients function in the PDE Toolbox.

symCoeffs = subs(symCoeffs,f,-3);
coeffs = pdeCoefficientsToDouble(symCoeffs)
coeffs = struct with fields:
    a: 0
    c: [4x1 double]
    m: 0
    d: 0
    f: -3

Solve a system of two second-order PDEs. You can solve the PDE system by extracting the PDE coefficients symbolically using pdeCoefficients, converting the coefficients to double-precision numbers using pdeCoefficientsToDouble, and specifying the coefficients in the PDE model using specifyCoefficients.

The system of PDEs represents the deflection of a clamped structural plate under a uniform pressure load. The system of PDEs with the dependent variables u1 and u2 is given by

-2u1+u2=0,

-D2u2=p,

where D is the bending stiffness of the plate given by

D=Eh312(1-ν2),

and E is the modulus of elasticity, ν is Poisson's ratio, h is the plate thickness, u1 is the transverse deflection of the plate, and p is the pressure load.

Create a PDE model for the system of two equations.

model = createpde(2);

Create a square geometry. Specify the side length of the square. Then include the geometry in the PDE model.

len = 10.0;
gdm = [3 4 0 len len 0 0 0 len len]';
g = decsg(gdm,'S1',('S1')');
geometryFromEdges(model,g);

Specify the values of the physical parameters of the system. Let the external pressure p be a symbolic variable pres that can take any value.

E = 1.0e6;
h_thick = 0.1;
nu = 0.3;
D = E*h_thick^3/(12*(1 - nu^2));
syms pres

Declare the PDE system as a system symbolic equations. Extract the coefficients of the PDE and return them in symbolic form.

syms u1(x,y) u2(x,y)
pdeeq = [-laplacian(u1) + u2; -D*laplacian(u2) - pres];
symCoeffs = pdeCoefficients(pdeeq,[u1 u2],'Symbolic',true)
symCoeffs = struct with fields:
    m: 0
    a: [2x2 sym]
    c: [4x4 sym]
    f: [2x1 sym]
    d: 0

Display the coefficients m, a, c, f, and d.

structfun(@disp,symCoeffs)
0

(0100)

(100001000025000273000025000273)

(0pres)

0

Substitute a value for pres using the subs function. Since the outputs of subs are symbolic objects, use the pdeCoefficientsToDouble function to convert the coefficients to the double data type, which makes them valid inputs for the PDE Toolbox.

symCoeffs = subs(symCoeffs,pres,2);
coeffs = pdeCoefficientsToDouble(symCoeffs)
coeffs = struct with fields:
    a: [4x1 double]
    c: [16x1 double]
    m: 0
    d: 0
    f: [2x1 double]

Specify the PDE coefficients for the PDE model.

specifyCoefficients(model,'m',coeffs.m,'d',coeffs.d, ...
    'c',coeffs.c,'a',coeffs.a,'f',coeffs.f);

Specify spring stiffness. Specify boundary conditions by defining distributed springs on all four edges.

k = 1e7;
bOuter = applyBoundaryCondition(model,'neumann','Edge',(1:4), ...
    'g',[0 0],'q',[0 0; k 0]);

Specify the mesh size of the geometry and generate a mesh for the PDE model.

hmax = len/20;
generateMesh(model,'Hmax',hmax);

Solve the model.

res = solvepde(model);

Access the solution at the nodal locations.

u = res.NodalSolution;

Plot the transverse deflection of the plate.

numNodes = size(model.Mesh.Nodes,2);
figure;
pdeplot(model,'XYData',u(1:numNodes),'contour','on')
title 'Transverse Deflection'

Figure contains an axes object. The axes object with title Transverse Deflection contains 12 objects of type patch, line.

Find the transverse deflection at the plate center.

wMax = min(u(1:numNodes,1))
wMax = -0.2763

Compare the result with the deflection at the plate center computed analytically.

pres = 2;
wMax = -.0138*pres*len^4/(E*h_thick^3)
wMax = -0.2760

Since R2023a

Create a PDE system of four equations with four dependent variables, A, E, P, and T.

syms A(x,y,z,t) E(x,y,z,t) P(x,y,z,t) T(x,y,z,t)
eqn1 = diff(A,t) == -A*E*(-exp(T));
eqn2 = diff(P,t) == -P*E*(-exp(T));
eqn3 = diff(E,t) == -A*E*(-exp(T)) - P*E*(-exp(T));
eqn4 = diff(T,t) == laplacian(T,[x y z]) + A*E*(-exp(T)) + P*E*(-exp(T));
pdeeq = [eqn1; eqn2; eqn3; eqn4]
pdeeq(x, y, z, t) = 

(t A(x,y,z,t)=σ2t P(x,y,z,t)=σ1t E(x,y,z,t)=σ2+σ1t T(x,y,z,t)=2x2 T(x,y,z,t)+2y2 T(x,y,z,t)+2z2 T(x,y,z,t)-σ2-σ1)where  σ1=eT(x,y,z,t)E(x,y,z,t)P(x,y,z,t)  σ2=eT(x,y,z,t)A(x,y,z,t)E(x,y,z,t)

Extract the coefficients of the PDE system as symbolic expressions. Specify the variable u to represent the dependent variables.

u = [A E P T];
symCoeffs = pdeCoefficients(pdeeq,u,'Symbolic',true)
symCoeffs = struct with fields:
    m: [4x4 sym]
    a: [4x4 sym]
    c: [12x12 sym]
    f: [4x1 sym]
    d: [4x4 sym]

Display the coefficients m, a, c, f, and d.

structfun(@disp,symCoeffs)

(0000000000000000)

(-σ20000σ100-σ2σ100σ2eT(x,y,z,t)P(x,y,z,t)00)where  σ1=-eT(x,y,z,t)P(x,y,z,t)  σ2=eT(x,y,z,t)E(x,y,z,t)

(000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000100000000000010000000000001)

(0000)

(1000001001000001)

Convert the coefficients to the double data type so that they are valid inputs for the specifyCoefficients function in Partial Differential Equation Toolbox. Because the a coefficient in symCoeffs contains the dependent variables and is not constant, calling pdeCoefficientsToDouble(symCoeffs) without the second input argument can return an error. Instead, specify the second argument as the dependent variables u when using pdeCoefficientsToDouble.

coeffs = pdeCoefficientsToDouble(symCoeffs,u)
coeffs = struct with fields:
    a: @makeCoefficient/coefficientFunction
    c: [144x1 double]
    m: 0
    d: [16x1 double]
    f: 0

Input Arguments

collapse all

Coefficients of PDE in symbolic form, specified as a structure of symbolic expressions.

Since R2023a

Dependent variables of PDE, specified as a symbolic function. The argument u must contain stationary or time-dependent variables in two or three dimensions.

Example: syms E(x,y,z,t) B(x,y,z,t); u = [E B];

Output Arguments

collapse all

Coefficients of PDE operating on doubles, returned as a structure of double-precision numbers and function handles as required by the specifyCoefficients function. The fields of the structure are a, c, m, d, and f. For details on interpreting the coefficients in the format required by Partial Differential Equation Toolbox, see:

When pdeCoefficientsToDouble returns a coefficient as a function handle, the function handle takes two structures as input arguments, location and state, and returns double-precision output. The function handle is displayed as @makeCoefficient/coefficientFunction in the Command Window. To display the formula of the function handle in coeffs.f in symbolic form, use coeffs.f('show').

In some cases, not all generated coefficients can be used by specifyCoefficients. For example, the d coefficient must take a special matrix form when m is nonzero. For more details, see d Coefficient When m Is Nonzero (Partial Differential Equation Toolbox).

Version History

Introduced in R2021a

expand all