检查或修改优化问题
使用 show 或 write 检查问题
创建优化问题后，您可以使用 show 检查其表示。对于较大的问题，请改用 write。例如，
prob = optimproblem;
x = optimvar('x',2,'LowerBound',0);
prob.Objective = x(1) - 2*x(2);
prob.Constraints.cons1 = x(1) + 2*x(2) <= 4;
prob.Constraints.cons2 = -x(1) + x(2) <= 1;
show(prob) OptimizationProblem :
Solve for:
x
minimize :
x(1) - 2*x(2)
subject to cons1:
x(1) + 2*x(2) <= 4
subject to cons2:
-x(1) + x(2) <= 1
variable bounds:
0 <= x(1)
0 <= x(2)此次检查展示了问题的基本要素，例如问题是求最小化还是最大化，以及变量边界。检查显示变量中使用的索引名称（如果有）。检查并未表明变量是否为整数值。
更改默认求解器或选项
为了尝试改进解或解速度，请检查并更改默认求解器或选项。
要查看默认求解器和选项，请使用 optimoptions(prob)。例如，
rng default
x = optimvar('x',3,'LowerBound',0);
expr = sum((rand(3,1).*x).^2);
prob = optimproblem('Objective',expr);
prob.Constraints.lincon = sum(sum(randn(size(x)).*x)) <= randn;
options = optimoptions(prob)options =
lsqlin options:
Options used by current Algorithm ('interior-point'):
(Other available algorithms: 'trust-region-reflective')
Set properties:
No options set.
Default properties:
Algorithm: 'interior-point'
ConstraintTolerance: 1.0000e-08
Display: 'final'
LinearSolver: 'auto'
MaxIterations: 200
OptimalityTolerance: 1.0000e-08
StepTolerance: 1.0000e-12
Show options not used by current Algorithm ('interior-point')该问题的默认求解器是 lsqlin，您可以看到默认选项。
要更改求解器，请在 'Solver' 中设置 solve 名称-值对。要查看不同求解器适用的选项，请使用 optimoptions 将当前选项传递给不同的求解器。例如，继续这个问题，
options = optimoptions('quadprog',options)options =
quadprog options:
Options used by current Algorithm ('interior-point-convex'):
(Other available algorithms: 'trust-region-reflective')
Set properties:
ConstraintTolerance: 1.0000e-08
MaxIterations: 200
OptimalityTolerance: 1.0000e-08
StepTolerance: 1.0000e-12
Default properties:
Algorithm: 'interior-point-convex'
Display: 'final'
LinearSolver: 'auto'
Show options not used by current Algorithm ('interior-point-convex')要更改选项，请使用 optimoptions 或点符号来设置选项，并将选项传递给 solve 名称-值对中的 'Options'。请参阅常用的优化选项：调整和故障排除。继续这个示例，
options.Display = 'iter'; sol = solve(prob,'Options',options,'Solver','quadprog');
Iter Fval Primal Infeas Dual Infeas Complementarity
0 1.500359e+00 3.068423e-01 2.275437e+00 2.500000e-01
1 1.728717e-01 0.000000e+00 7.719860e-03 3.637874e-02
2 2.604108e-02 0.000000e+00 0.000000e+00 5.245260e-03
3 7.822161e-03 0.000000e+00 2.775558e-17 1.407915e-03
4 2.909218e-03 0.000000e+00 6.938894e-18 2.070784e-04
5 1.931264e-03 0.000000e+00 1.734723e-18 2.907724e-05
6 1.797508e-03 0.000000e+00 2.602085e-18 4.083167e-06
7 1.775398e-03 0.000000e+00 4.336809e-19 5.102453e-07
8 1.772971e-03 0.000000e+00 2.632684e-19 3.064243e-08
9 1.772848e-03 0.000000e+00 5.228973e-19 4.371356e-11
Minimum found that satisfies the constraints.
Optimization completed because the objective function is non-decreasing in
feasible directions, to within the value of the optimality tolerance,
and constraints are satisfied to within the value of the constraint tolerance.纠正错误指定的问题
要检查您的问题是否正确，请检查其所有方面。例如，通过运行此脚本创建一个优化问题来求解数独问题。
x = optimvar('x',9,9,9,'LowerBound',0,'UpperBound',1);
cons1 = sum(x,1) == 1;
cons2 = sum(x,2) == 1;
cons3 = sum(x,3) == 1;
prob = optimproblem;
prob.Constraints.cons1 = cons1;
prob.Constraints.cons2 = cons2;
prob.Constraints.cons3 = cons3;
mul = ones(1,1,9);
mul = cumsum(mul,3);
prob.Objective = sum(sum(sum(x,1),2).*mul);
cons4 = optimconstr(3,3,9);
for u = 1:3
for v = 1:3
arr = x(3*(u-1)+1:3*(u-1)+3,3*(v-1)+1:3*(v-1)+3,:);
cons4(u,v,:) = sum(sum(arr,1),2) <= ones(1,1,9);
end
end
prob.Constraints.cons4 = cons4;
B = [1,2,2;
1,5,3;
1,8,4;
2,1,6;
2,9,3;
3,3,4;
3,7,5;
4,4,8;
4,6,6;
5,1,8;
5,5,1;
5,9,6;
6,4,7;
6,6,5;
7,3,7;
7,7,6;
8,1,4;
8,9,8;
9,2,3;
9,5,4;
9,8,2];
for u = 1:size(B,1)
x.LowerBound(B(u,1),B(u,1),B(u,1)) = 1;
end该脚本有一些错误，您可以通过检查变量、目标和约束来发现这些错误。首先，检查变量 x。
x
x =
9×9×9 OptimizationVariable array with properties:
Array-wide properties:
Name: 'x'
Type: 'continuous'
IndexNames: {{} {} {}}
Elementwise properties:
LowerBound: [9×9×9 double]
UpperBound: [9×9×9 double]
See variables with show.
See bounds with showbounds.该显示表明变量的类型是连续的。该变量应为整数值。更改类型。
x.Type = 'integer'
x =
9×9×9 OptimizationVariable array with properties:
Array-wide properties:
Name: 'x'
Type: 'integer'
IndexNames: {{} {} {}}
Elementwise properties:
LowerBound: [9×9×9 double]
UpperBound: [9×9×9 double]
See variables with show.
See bounds with showbounds.检查边界。应该有 21 个下界，值为 1，B 的每一行一个。因为 x 是一个大数组，所以将边界写入文件而不是在命令行上显示它们。
writebounds(x,'xbounds.txt')在文件 xbounds.txt 中搜索 1 <= 的所有实例。在变量 x(1,1,1)、x(2,2,2)、…、x(9,9,9) 中，只有 9 个下界的值为 1。为了调查这种差异，请检查设置下界的代码：
for u = 1:size(B,1)
x.LowerBound(B(u,1),B(u,1),B(u,1)) = 1;
end循环内的行应该说 x.LowerBound(B(u,1),B(u,2),B(u,3)) = 1;。将所有下界重置为零，然后运行更正后的代码。
x.LowerBound = 0;
for u = 1:size(B,1)
x.LowerBound(B(u,1),B(u,2),B(u,3)) = 1;
end
writebounds(x,'xbounds.txt')xbounds.txt 现在具有正确的下界数量，即 1。
检查目标函数。目标函数表达式很大，因此将表达式写入文件中。
write(prob.Objective,'objectivedescription.txt') x(1, 1, 1) + x(2, 1, 1) + x(3, 1, 1) + x(4, 1, 1) + x(5, 1, 1) + x(6, 1, 1) + x(7, 1, 1) + x(8,
1, 1) + x(9, 1, 1) + x(1, 2, 1) + x(2, 2, 1) + x(3, 2, 1) + x(4, 2, 1) + x(5, 2, 1) + x(6, 2,
...
9*x(7, 8, 9) + 9*x(8, 8, 9) + 9*x(9, 8, 9) + 9*x(1, 9, 9) + 9*x(2, 9, 9) + 9*x(3, 9, 9) +
9*x(4, 9, 9) + 9*x(5, 9, 9) + 9*x(6, 9, 9) + 9*x(7, 9, 9) + 9*x(8, 9, 9) + 9*x(9, 9, 9)目标函数看起来合理，因为它是标量表达式的总和。
将约束写入文件以供检查。
write(prob.Constraints.cons1,'cons1.txt') write(prob.Constraints.cons2,'cons2.txt') write(prob.Constraints.cons3,'cons3.txt') write(prob.Constraints.cons4,'cons4.txt')
检查 cons4.txt，您会发现一个错误。所有的约束都是不等式而不是等式。更正创建此约束的代码行，并将更正后的约束放入问题中。
cons4 = optimconstr(3,3,9);
for u = 1:3
for v = 1:3
arr = x(3*(u-1)+1:3*(u-1)+3,3*(v-1)+1:3*(v-1)+3,:);
cons4(u,v,:) = sum(sum(arr,1),2) == ones(1,1,9);
end
end
prob.Constraints.cons4 = cons4;经过这些改变之后，您就可以成功求解问题了。
sol = solve(prob);
x = round(sol.x);
y = ones(size(x));
for k = 2:9
y(:,:,k) = k; % multiplier for each depth k
end
S = x.*y; % multiply each entry by its depth
S = sum(S,3); % S is 9-by-9 and holds the solved puzzle
drawSudoku(S)变量名称重复
如果您重新创建一个变量，但已经有一个使用旧变量的表达式，那么在将表达式合并到单个问题中时可能会出现错误。请参阅变量名称不允许重复。
另请参阅
show | showbounds | write | writebounds | OptimizationConstraint | OptimizationVariable | OptimizationProblem | OptimizationExpression
