I want to calculate the minimum value.
And Already the function I made is proved as a convex function.
Actually, I am looking for a function such as “fminsearch”.
So I choose the “minimize” function in cvx, but the error comes up.
previous by using fminsearch:
len1 = @(x) ~~~~;
...
len16 = @(x) ~~~~;
myfun = @(x) len1(x)+len2(x)+len3(x)+....+len16(x);
[x,fval] = fminsearch(myfun, [0,0,0,0,0,0,0,0,0,0,0,0]);
-> Bad output
now by using cvx:
variables x1 x2 ..... x12
len1 = norm( using x1~x12);
...
len16 = norm( using x1~x12);
myfun = minimize(len1+len2+....+len16);
-> Error using minimize
Too many output arguments.
Should I make the function as an input for minimize
such as A*x - b ??
Help me…!
Please show exact and complete code of what you tried and the output which was produced, or at least a minimal reproducible example which is complete and illustrates your problem.
why are you trying to assign the output of minimize to a variable? Just do
As mcg wrote, don’t use
tolen(n) = minimize(...);
Get rid of
tolen(n) =
Are you trying to save the optimal objective value each time minimize is called? if so, insert the line
tolen(n) = cvx_optval;
after the minimize statement.
Oh, I tried but error occurs:
`Undefined function or variable 'cvx_optval'.
Sorry. cvx_optval needs to come after cvx_end
If you want to call minimize each time through the nested loops, you need to put cvx_begin, CVX variable declarations, and all expressions (and constraints if you had any) between the cvx_begin and cvx_end, which should all be inside the innermost loop.
Thank you very much.
Now it works and the results seem satisfying.
One more question here:
The cvx output(like the below) is very long and much.(Since there’re more than 1000 iterations…)
I want to save(hide) the outputs in a file in order not to show on the command window.
Calling SDPT3 4.0: 12 variables, 2 equality constraints
------------------------------------------------------------
num. of constraints = 2
dim. of socp var = 12, num. of socp blk = 4
*******************************************************************
SDPT3: Infeasible path-following algorithms
*******************************************************************
version predcorr gam expon scale_data
NT 1 0.000 1 0
it pstep dstep pinfeas dinfeas gap prim-obj dual-obj cputime
-------------------------------------------------------------------
0|0.000|0.000|3.8e-01|6.0e+00|2.1e+01| 0.000000e+00 0.000000e+00| 0:0:00| chol 1 1
1|1.000|0.997|4.6e-09|1.2e-01|3.9e-01| 0.000000e+00 5.244599e-03| 0:0:00| chol 1 1
2|1.000|1.000|6.8e-08|1.0e-02|2.8e-02| 0.000000e+00 7.256085e-04| 0:0:00| chol 1 1
3|1.000|1.000|3.2e-08|1.0e-03|2.2e-03| 0.000000e+00 1.046287e-04| 0:0:00| chol 1 1
4|1.000|1.000|1.1e-08|1.0e-04|1.8e-04| 0.000000e+00 1.404183e-05| 0:0:00| chol 1 1
5|1.000|1.000|1.6e-10|1.0e-05|1.4e-05| 0.000000e+00 1.779475e-06| 0:0:00| chol 1 1
6|1.000|0.989|1.1e-11|1.1e-07|1.5e-07| 0.000000e+00 1.939589e-08| 0:0:00| chol 1 1
7|1.000|0.988|6.3e-14|1.3e-09|1.9e-09| 0.000000e+00 1.779273e-10| 0:0:00|
stop: max(relative gap, infeasibilities) < 1.49e-08
-------------------------------------------------------------------
number of iterations = 7
primal objective value = 0.00000000e+00
dual objective value = 1.77927349e-10
gap := trace(XZ) = 1.91e-09
relative gap = 1.91e-09
actual relative gap = -1.78e-10
rel. primal infeas (scaled problem) = 6.33e-14
rel. dual " " " = 1.32e-09
rel. primal infeas (unscaled problem) = 0.00e+00
rel. dual " " " = 0.00e+00
norm(X), norm(y), norm(Z) = 1.6e+00, 2.9e-10, 1.4e-09
norm(A), norm(b), norm(C) = 3.8e+00, 1.6e+00, 1.0e+00
Total CPU time (secs) = 0.06
CPU time per iteration = 0.01
termination code = 0
DIMACS: 6.3e-14 0.0e+00 1.3e-09 0.0e+00 -1.8e-10 1.9e-09
-------------------------------------------------------------------
------------------------------------------------------------
Status: Solved
Optimal value (cvx_optval): +0
You can use
cvx_begin quiet
in order to suppress all the CVX output. You can rely on MATLAB inspection in your code and/or saving the values of the CVX variable output values, cvx_optval, and cvx_status to assess the results of each CVX invocation. You can use the MATLAB diary command if you want to save a copy of all the CVX output (without using quiet option).
If you have 1296 optimization problems to solve, then you need to have 1296 occurrences of cvx_begin … cvx_end.
Are you getting the correct answer as far as you know? if so, it is a question of run time. You can try moving everything possible out of the for loops, but perhaps yuu have already done that. If so, if you want to speed things up, you may have to use a different tool which doesn’t incur CVX’s overhead
Fortunately, I am getting the correct answer as far as I know.
And it takes less than 10 minutes for 1296 iteration.
I already used
- fminsearch : wrong results / more than 30 minutes.
- global search : correct results / more than 16 hours.
Compare with the above two tools, cvx tool is very strong and efficient I think…
Thank you very much for your help:)
I am really enjoying this process.
