Timeline for How to write this objective in CVXPY for quasiconvex programming?
Current License: CC BY-SA 4.0
22 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Dec 1, 2023 at 17:26 | review | Close votes | |||
| Dec 6, 2023 at 3:08 | |||||
| Nov 14, 2023 at 20:59 | history | edited | Uomond | CC BY-SA 4.0 |
found an approximate representation for -log(t)log(1-t)
|
| Nov 6, 2023 at 14:58 | history | edited | Uomond | CC BY-SA 4.0 |
fixed typo
|
| Nov 6, 2023 at 13:55 | comment | added | Uomond | @MarkL.Stone I don't think they are, but it makes sense what you said, oh well | |
| Nov 6, 2023 at 10:28 | comment | added | Mark L. Stone | If I understand correctly, ptp(x) = max(x) -min(x). So I don’t see how ptp could accept a non-affine argument, as you are proposing. because max requires a convex argument and min` requires a concave argument. Am I wrong? cvxpy.org/api_reference/cvxpy.atoms.other_atoms.html#ptp doesn’t indicate argument requirements. Are they documented somewhere? | |
| Nov 6, 2023 at 8:14 | history | edited | Uomond | CC BY-SA 4.0 |
Added some advances on the dcp objective
|
| Nov 3, 2023 at 22:16 | review | Close votes | |||
| Nov 10, 2023 at 3:04 | |||||
| Nov 3, 2023 at 18:46 | comment | added | Mark L. Stone | @Uomond If you have a DCP compliant representation, you should post it here and at ask.cvxr.com/t/dcp-representation-of-1-2-log-z-1-z-log-z-2z-1/… | |
| Nov 3, 2023 at 14:25 | comment | added | Uomond | @BrianBorchers I didn't know that was a thing. Optimization is not my expertise. Do you think this would help? How would that work in this case? If it solves the question practically I'm more than willing to accept it as an answer! I think I found a way to represent it as a DCP compliant program, but I get ill-posedness in this reformulation. I'll check my calculations and edit the post if it is correct. Thanks again for all the help! | |
| Nov 2, 2023 at 22:29 | comment | added | Brian Borchers | Have you considered the trick of rewriting this quasiconvex optimization problem via a family of convex functions as in section 3.4.5. of Boyd and Vandenberghe? | |
| Nov 2, 2023 at 19:04 | history | edited | Uomond | CC BY-SA 4.0 |
Added what can be done about it question. Fixed typos
|
| Nov 2, 2023 at 17:48 | comment | added | Mark L. Stone | Those additive terms aren't convex or concave over the entirety of their natural domain, therefore they are unlikely to have a DCP representation. The same with T. | |
| Nov 2, 2023 at 17:25 | history | edited | Uomond | CC BY-SA 4.0 |
Slight progress: We can write $T$ as sum of simpler concave functions, maybe now we do it?
|
| Nov 1, 2023 at 23:21 | comment | added | Mark L. Stone | I doubt there is a DCP representation for T. But if you post at ask.cvxr.com only about a DCP representation for T, and don't even mention anything else, you will get focused attention from a collection of DCP formulation wizards. I suggest putting the expression in the title. T seems a likely candidate for @ErlingMOSEK 's challenge ask.cvxr.com/t/express-1-a-2-x-2-in-cvx/5642/4 . | |
| Nov 1, 2023 at 22:41 | history | edited | Uomond | CC BY-SA 4.0 |
Added my work thus far
|
| Oct 29, 2023 at 20:42 | history | edited | Uomond | CC BY-SA 4.0 |
edited body
|
| Oct 29, 2023 at 2:21 | history | edited | Uomond | CC BY-SA 4.0 |
deleted 13 characters in body
|
| Oct 29, 2023 at 0:56 | history | edited | Mark L. Stone | CC BY-SA 4.0 |
Changed title to CVXPY, because CVX is a different tool, and does not support DQCP
|
| Oct 28, 2023 at 21:27 | history | edited | Uomond | CC BY-SA 4.0 |
added 164 characters in body
|
| Oct 28, 2023 at 20:32 | history | edited | Uomond | CC BY-SA 4.0 |
added 73 characters in body
|
| S Oct 28, 2023 at 20:20 | review | First questions | |||
| Oct 28, 2023 at 22:25 | |||||
| S Oct 28, 2023 at 20:20 | history | asked | Uomond | CC BY-SA 4.0 |