Improve Constants Optimization for Symbolic Regression

[HeuristicLab-Trac-Bot]

Issue migrated from trac ticket # 2852

milestone: HeuristicLab 3.3.15 | component: Problems.DataAnalysis.Symbolic.Regression | priority: medium | resolution: done

2017-11-03 15:14:42: @mkommend created the issue

---

• Constants optimization should work without linear scaling
• Constants optimization should report the number of function and gradient evaluations

[HeuristicLab-Trac-Bot]

2017-11-03 15:27:30: @mkommend changed status from new to accepted

[HeuristicLab-Trac-Bot]

2017-11-03 15:29:48: @mkommend commented

---

r15447: Adapted constants optimization and auto diff converter to not add linear scaling terms.

[HeuristicLab-Trac-Bot]

2017-11-03 17:10:04: @mkommend changed status from accepted to reviewing

[HeuristicLab-Trac-Bot]

2017-11-03 17:10:04: @mkommend changed owner from @mkommend to @gkronber

[HeuristicLab-Trac-Bot]

2017-11-03 17:10:04: @mkommend changed milestone from HeuristicLab 3.3.16 to HeuristicLab 3.3.15

[HeuristicLab-Trac-Bot]

2017-11-03 17:10:04: @mkommend commented

---

r15448: Added counts of function and gradient evaluations performed by constants optimization as results.

Maximum gradient evaluations = maximum iterations + 1 \ Maximum function evaluations = maximum iterations * 2 + 1;

[HeuristicLab-Trac-Bot]

2017-11-07 15:28:49: @gkronber commented

---

Reviewed r15448.

Are 'result parameters' already available in the trunk?

Reading the code it is odd to find a 'ResultParameter'. I think at least the properties should be renamed (FunctionEvaluationsResultParameter -> FunctionEvaluationsResult, FunctionEvaluationsResultParameterName -> FunctionEvaluationsResultName)

[HeuristicLab-Trac-Bot]

2017-11-08 10:25:18: @mkommend commented

---

Replying to [comment:4 gkronber]:

Are 'result parameters' already available in the trunk? Resultparameter were introduced with #2281 over a year ago!

\

Reading the code it is odd to find a 'ResultParameter'. I think at least the properties should be renamed (FunctionEvaluationsResultParameter -> FunctionEvaluationsResult, FunctionEvaluationsResultParameterName -> FunctionEvaluationsResultName)

I have already implemented the proposed changes, but during a final review and test I decided against committing them. The reason therefore is that an execution context must be present (and manually set line 169) to allow access to the result and the postfix parameter indicates that.

[HeuristicLab-Trac-Bot]

2017-11-20 20:36:43: @gkronber commented

---

r15480: moved the scaling parameters to the end of the parameter vector to be able to remove the c.Skip(2).ToArray() call and removed unnecessary .ToArray() calls

[HeuristicLab-Trac-Bot]

2017-11-20 21:06:11: @gkronber changed status from reviewing to readytorelease

[HeuristicLab-Trac-Bot]

2017-11-20 21:06:11: @gkronber changed owner from @gkronber to @mkommend

[HeuristicLab-Trac-Bot]

2017-11-20 21:06:11: @gkronber commented

---

Reviewed r15447. Concerning r15448 I have the feeling that the algorithm is slower because of the updates of function and gradient evaluation results. However, I have not really tested this.

Please check r15480 and release.

[HeuristicLab-Trac-Bot]

2017-11-21 10:51:14: @mkommend changed status from readytorelease to assigned

[HeuristicLab-Trac-Bot]

2017-11-21 10:51:14: @mkommend changed owner from @mkommend to @gkronber

[HeuristicLab-Trac-Bot]

2017-11-21 10:51:14: @mkommend commented

---

r15480 is a good idea (too get rid of the .ToArray() calls), but is not implemented correctly.

It might be that r15448 slows down constants optimization. An option would be to make this analysis optional and disabled per default.

[HeuristicLab-Trac-Bot]

2017-11-21 15:40:29: @gkronber commented

---

r15481: reverted changeset r15480 partially to fix unit test

[HeuristicLab-Trac-Bot]

2017-11-21 15:55:07: @gkronber commented

---

After some tests I conclude that the order of parameters in lsfit() has an influence on the optimization results. Therefore, we cannot put the scaling parameters at the end of the parameter vector without breaking backwards compatibility.

Please add a parameter to make reporting of the number of gradient and function evaluations optional.

[HeuristicLab-Trac-Bot]

2017-11-21 15:55:26: @gkronber changed owner from @gkronber to @mkommend

[HeuristicLab-Trac-Bot]

2017-11-21 17:12:43: @mkommend changed status from assigned to reviewing

[HeuristicLab-Trac-Bot]

2017-11-21 17:12:43: @mkommend changed owner from @mkommend to @gkronber

[HeuristicLab-Trac-Bot]

2017-11-21 17:12:43: @mkommend commented

---

r15483: Added option for reporting function and gradient evaluation in the constants optimization evaluator.

Unfortunately, you have been right and reporting the evaluations has a huge impact on the performance. Quick tests on my work station for the OSGP SymReg sample show execution times of ~ 5.1 min with and ~ 3.02 min without reporting (fixed seed of 0).

[HeuristicLab-Trac-Bot]

2017-11-22 12:35:29: @gkronber commented

---

Reviewed r15483. Thanks for testing this.

[HeuristicLab-Trac-Bot]

2017-11-22 12:35:40: @gkronber changed status from reviewing to readytorelease

[HeuristicLab-Trac-Bot]

2017-11-22 12:35:40: @gkronber changed owner from @gkronber to @mkommend

[HeuristicLab-Trac-Bot]

2017-12-12 16:16:32: @mkommend commented

---

r15515: Merged r15447, r15448, r15480, r15481, r15483 into stable.

[HeuristicLab-Trac-Bot]

2017-12-27 11:39:07: @mkommend changed status from readytorelease to closed

[HeuristicLab-Trac-Bot]

2017-12-27 11:39:07: @mkommend set resolution to done