Closed iljamaki closed 1 year ago
hey @iljamaki,
I just tested the latest extension v1.0.3
with floating point parameter strategy and didn't see any kind of issue.
Please make sure that your strategy parameters matches with the OptiPie extension parameters you give.
Here is the sample strategy optimization result with floating point parameter
Note: v1.0.3
didn't have any kind of update/code changes to the optimization process as well.
I have made sure. The first input in the strategy is a floating point value with minimum set to 1.0. In the optimization I only optimize this single parameter, trying to see which value between 1.5 - 2.0 is the optimal one. When the optimization runs, from the settings dialog I see that it tries values 1.4, 2.4, 3.4, 4.4, 5.4, 6.4, and 7.4 and then fails. Since I set 0.1 as the step, I would expect it to try values 1.5, 1.6, 1.7, 1.8, 1.9, 2.0, and succeed. As I said before, I'm not sure if this issue existed in version 1.0.2 or not.
In the previous issue I reported, the case was that you were testing in a different setup and therefore you couldn't reproduce the issue. I'm pretty sure it's something similar this time again. I'm just trying to help make this software better and bug free but I recognize you are not interested, so apologies for the disturbance.
hey @iljamaki As I said, I was not able produce that error from multiple tests(that's why I have closed the issue, tried with 4 different strategies with floating point parameter) which means that yours would be edge case situation. To understand and analyze the problem, we need more details for sure. Could you please provide the steps below?
Let me produce the same error that you have so that I can easily address the issue and fix it. Thanks a lot for the collaboration, appreicated.
closing as stale issue
In version 1.0.3, when you try to optimize a floating point parameter value, for example start=1.5, end=2.0, step=0.1, the optimization will fail. To me it seems it's trying to step with integer steps even though 0.1 step is given, but that is just a guess. I think this worked in version 1.0.2 but I'm not sure.