英文:
HiGHS solver for PyPSA-EUR
问题
我正在尝试在PyPSA-Eur中使用HiGHS求解器。通过在config.yaml中进行如下设置：
solver:
name: highs
options: {}
它会抛出以下错误：
> RuntimeError: 尝试使用不可用的求解器。
>
> SolverFactory 无法创建求解器 "highs"，并返回了一个 UnknownSolver 对象。此错误是在使用 UnknownSolver 对象作为有效对象（通过调用 "has_capability" 方法）的地方引发的。
>
> 最初的求解器是以以下参数创建的：
> executable: highs
> type: highs
> _args: ()
> options: {}
在创建高级环境时，highspy被正确安装，没有任何问题。
而在pypsa-eur环境中，如果我运行 python -c "import highspy"，它不会抛出任何错误，这意味着highspy已经可用于使用。
我们如何配置pypsa以使用HiGHS作为求解器？
英文:
I am trying to use HiGHS solver for PyPSA-Eur. By setting it in config.yaml like:
solver:
name: highs
options: {}
It throws following error:
> RuntimeError: Attempting to use an unavailable solver.
>
> The SolverFactory was unable to create the solver "highs" and returned
> an UnknownSolver object. This error is raised at the point where the
> UnknownSolver object was used as if it were valid (by calling method
> "has_capability").
>
> The original solver was created with the following parameters:
> executable: highs
> type: highs
> _args: ()
> options: {}
while creating the environment highspy gets installed properly and no issues there.
While in the pypsa-eur environment if i do python -c "import highspy" it does not throw any error that means highspy is there to be used.
How can we configure pypsa to use HiGHS as a solver?
答案1
得分: 2
应该可以正常工作，如果你使用以下配置：
solver:
name: highs
options: highs-default
第一行定义了求解器，第二行定义了用于求解器的选项。
英文:
It should work if you use
solver:
name: highs
options: highs-default
First line defines the solver, second line defines the options to use for the solver.
答案2
得分: 1
如果您使用的是Windows，可能需要手动安装HiGHS二进制文件。不确定。
https://ergo-code.github.io/HiGHS/stable/installation/
英文:
Should you be on Windows, it might be that you have to install HiGHS binaries manually. Not sure.
答案3
得分: 1
需要从https://github.com/ERGO-Code/HiGHS构建HiGHS，并且它起作用。还必须在LD_LIBRARY_PATH中包含HiGHS共享库路径，以便可以找到动态链接的库。
从源代码构建HiGHS后，可在以下位置找到可执行文件：
> <HiGHS-isntallation-dir>/HiGHS/build/bin/
可能需要创建符号链接或提供可执行文件的完整路径。
似乎HiGHS无法处理PyPSA-Eur中的网络聚类
如果不支持quadratic_objective"):
logger.warning(
f"配置的求解器`{solver_name}`不支持二次目标。回退到`ipopt`。"
)
opt = po.SolverFactory("ipopt")
results = opt.solve(m, tee=True)
在使用HiGHS时，在results = opt.solve(m, tee=True)处遇到了错误，因此作为一种解决方法，我默认使用ipopt来进行cluster_network规则（snakemake）并使用HiGHS来解决模型。
英文:
Had to build HiGHS from https://github.com/ERGO-Code/HiGHS and it worked. Also HiGHS shared lib path has to be included in LD_LIBRARY_PATH so the dynamically linked lib can be found.
After building HiGHS from source the executable can be found in:
> <HiGHS-isntallation-dir>/HiGHS/build/bin/
One might need to create a symlink or provide a complete path to the executable.
It seems that HiGHS was not able to handle network clustering in PyPSA-Eur
if not opt.has_capability("quadratic_objective"):
logger.warning(
f"The configured solver `{solver_name}` does not support quadratic objectives. Falling back to `ipopt`."
)
opt = po.SolverFactory("ipopt")
results = opt.solve(m, tee=True)
When using HiGHS the error was encountered at results = opt.solve(m, tee=True) so as a workaround I used ipopt by default for cluster_network rule (snakemake) and used HiGHS to solve the model.
通过集体智慧和协作来改善编程学习和解决问题的方式。致力于成为全球开发者共同参与的知识库，让每个人都能够通过互相帮助和分享经验来进步。
评论