The command cmake --install build/release in Windows powershell

Returns

CMake Error at build/release/ext/sgtelib/cmake_install.cmake:39 (file):
  file INSTALL cannot find
  "C:/......../nomad-v.4.2.0/build/release/ext/sgtelib/Release/sgtelib_main.exe":

A temporary fix consists of replacing the file CMakeLists.txt in C:/..../nomad-v.4.2.0/ext/sgtelib by the one in attachment. I just commented "sgtelib_main.exe" which is not necessary for Nomad optimization. CMakeLists.txt

I am on windows server 2019. The command cmake -S . -B build/release returns an error.

CMake Error at CMakeLists.txt:9 (project):
  Running
    'nmake' '-?'
  failed with:
     The system cannot find the file specified

CMake Error: CMAKE_C_COMPILER not set, after EnableLanguage
CMake Error: CMAKE_CXX_COMPILER not set, after EnableLanguage
-- Configuring incomplete, errors occurred!
See also "C:/Nomad/nomad-4.1.0/build/release/CMakeFiles/CmakeOutput.log".

Any ideas?

see https://dev.azure.com/JuliaPackaging/Yggdrasil/_build/results?buildId=20791&view=logs&jobId=eb643228-59a0-57e5-a6a5-6cb55400d5ea&j=26a97fd6-2070-5ccc-98fe-466416439df2&t=abe13601-88ab-5e8d-6817-533311385b0c

OS: Debian "bullseye" Nomad version: 4.2

When building the Java wrapper, following the instructions in interfaces/jNomad/Readme, I get the following error:

[ 93%] Built target jNomad
[ 93%] Linking CXX executable NMOpt.exe
Scanning dependencies of target jNomad_jar
gmake[2]: *** No rule to make target 'interfaces/jNomad/SWIGTYPE_p_NOMAD_4_1__EvalParameters.java', needed by 'interfaces/jNomad/CMakeFiles/jNomad_jar.dir/java_compiled_jNomad_jar'.  Stop.
gmake[1]: *** [CMakeFiles/Makefile2:1366: interfaces/jNomad/CMakeFiles/jNomad_jar.dir/all] Error 2
gmake[1]: *** Waiting for unfinished jobs....

and cmake exits with error code 2.

Here is the complete Dockerfile I'm using:

FROM openjdk:8u332-bullseye

ENV DEBIAN_FRONTEND=noninteractive

# Install dependencies
RUN apt-get update && apt-get install -y --no-install-recommends \
	build-essential	\
	m4 \
    cmake \
    nano \
    curl

WORKDIR /

# First install SWIG
# Install dependencies
RUN apt-get update && apt-get install -y --no-install-recommends \
	git \
	autotools-dev \
	automake \
	byacc \
	libpcre2-dev

RUN git clone https://github.com/swig/swig.git

WORKDIR swig
RUN ./autogen.sh
RUN ./configure --prefix=/swig
RUN make
RUN make install

RUN ln -s /swig/bin/swig /usr/local/bin/swig

# Compile and install NOMAD
WORKDIR /

RUN curl https://codeload.github.com/bbopt/nomad/tar.gz/refs/tags/v.4.2.0 --output NOMAD42.tar.gz
RUN tar -xf NOMAD42.tar.gz && mv nomad-v.4.2.0 NOMAD42 && rm -r NOMAD42.tar.gz

ENV NOMAD_HOME="/NOMAD42"
WORKDIR /NOMAD42

RUN cmake -DTEST_OPENMP=OFF -DBUILD_INTERFACE_C=ON -DBUILD_INTERFACE_JAVA=ON  -S . -B build/release && \
    cmake --build build/release --parallel 16 && \
    cmake --install build/release

ENV LD_LIBRARY_PATH="/NOMAD42/build/release/lib"

WORKDIR /
CMD /bin/bash

I was able to build the C and Python interfaces without a problem.

Any help would be greatly appreciated!

Hi, I'm devising an application for which I'm using Nomad in python as an MIP solver. I want to use PyNomad recurrently to solve several subproblems but I get the following message from the second execution of Nomad (it works well in the first one) :

How can I avoid such problem?

Best regards, Juan José

Hi, I am new using the NOMAD to do a evaluation of a problem. When I download the NOMAD 4 and modify the single_obj file to a simple objective function like y=sin(x) * x *x, every parameter seems set good. NOMAD4 will run into a dead loop while NOMAD3.9.1 will run with best feasible solution immediately. They are the same project in different libraries. I am not sure this is normal for using NOMAD. Thank you

The size of this repository is over 300 Mb, mostly because of the compiled docs. I'm wondering if you could shrink it, by making the compiled docs available by other means. This might benefit users with poor connection, by reducing network traffic and storage.

Thanks for making this great tool available here!

The comment claims the function AllParameters::readParamLine throws for unknown parameters.

The implementation, however, does not throw anything. Instead, the errors are printed out to the standard error stream.

Hello Nomad team,

for Nomad 3.9.1, there is an installer for end user so the user can install with a simple click of the installer file.

Could we also have an installer for Nomad 4 for end user? Typically I am asking for Windows OS.

Thank you, Jenny

Would it be pertinent to have a section about the Julia interface Nomad.jl in this section https://nomad-4-user-guide.readthedocs.io/en/latest/LibraryMode.html ?

Hi, I am using the NOMAD3.9.1 to evaluate a blackbox task. It will drop to obvious local optimum somethings. I notice that the outputs are nearly same or even get very small differences at the end of the evaluation. In some conditions, it will last for hundres of iterations. More compuating resources and time have been put into the final result, but the result has satisfied the requirements. So, my question is how can I alter some arguments or is there a mechanism to terminate the evaluaion in NOMAD to solve this problem? I know user can use CTRL-C to terminate the algorithm mamually. Thank you very much.

The documentation for the MATLAB interface diverges from the actual implementation.

1. According to the documentation, nomadOpt returns [ x, fval, exitflag, iter, nfval ], however, the actual implementation returns [ x, fval, hinf, exitflag, nfval ]. The example reflects the actual implementation.
2. Values of the exitflag do not follow the documented scheme. Instead of the declared spectrum of return values, only two values are returned and their meaning differs from the documentation. The value 0 represents generic success, whereas -1 indicates no feasible result was found.

I believe it may be possible to modify nomadmex.cpp to approach the documented behavior. The following pseudo-code might do that.

bool hasConverged = (nbBestFeas > 0);
bool hasInfeasible = (nbBestInf > 0);

bool exitInitializationError = AllStopReasons::testIf(INITIALIZATION_FAILED);
bool exitUser = (
  AllStopReasons::testIf(BaseStopType::CTRL_C) ||
  AllStopReasons::testIf(BaseStopType::USER_STOPPED));
bool exitNomadError = (
  AllStopReasons::testIf(BaseStopType::ERROR) ||
  AllStopReasons::testIf(BaseStopType::UNKNOWN_STOP_REASON));
bool exitExceededEvaluations =  (
  AllStopReasons::testIf(EvalGlobalStopType::MAX_EVAL_REACHED) ||
  AllStopReasons::testIf(EvalGlobalStopType::MAX_BB_EVAL_REACHED) ||
  AllStopReasons::testIf(EvalGlobalStopType::MAX_BLOCK_EVAL_REACHED) ||
  AllStopReasons::testIf(EvalGlobalStopType::MAX_SURROGATE_EVAL_OPTIMIZATION_REACHED));

if (exitUser) {
  * exitflag = -5;
}
else if (exitNomadError) {
  * exitflag = -3;
}
else if (exitInitializationError) {
  * exitflag = -2;
}
else if (hasConverged) {
  * exitflag = 1;
}
else if (exitExceededEvaluations) {
  * exitflag = 0;
}
else if (hasInfeasible) {
  * exitflag = -1;
}
else {
  // Something else must have happened.
  // There are neither feasible nor infeasible points and there is no clear reason for termination. 
  // Let's chalk that up as 'nomad error'.
  * exitflag = -3;
}

REMARKS

• I see a potential issue regarding the definition of converged / target reached within the above code: should we consider the search successful if at least one feasible point was found even though the optimizer eventually ran out of evaluations?
• I currently do not have access to MATLAB-compatible version of GCC. Once I do, and if desired, I can eventually turn this into a pull request.
• Essentially identical code could be used to resolve https://github.com/bbopt/nomad/issues/104.

The current installation instructions make very hard to install PyNomad in virtual enviroment, in fact the --user option in the line COMMAND python setup_PyNomad.py ${CMAKE_BINARY_DIR} ${NOMAD_VERSION} install inside interfaces/PyNomad/CMakeLists.txt overwrite the default setting of a virtual environment.

A separate parameter PYTHON_DIR defined in CMake could allow the user to specify the correct Python location, otherwise the --user flag could be removed to let the system the best location by itself.

Is there a way to install PyNomad in a virtual environment with the current code?

What are the NOMAD 4.x equivalent parameters to the 3.x parameters "h_min" and "h_norm"? I don't see anything that is roughly the same as these in the 4.x parameter list.

I found an unofficial pip installation path to an older and possibly modified version of NOMAD. Would be convenient to have an official pip installation path to the latest version.

Both python and matlab interfaces should provide complete exit status range: % 1 - converged / target reached % 0 - maximum iterations / function evaluations exceeded % -1 - infeasible / mesh limit reached % -2 - initialization error % -3 - nomad error % -5 - user exit

This issue has been updated using issue #124. Maybe we need more exit statuses (and clarify the meaning):

• (feasible+max eval/iter reached) from (infeasible+max eval/iter reached).
• (feasible+mesh limit reached-> equivalent to 1-converged) from (infeasible+mesh limit reached).

My problem

I am building nomad on macOS for MATLAB. I hence ran

cmake -DBUILD_INTERFACE_MATLAB=ON -DMatlab_ROOT_DIR=/Applications/MATLAB_R2022b.app -S . -B build/release

This completed successfully. However, when building the library with

cmake --build build/release

I get the following error.

[ 99%] Linking CXX shared library nomadOpt.mexmaci64
Undefined symbols for architecture x86_64:
  "_utIsInterruptPending", referenced from:
      matlabEval::eval_x(NOMAD_4_2::EvalPoint&, NOMAD_4_2::Double const&, bool&) const in nomadmex.cpp.o
  "_utSetInterruptPending", referenced from:
      matlabEval::eval_x(NOMAD_4_2::EvalPoint&, NOMAD_4_2::Double const&, bool&) const in nomadmex.cpp.o
ld: symbol(s) not found for architecture x86_64
clang: error: linker command failed with exit code 1 (use -v to see invocation)
make[2]: *** [interfaces/Matlab_MEX/nomadOpt.mexmaci64] Error 1
make[1]: *** [interfaces/Matlab_MEX/CMakeFiles/nomadOpt.dir/all] Error 2
make: *** [all] Error 2

I set up mex to run with clang.

My architecture

• macOS: version 12.6 (Monterey, x86_64)
• clang: version 14.0.0 (clang-1400.0.29.102)
• MATLAB: version 9.13.0.2049777 (R2022b)
• CMake: version 3.24.1

Do you have any idea how I can fix my problem? Thank you very much for your help :grin: