LinuxQuestions.org - Fedora 28, 'Nothing provides libmpi.so.20()(64bit)' with openmpi and rpm-mpi-hooks installed

As the title says, on Fedora 28, I'm receiving errors from rpm and dnf won't install my package because libmpi.so.20()(64bit). I've verified the package is in /usr/lib64/openmpi/lib and PATH includes it. OpenMPI and rpm-mpi-hooks are installed. The complete .spec file is below

Code:

# spec file for Bertini v1.6



# define _topdir /home/rcx11/Bertini

%define name Bertini

%define release 0

%define version 1.6

# define buildroot %{_topdir}/%{name}-%{version}-root



BuildRoot:        /var/tmp/%{name}-%{version}-root

Summary:        Bertini - Numerical Algebraic Geometry

License:        Proprietary

Name:                %{name}

Version:        %{version}

Release:        %{release}

Source:                %{name}-%{version}.tar.gz

Prefix:                /usr/local

Group:                Education



Requires:        bison

Requires:        flex

Requires:        gmp

Requires:        mpfr

Requries:        openmpi

Requires:        openmpi-devel

Requires:        rpm-mpi-hooks



BuildRequires:        gmp

BuildRequires:        flex

BuildRequires:        bison

BuildRequires:        mpfr

BuildRequires:        openmpi

BuildRequires:        openmpi-devel

BuildRequires:        rpm-mpi-hooks



%description

Software for solving numerical solutions to polynomial systems using Homotopy continuation.

Bertini is a general-purpose solver, written in C, that was created for research about polynomial continuation. 



 Finds isolated solutions using total-degree start systems, multihomogeneous-degree start systems, and also user defined homotopies.

 Implements parameter continuation for families of systems, such as the inverse kinematics of six-revolute serial-link arms, or the forward kinematics of Stewart-Gough parallel-link robots.

 Adaptive multiprecision implemented for finding isolated solutions and for the numerical irreducible decomposition.

 Treats positive-dimensional solutions by computing witness sets.

 Has automatic differentiation which preserves the straightline quality of an input system.

 Uses homogenization to accurately compute solutions "at infinity."

 Provides endgames to accurately compute singular roots.

 Allows for subfunctions.

 Allows for witness set manipulation via both sampling and membership testing.

 Accepts underdetermined, exactly determined, and overdetermined systems (i.e., the number of variables and equations do not have to be equal). 



%prep

%setup -q



%build

./configure

make



%install

make install prefix=$RPM_BUILD_ROOT/usr/local



%files

%defattr(-,root,root)

/usr/local/bin/bertini

/usr/local/include/bertini.h

/usr/local/include/cascade.h

/usr/local/include/diff.h

/usr/local/include/dimbydim.h

/usr/local/include/eqbyeq.h

/usr/local/include/localdim.h

/usr/local/include/parallel.h

/usr/local/include/pos_dim.h

/usr/local/include/ppParse.h

/usr/local/include/regeneration.h

/usr/local/include/regen_pos_dim.h

/usr/lib/debug/usr/local/bin/bertini-parallel-1.6-0.x86_64.debug

/usr/lib/debug/usr/local/bin/bertini-serial-1.6-0.x86_64.debug

/usr/lib/debug/usr/local/lib/libbertini-parallel.so.0.0.0-1.6-0.x86_64.debug

/usr/lib/debug/usr/local/lib/libbertini-serial.so.0.0.0-1.6-0.x86_64.debug

/usr/local/bin/bertini-parallel

/usr/local/bin/bertini-serial

/usr/local/lib/libbertini-parallel.a

/usr/local/lib/libbertini-parallel.la

/usr/local/lib/libbertini-parallel.so

/usr/local/lib/libbertini-parallel.so.0

/usr/local/lib/libbertini-parallel.so.0.0.0

/usr/local/lib/libbertini-serial.a

/usr/local/lib/libbertini-serial.la

/usr/local/lib/libbertini-serial.so

/usr/local/lib/libbertini-serial.so.0

/usr/local/lib/libbertini-serial.so.0.0.0





# %doc %attr(0444,root,root)