Merge

Examples/CONTENT.md

@@ -13,7 +13,7 @@ In the following, the name of the example is the name of the directory where the
 
 ## Utilities ##
   - `Utilities` A set of utilities that you may use also in your code. All described in the local `README.md` file. 
-  **You should go in the directory and type**
+  **You should go in the `Utilities` directory and type**
 
 ``` 
     make

Examples/Makefile.user

@@ -2,12 +2,12 @@
 # This file must be renamed into Makefile.inc in order to work!!
 #
 
-# *****  This is the user defined part of the Makefile
+# *****  This is the user defined part of the Makefile ****
 #
 #  It is included in almost all Makefiles of the example (a part some particulare cases)
 #
-# It has been modified so to read some data from environmental variables
-# and be compatible with the system of modules used at MOX
+# It has been modified so to read some data from environmental variables (if set!)
+# and be compatible with the system of modules used at MOX Laboratory
 #
 # YOU CAN change it to suit your need!!!!
 #
@@ -21,12 +21,14 @@
 #
 
 # The directory where the examples reside (you HAVE to change this)
-PACS_ROOT?=/home/forma/PacsVirtualMachine/pacs/PACSCourse/Material/Examples
+export AMSC_ROOT?=/home/forma/Work/Didattica/HPC_Courses/AMSC-CodeExamples/Examples
+# For compatibility with the PACS course. Do not change it
+export PACS_ROOT=${AMSC_ROOT}
 # The directory with the Eigen (leave blank if they are in a standard directory)
 mkEigenInc?=/usr/local/include/eigen3/
 EIGEN_DIR=$(mkEigenInc)
 #
-# Mpi directories for library and include files
+# Mpi directories for library and include files (change if needed)
 #
 mkOpenmpiLib?=/usr/lib/x86_64-linux-gnu
 mkOpenmpiInc?=/usr/include
@@ -63,7 +65,7 @@ CXX = ${mkCxxCompiler}
 mkCCompiler?=gcc
 CC = ${mkCxxCompiler}
 
-# Since we are only uning c++ I set the linker LINK.o for object files
+# Since we are only using c++ I set the linker LINK.o for object files
 # so that it uses the c++ compiler. Make has no separate macro for the linker!
 LINK.o = $(CXX) $(LDFLAGS) $(TARGET_ARCH)
 #

Examples/README.md

@@ -1,6 +1,8 @@
-# Examples of the Course on Advanced Programming for Scientific Computing (aka PACS)#
-**Mathematical Engineering, Politecnico di Milano**
-**Copyright Luca Formaggia 2012-2024**
+# Course on Advanced programming for scientific computing #
+## Mathematical Engineering, Politecnico di Milano ##
+# Course on Advanced Modelling for scientific computing #
+## HPC and Big Data, Politecnico di Milano ##
+## Copyright Luca Formaggia 2012-2023 ##
 
 ## LICENCING ##
 The software contained in the subfolders of this directory is free-software released under a GNU General Public Licens (GPL). The details of the licence are availabe in the `COPYRIGHT` file in the root directory.
@@ -38,7 +40,7 @@ The first operation is to copy the  file `Makefile.user` to `Makefile.inc`:
 
     $ cp Makefile.user Makefile.inc
 
-and edit the latter to suit your system. In particular, you have to set `PACS_ROOT` to the root directory (with the full path, i.e. the path should start with `/`).
+and edit the latter to suit your system. In particular, you have to set `AMSC_ROOT` to the root directory (with the full path, i.e. the path should start with `/`).
 If you do not know it
 just type
 
@@ -53,18 +55,19 @@ compilation options. But normally you don't need it.
 
 As an example:
 
-    PACS_ROOT=/home/myname/pacs/Examples/
+    AMSC_ROOT=/home/myname/pacs/Examples/
 
 Alternatively, to editing the file, you may set an environmental
 variable with the same name using the command 
 
-    export PACS_ROOT=/home/myname/pacs/Examples/ 
+    export AMSC_ROOT=/home/myname/pacs/Examples/ 
 
 in the `.profile` or in the `.bashrc` file in your home directory (see Note
 at the end of the file). The file `Makefile.inc` is included in all the Makefiles and defines and
 exports the following make macros:
 
-- `PACS_ROOT` the directory where the Examples resides
+- `AMSC_ROOT` the directory where the Examples resides
+- `PACS_ROOT` for compatibility with the APSC course
 - `CXX` the c++ compiler of choice
 - `STANDARD` contains the C++ standard used in compilation
 - `MPI_LIBDIR` and `MPI_INCDIR` where to find mpi libraries and header files, respectively
@@ -155,8 +158,9 @@ env | grep mk
 
   * You do not use the module system. Then you have again two choices
    * You set the various macro in your `Makefile.inc` (and possibly the other Makefiles, if needed) by yourself
-   * You simulate the module environment by creating the environmental variables: you have to put in the `.profile `file in your home directory the corresponding instructions
-        for the bash (the `~/.profile` file is read by the bash shell every time you do a login).  For example, in my `.profile` I have:
+   * You simulate the module environment by creating the environmental variables: you have to put in the `.bash-profile` (or `.profile `) 
+   file in your home directory the corresponding instructions
+        for the bash (the `.bash-profile` and `~/.profile` files are sourced by the bash shell every time you do a login).  For example, in my `.profile` I have:
 
 -------------------------------------------------------------------------------
 
@@ -258,13 +262,13 @@ content is run as if typed on a terminal). Everytime you open a bash
 shell the `.bashrc` file is sourced. 
 
 What's the difference? Well today people prefer to put everything in
-the `.bashrc` file, but in principle envronmental variables
-definitions should be put in the `.profile` (or `.bash_profile`),
+the `.bashrc` file, but in principle definitions of environmental variables
+should be put in the `.bash_profile`  (or `.profile`),
 while `.bashrc` should be reserved for command alias or customization
-that need to change wether you are in a interactive shell (terminal)
+that need to change depending whether you are in a interactive shell (i.e. a terminal)
 or not (batch job). 
 
-A practical difference is that if you change `.profile` (or `.bash_profile`)
+A practical difference is that if you change `.bash_profile`  (or `.profile`)
 the changes become operative only after a new login. 
 Changes in `.bashrc` becomes active just by opening a new terminal.
 

Examples/src/Bindings/main.cpp

@@ -2,6 +2,7 @@
 // parameters are not used
 #pragma GCC diagnostic push
 #pragma GCC diagnostic ignored "-Wunused-parameter"
+#pragma GCC diagnostic ignored "-Wunused-variable"
 
 #include <iostream>
 #include <typeinfo>
@@ -52,6 +53,8 @@ foo(int &a)
 void
 foo(const int &a)
 {
+  auto av=5*a; // ok i can read a
+  // a=10.// Error! I cannot change a
   std::cout << "using void foo(const int&)" << std::endl;
 }
 
@@ -76,6 +79,7 @@ void
 goo(std::vector<int> &&a)
 {
   std::vector<int> b=std::move(a);
+  // now a is empty
   std::cout << "using void goo(vector<int> &&)" << std::endl;
 }
 
@@ -87,6 +91,8 @@ goo(std::vector<int> &&a)
 void
 goo(const std::vector<int> &a)
 {
+  auto av=a[0]; // ok i can read a
+  // a[0]=10.// Error! I cannot change a
   std::cout << "using void goo(const vector<int>&)" << std::endl;
 }
 
@@ -97,7 +103,10 @@ goo(const std::vector<int> &a)
  */
 void
 goo(const std::vector<int> &&a)
-{
+{  
+  auto av=a[0];
+  // ok i can read a
+  // a[0]=10.// Error! I cannot change a
   std::cout << "using void goo(const vector<int> &&)" << std::endl;
 }
 
@@ -127,7 +136,7 @@ gvect(T &&x)
   T v = std::forward<T>(x);
   std::cout << "using void gvect<T>(T&&) with std::forward\n"
             << " created vector size is " << v.size()
-            << ". Input vector size is now " << x.size() << std::endl;
+            << " Input vector size is now " << x.size() << std::endl;
 }
 
 int

Examples/src/FloatingPoint/FPExceptions/README.md

@@ -4,13 +4,13 @@ When you perform an invalid floating point operation (like the square root
 of a negative number) or you have overflows/underflows, the execution
 does not stop. Yet in many architectures a flag is raised to indicate
 that a **floating point exception** has occurred. You may test if the
-flag is raised, and take the appropriate action, if you wish. But it require  some work.
+flag is raised, and take the appropriate action, if you wish. But it requires  some work.
 
 This directory contains two utilities
 
 * `trapfpe.hpp`, a header only utility that you should include in your main source file if you want to be able to activate *abort on major fp exceptions*.  To activate this feature, you have to compile your code with the cpp option `-DFPE_ABORT`. If not, you have the default behaviour in case of floating point exceptions.  With major floating point exceptions here I mean: invalid operation, divide by zero, overflow.
 
-* `fpeUtilities.[hc]pp` the sources of a little library that introduces some exception classes inherited from std::runtime_error and a function that tests if major fp exception has been raised, and in that case launches the relevant exception, or returns a boolean equal to true. To use this utility you need to link with the libfpe library created in this example. 
+* `fpeUtilities.[hc]pp` the sources of a little library that introduces some exception classes inherited from std::runtime_error and a function that tests if major fp exception has been raised, and in that case launches the relevant exception, or returns a boolean equal to true. To use this utility you need to link the `libfpe.so` library created by this example. 
 
 A test is available, `main_fpe.cpp`, to test the utilities.
 
@@ -35,7 +35,7 @@ Remember that you have to make
 ```
 make install 
 ```
-to install in the correct directory all headers and library
+to install in the correct directory all headers and the library
 `libfpe.[a|so]` containing the compiled fpe Utility for use in other
 examples.
 

Examples/src/FloatingPoint/FPFailure/explanation.tex

@@ -45,7 +45,7 @@ \section{Wrongly converging sequence}.
 \]
     It can be easily found, by looking at the roots of the polynomial
     \[
-    x^3-111+1130x^2-3000,
+    x^3-111x^2+1130x-3000,
     \]
     that, if the sequence converges, it converges necessarily to either
     $5$, $6$, or $100$.

Examples/src/FloatingPoint/FinDiff/gnuplot.txt

@@ -7,9 +7,14 @@ set style line 2 lc 2 ps 2
 set style line 4 lc 3 ps 1
 set format y "%3.1e"
 set format x "%3.1e"
-plot "single.dat" using 1:3 title "Error float 2" ls 1 with line, "single.dat" using 1:4 title "Roundoff error" ls 2 with points, "single.dat" using 1:5 title "Error float 4" ls 3 with points, "single.dat" using 1:6 title "Error float wd" ls 4 with points
-pause mouse
-plot "double.dat" using 1:3 title "Error double 2" ls 1 with line, "double.dat" using 1:4 title "Roundoff error" ls 2 with points, "double.dat" using 1:5 title "Error double 4" ls 3 with points, "double.dat" using 1:6 title "Error double wd" ls 4 with points
-pause mouse
-plot "extended.dat" using 1:3 title "Error extended 2" ls 1 with line, "extended.dat" using 1:4 title "Roundoff error" ls 2 with points, "extended.dat" using 1:5 title "Error extended 4" ls 3 with points, "extended.dat" using 1:6 title "Error ext. wd" ls 4 with points
-pause mouse
+#plot into three separate windows using gnuplot
+set terminal x11 0
+set title "Single Precision Errors"
+plot "single.csv" using 1:3 title "Error 2nd order" ls 1 with line, "single.csv" using 1:4 title "Roundoff error" ls 2 with points, "single.csv" using 1:5 title "Error 4th order" ls 3 with points, "single.csv" using 1:6 title "Error wd" ls 4 with points
+set terminal x11 1
+set title "Double Precision Errors"
+plot "double.csv" using 1:3 title "Error 2nd order" ls 1 with line, "double.csv" using 1:4 title "Roundoff error" ls 2 with points, "double.csv" using 1:5 title "Error 4th order" ls 3 with points, "double.csv" using 1:6 title "Error wd" ls 4 with points
+set terminal x11 2
+set title "Extended Precision Errors"
+plot "extended.csv" using 1:3 title "Error 2nd order" ls 1 with line, "extended.csv" using 1:4 title "Roundoff error" ls 2 with points, "extended.csv" using 1:5 title "Error 4th order" ls 3 with points, "extended.csv" using 1:6 title "Error wd" ls 4 with points
+pause -1

Examples/src/FloatingPoint/FinDiff/main_findiff.cpp

@@ -8,26 +8,26 @@
 #include <vector>
 #if defined(SINGLE_PRECISION)
 using Real = float;
-std::string filename("single.dat");
+std::string filename("single.csv");
 #elif defined(EXTENDED_PRECISION)
 using Real = long double;
-std::string filename("extended.dat");
+std::string filename("extended.csv");
 #else
 using Real = double;
-std::string filename("double.dat");
+std::string filename("double.csv");
 #endif
 
 // Global values (are outside any scope so I can change them in
 // funct and dfunct in just one go)
 constexpr Real a = 100.0;
 constexpr Real b = 4.0;
 
-// I use lambdas and I exploit the fact that constexpr are capured automatically!
+// I use lambdas and I exploit the fact that constexpr are capured
+// automatically!
 //! \f$ f(x)=a e^{bx}\f$
-auto funct= [](auto const & x){return a * std::exp(b * x);};
+auto funct = [](auto const &x) { return a * std::exp(b * x); };
 //! \f$ f^\prime(x)= ab e^{bx}\f$
-auto dfunct= [](auto const & x){return a * b* std::exp(b * x);};
-
+auto dfunct = [](auto const &x) { return a * b * std::exp(b * x); };
 
 //! Tests of accuracy of finite differences
 /*!
@@ -38,16 +38,16 @@ auto dfunct= [](auto const & x){return a * b* std::exp(b * x);};
 int
 main()
 {
-  // Get roundoff unit
-  const Real u = 0.5 * std::numeric_limits<Real>::epsilon();
+  // Get roundoff unit (a use of auto)
+  const auto u = 0.5 * std::numeric_limits<Real>::epsilon();
   std::cout << "Roundoff unit=" << u << std::endl;
   // Read some data
   GetPot ifl("data.pot");
   // Initial spacing
   Real h = ifl("h", 0.1);
   std::cout << " Initial Spacing " << h << std::endl;
   // Number of subdivisions
-  const unsigned int n = ifl("n", 10);
+  const auto n = ifl("n", 10u);
   std::cout << " Number of subdivisions " << n << std::endl;
   // Point where derivative is computed
   const Real x = ifl("x", 3.0);
@@ -70,9 +70,9 @@ main()
   spacing.reserve(n);
   roundoffErrorEstimate.reserve(n);
 
-  Real constexpr half(0.5);
-  Real constexpr twelvth(1. / 12.);
-  Real constexpr twothird(2. / 3);
+  Real constexpr half = 0.5;
+  Real constexpr twelvth = 1. / 12.;
+  Real constexpr twothird = 2. / 3;
 
   for(unsigned int i = 0; i < n; ++i)
     {
@@ -99,14 +99,18 @@ main()
     }
   // Write data
   std::ofstream file(filename.c_str());
-  file.width(15);
-  file<<"# h\t d2\t e2\t estim\t d4\t without_a diff\n";
+  auto constexpr max_prec = std::numeric_limits<Real>::max_digits10;
+  file << "# h,\t diff2,\t error2,\t truncation error,\t error4,\t error "
+          "without_a diff\n";
+  file.precision(
+    max_prec); // do this if you want the data printed in full precision!
+  // file.precision(8); // do this if you want the data printed in a smaller
+  // precision!
   for(unsigned int i = 0; i < n; ++i)
     {
-      file.width(15);
-      file << std::scientific<< spacing[i] << "\t" << derNumer[i] << "\t" << Error[i] << "\t"
-           << roundoffErrorEstimate[i] << "\t" << Error4[i] << "\t" << Errorcx[i]
-           << std::endl;
+      file << std::scientific << spacing[i] << ",\t" << derNumer[i] << ",\t"
+           << Error[i] << ",\t" << roundoffErrorEstimate[i] << ",\t"
+           << Error4[i] << ",\t" << Errorcx[i] << std::endl;
     }
   file.close();
 }