Conda Packages

README.md

@@ -1,5 +1,9 @@
-# Detailed installation instructions (64-bit Ubuntu 16 or 18):
+# Installation instructions:
+In principle you should be able to install this on Windows / OS X / Linux using conda and 'conda install -c marcus-o s4'.
+Install in a new environment to avoid incompatabilities.
+I don't have a large test base, so I would appreciate if you let me know if it works (marcus.ossiander at gmail) or not, then I'll try to help).
 
+# Detailed installation instructions (64-bit Ubuntu 16 or 18):
 ## Key steps:
 
 ```

S4/RNP/Eigensystems.h

@@ -1,6 +1,10 @@
 #ifndef _RNP_EIGENSYSTEMS_H_
 #define _RNP_EIGENSYSTEMS_H_
 
+#ifdef _MSC_VER 
+#include <algorithm>
+#endif
+
 #include <cstddef>
 #include <complex>
 

S4/S4.cpp

@@ -61,7 +61,7 @@ namespace bs = boost::serialization;
 
 
 void* S4_malloc(size_t size){ // for debugging
-	void* ret = malloc_aligned(size, 16);
+	void* ret = malloc_aligned(size, 128);
 	// memset(ret, 0x0, size);
 	return ret;
 }
@@ -2939,7 +2939,8 @@ int Simulation_GetField(Simulation *S, const double r[3], double fE[6], double f
 	}
 	std::complex<double> *work = ab + n4;
 
-	RNP::TBLAS::Copy(n4, Lsoln->ab,1, ab,1);
+	memcpy(ab, Lsoln->ab, sizeof(std::complex<double>) * n4);
+	//RNP::TBLAS::Copy(n4, Lsoln->ab,1, ab,1);
 	//RNP::IO::PrintVector(n4, ab, 1);
 	TranslateAmplitudes(S->n_G, Lbands->q, L->thickness, dz, ab);
 	std::complex<double> efield[3], hfield[3];
@@ -3051,7 +3052,8 @@ int Simulation_GetFieldPlane(Simulation *S, int nxy[2], double zz, double *E, do
 	}
 	std::complex<double> *work = ab + n4;
 
-	RNP::TBLAS::Copy(n4, Lsoln->ab,1, ab,1);
+	memcpy(ab, Lsoln->ab, sizeof(std::complex<double>) * n4);
+	//RNP::TBLAS::Copy(n4, Lsoln->ab,1, ab,1);
 	//RNP::IO::PrintVector(n4, ab, 1);
 	TranslateAmplitudes(S->n_G, Lbands->q, L->thickness, dz, ab);
 	size_t snxy[2] = { nxy[0], nxy[1] };

S4/config.h

@@ -8,3 +8,7 @@
 #define PACKAGE_URL ""
 #define PACKAGE_VERSION "1.1.1"
 #define VERSION "1.1.1"
+
+#ifdef _MSC_VER 
+#define strcasecmp _stricmp
+#endif

S4/fmm/fft_iface.cpp

@@ -20,7 +20,7 @@
 #include "fft_iface.h"
 #include <cstdlib>
 
-#ifdef HAVE_LIBFFTW3
+#ifdef HAVE_FFTW3
 #include <fftw3.h>
 #else
 #include <kiss_fft.h>
@@ -46,23 +46,23 @@ int fft_next_fast_size(int n){
 }
 
 std::complex<double> *fft_alloc_complex(size_t n){
-#ifdef HAVE_LIBFFTW3
+#ifdef HAVE_FFTW3
 	return (std::complex<double>*)(fftw_complex*)fftw_malloc(sizeof(fftw_complex) * n);
 #else
 	return (std::complex<double>*)KISS_FFT_MALLOC(sizeof(std::complex<double>) * n);
 #endif
 }
 
 void fft_free(void *p){
-#ifdef HAVE_LIBFFTW3
+#ifdef HAVE_FFTW3
 	fftw_free(p);
 #else
 	KISS_FFT_FREE(p);
 #endif
 }
 
 struct tag_fft_plan{
-#ifdef HAVE_LIBFFTW3
+#ifdef HAVE_FFTW3
 	fftw_plan plan;
 #else
 	kiss_fftnd_cfg cfg;
@@ -76,7 +76,7 @@ fft_plan fft_plan_dft_2d(
 	int sign
 ){
 	fft_plan plan = NULL;
-#ifdef HAVE_LIBFFTW3
+#ifdef HAVE_FFTW3
 # ifdef HAVE_LIBPTHREAD
 	pthread_mutex_lock(&mutex);
 # endif
@@ -103,7 +103,7 @@ fft_plan fft_plan_dft_2d(
 }
 
 void fft_plan_exec(const fft_plan plan){
-#ifdef HAVE_LIBFFTW3
+#ifdef HAVE_FFTW3
 	fftw_execute(plan->plan);
 #else
 	kiss_fftnd(plan->cfg, (const kiss_fft_cpx *)plan->in, (kiss_fft_cpx *)plan->out);
@@ -112,7 +112,7 @@ void fft_plan_exec(const fft_plan plan){
 
 void fft_plan_destroy(fft_plan plan){
 	if(NULL == plan){ return; }
-#ifdef HAVE_LIBFFTW3
+#ifdef HAVE_FFTW3
 # ifdef HAVE_LIBPTHREAD
 	pthread_mutex_lock(&mutex);
 # endif
@@ -135,7 +135,7 @@ void fft_init(){
 }
 
 void fft_destroy(){
-#ifdef HAVE_LIBFFTW3
+#ifdef HAVE_FFTW3
 	fftw_cleanup();
 #endif
 #ifdef HAVE_LIBPTHREAD

S4/fmm/fmm_FFT.cpp

@@ -32,8 +32,8 @@
 #include "fmm.h"
 
 #include <limits>
-#include <kiss_fft.h>
-#include <tools/kiss_fftnd.h>
+//#include <kiss_fft.h>
+//#include <tools/kiss_fftnd.h>
 #include "fft_iface.h"
 
 int FMMGetEpsilon_FFT(const Simulation *S, const Layer *L, const int n, std::complex<double> *Epsilon2, std::complex<double> *Epsilon_inv){

S4/fmm/fmm_PolBasisJones.cpp

@@ -19,6 +19,7 @@
 
 #include <config.h>
 
+#define _USE_MATH_DEFINES
 #include <cmath>
 #include <S4.h>
 #include "../RNP/TBLAS.h"

S4/fmm/fmm_PolBasisVL.cpp

@@ -238,7 +238,7 @@ int FMMGetEpsilon_PolBasisVL(const Simulation *S, const Layer *L, const int n, s
 		}
 
 
-		fft_plan plan = fft_plan_dft_2d(ngrid, Ffrom, Fto, -1);
+		fft_plan plan = fft_plan_dft_2d(ngrid, Ffrom, Fto, 1);
 
 		// We fill in the quarter blocks of F in Fortran order
 		for(int w = 0; w < 4; ++w){

S4/main_python.c

@@ -30,7 +30,7 @@
 #include <string.h>
 #include <stdarg.h>
 
-#ifdef WIN32
+#ifdef _WIN32
 #define _USE_MATH_DEFINES
 #endif
 
@@ -2169,7 +2169,7 @@ static PyObject *S4_NewInterpolator(PyObject *self, PyObject *args, PyObject *kw
 static PyObject *S4_SolveInParallel(PyObject *Self, PyObject *args, PyObject *kwds)
 {
 	static char *kwlist[] = {"Layer", "Simulations", NULL};
-	const char *layerName;
+	//const char *layerName;
 	//S4_solve_in
 	Py_RETURN_NONE;
 }

S4/numalloc.c

@@ -20,10 +20,10 @@
 #include <stdlib.h>
 #include <stdio.h>
 
-#ifdef WIN32
+#ifdef _WIN32
 # include <malloc.h>
-void * _aligned_malloc(size_t size, size_t alignment);
-void _aligned_free(void *ptr);
+// void * _aligned_malloc(size_t size, size_t alignment);
+// void _aligned_free(void *ptr);
 #else
 #include <inttypes.h>
 typedef uintptr_t malloc_aligned_ULONG_PTR;
@@ -33,7 +33,7 @@ typedef uintptr_t malloc_aligned_ULONG_PTR;
 //        The actual size of allocation will be greater than this size.
 // alignment : the alignment boundary
 void *malloc_aligned(size_t size, size_t alignment){
-#ifdef WIN32
+#ifdef _WIN32
 	return (void*)_aligned_malloc(size, alignment);
 #else
 	void *pa, *ptr;
@@ -51,7 +51,7 @@ void *malloc_aligned(size_t size, size_t alignment){
 }
 
 void free_aligned(void *ptr){
-#ifdef WIN32
+#ifdef _WIN32
 	_aligned_free(ptr);
 #else
 	if(ptr){

S4/pattern/predicates.c

@@ -116,7 +116,9 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <math.h>
+#ifndef _MSC_VER 
 #include <sys/time.h>
+#endif
 
 /* On some machines, the exact arithmetic routines might be defeated by the  */
 /*   use of internal extended precision floating-point registers.  Sometimes */

S4/rcwa.cpp

@@ -50,7 +50,7 @@
 #include <numalloc.h>
 
 static inline void* rcwa_malloc(size_t size){
-	void *ret = malloc_aligned(size, 16);
+	void *ret = malloc_aligned(size, 128);
 	//memset(ret, 0x0, size);
 	return ret;
 }

S4/sort.c

@@ -209,7 +209,9 @@ sort (void *const pbase, size_t total_elems, size_t size,
      of the array to sort, and END_PTR points at the very last element in
      the array (*not* one beyond it!). */
 
+#ifndef _MSC_VER
 #define min(x, y) ((x) < (y) ? (x) : (y))
+#endif
 
   {
     char *const end_ptr = &base_ptr[size * (total_elems - 1)];