improve gauss + fix latex

4 files changed, 21 insertions, 28 deletions

diff --git a/content/math/lgsFp.cpp b/content/math/lgsFp.cpp
index a4f3d0f..a8d7fc7 100644
--- a/content/math/lgsFp.cpp
+++ b/content/math/lgsFp.cpp
@@ -1,20 +1,20 @@
-vector<int> piv;
-void gauss(){
-	for(int r = 0, c = 0; c < m; c++){
-		for(int i = r; i < n; i++){
-			if(mat[i][c] != 0){
+vector<int> pivots; // ith pivot is in ith row
+void gauss(int n, int m) {
+	for (int r = 0, c = 0; c < m; c++) {
+		for (int i = r; i < n; i++) {
+			if (mat[i][c] != 0){
 				swap(mat[r], mat[i]);
 				break;
 		}}
-		if(mat[r][c] == 0) continue;
+		if (mat[r][c] == 0) continue;
 		ll f = multInv(mat[r][c], mod);
-		for(ll &x : mat[r]) x = x * f % mod;
-		for(int i = 0; i < n; i++){
-			if(i == r) continue;
+		for (ll& x : mat[r]) x = x * f % mod;
+		for (int i = 0; i < n; i++) {
+			if (i == r) continue;
 			f = mat[i][c];
-			for(int j = c; j < sz(mat[r]); j++){
+			for (int j = c; j < m; j++) {
 				mat[i][j] = (mat[i][j] - f * mat[r][j] % mod + mod) % mod;
 		}}
-		piv.push_back(c);
-		if(++r == n) break;
-}}
\ No newline at end of file
+		pivots.push_back(c);
+		if (++r == n) break;
+}} // no solution if pivots.back() == m-1
diff --git a/content/other/other.tex b/content/other/other.tex
index a38f6da..a5004fb 100644
--- a/content/other/other.tex
+++ b/content/other/other.tex
@@ -18,9 +18,9 @@
 	\begin{expandtable}

 		\begin{tabularx}{\linewidth}{|lR|}

 			\hline

-			Addition & \code{__builtin_saddll_overflow(a, b, &c)} \\

-			Subtraktion & \code{__builtin_ssubll_overflow(a, b, &c)} \\

-			Multiplikation & \code{__builtin_smulll_overflow(a, b, &c)} \\

+			Addition & \code{__builtin_saddll_overflow(a, b, \&c)} \\

+			Subtraktion & \code{__builtin_ssubll_overflow(a, b, \&c)} \\

+			Multiplikation & \code{__builtin_smulll_overflow(a, b, \&c)} \\

 			\hline

 		\end{tabularx}

 	\end{expandtable}

@@ -30,9 +30,9 @@
 	\begin{expandtable}

 	\begin{tabularx}{\linewidth}{|Ll|}

 		\hline

-		Bit an Position j lesen & \code{(x & (1 << j)) != 0} \\

+		Bit an Position j lesen & \code{(x \& (1 << j)) != 0} \\

 		Bit an Position j setzten & \code{x |= (1 << j)} \\

-		Bit an Position j löschen & \code{x &= ~(1 << j)} \\

+		Bit an Position j löschen & \code{x \&= ~(1 << j)} \\

 		Bit an Position j flippen & \code{x ^= (1 << j)} \\

 		Anzahl an führenden nullen	($x \neq 0$) & \code{__builtin_clzll(x)} \\

 		Anzahl an schließenden nullen ($x \neq 0$) & \code{__builtin_ctzll(x)} \\

diff --git a/tcr.pdf b/tcr.pdf
index eb0110f..16c465d 100644
--- a/tcr.pdf
+++ b/tcr.pdf
diff --git a/test/math/lgsFp.cpp b/test/math/lgsFp.cpp
index e529390..d7680ea 100644
--- a/test/math/lgsFp.cpp
+++ b/test/math/lgsFp.cpp
@@ -3,7 +3,6 @@
 vector<vector<ll>> mat;
 constexpr ll mod = 1'000'000'007;
 namespace lgs {
-	int n, m;
 	#include <math/lgsFp.cpp>
 }
 
@@ -16,9 +15,7 @@ vector<vector<ll>> inverseMat(const vector<vector<ll>>& m) {
 		mat[i].resize(2*n);
 		mat[i][n+i] = 1;
 	}
-	lgs::n = sz(mat);
-	lgs::m = sz(mat[0]);
-	lgs::gauss();
+	lgs::gauss(sz(mat), sz(mat[0]));
 	vector<vector<ll>> res(m);
 	for (int i = 0; i < n; i++) {
 		res[i] = vector<ll>(mat[i].begin() + n, mat[i].end());
@@ -57,9 +54,7 @@ void test_square() {
 		vector<vector<ll>> m(n);
 		for (auto& v : m) v = Random::integers<ll>(n, 0, mod);
 		mat = m;
-		lgs::n = sz(mat);
-		lgs::m = sz(mat[0]);
-		lgs::gauss();
+		lgs::gauss(sz(mat), sz(mat[0]));
 
 		for (int i = 0; i < n; i++) {
 			for (int j = 0; j < n; j++) {
@@ -104,11 +99,9 @@ void performance_test() {
 	vector<vector<ll>> m(N);
 	for (auto& v : m) v = Random::integers<ll>(N, 0, mod);
 	mat = m;
-	lgs::n = sz(mat);
-	lgs::m = sz(mat[0]);
 
 	t.start();
-	lgs::gauss();
+	lgs::gauss(sz(mat), sz(mat[0]));
 	t.stop();
 	hash_t hash = 0;
 	for (int i = 0; i < N; i++) {