Fibrations Over Singular K3 Surfaces and New Solutions to the Hull-Strominger System Anna Fino, Gueo Grantcharov, Jose Medel Abstract . Using fibrations over K3 orbisurfaces we construct new smooth solutions to the Hull-Strominger system. In particular, we prove that, for 4 ≤ k ≤ 22 4\leq k\leq 22 and 5 ≤ r ≤ 22 5\leq r\leq 22 , the smooth manifolds S 1 × ♯ k ​ ( S 2 × S 3 ) S^{1}\times\sharp_{k}(S^{2}\times S^{3}) and ♯ r ​ ( S 2 × S 4 ) ​ ♯ r + 1 ​ ( S 3 × S 3 ) \sharp_{r}(S^{2}\times S^{4})\sharp_{r+1}(S^{3}\times S^{3}) , have a complex structure with trivial canonical bundle and admit a solution to the Hull-Strominger system. 2000 Mathematics Subject Classification Primary 32J81; Secondary 53C07. 1. Introduction The Hull-Strominger system [ 44 , 28 ] is a system of nonlinear PDEs describing the geometry of compactification of heterotic strings with torsion to 4d Minkowski spacetime, which can be regarded as a generalization of Ricci-flat Kähler metrics coupled with Hermitian Yang-Mills equation on non-Kähler Calabi-Yau 3-folds. To describe the system, let M be a compact complex manifold of complex dimension 3 with holomorphically trivial canonical bundle, so that it admits a nowhere vanishing holomorphic (3,0) -form \psi . Let V be a holomorphic vector bundle over M with a Hermitian metric H along its fibers and let \alpha^{\prime}\in{\mathbb{R}} be a constant, also called the slope parameter. The Hull-Strominger system, for the fundamental form \omega of a Hermitian metric g on M , is given by: (1) \displaystyle F_{H}\wedge\omega^{2}=0;\quad F_{H}^{2,0}=F_{H}^{0,2}=0, (2) \displaystyle d(\|\psi\|_{\omega}\,\omega^{2})=0, (3) \displaystyle i\partial\bar{\partial}\omega=\frac{\alpha^{\prime}}{4}\,{\rm tr}\left(R_{\nabla}\wedge R_{\nabla}-F_{H}\wedge F_{H}\right), where F_{H} and R_{\nabla} are respectively the curvatures of H and of a metric connection \nabla on the tangent bundle TM . The equations ( 1 ) describe the Hermitian-Yang-Mills equations for the Chern connection A_{H} and the equation ( 2 ) says that \omega is conformally balanced. The equation ( 3 ) is the so-called Bianchi identity or anomaly cancellation equation. Note that in the equation ( 3 ) there is an ambiguity in the choice of a metric connection \nabla on TM , due to its origins in heterotic string theory [ 28 , 44 ] . Also from physical perspective one has \alpha^{\prime}\geq 0 with \alpha^{\prime}=0 corresponding to the Kähler case, but in mathematical literature the case \alpha^{\prime}<0 is also considered [ 41 ] . Different choices of the connection \nabla and their physical meaning are discussed in [ 9 ] . The first solutions of the Hull-Strominger system on compact non-Kähler manifolds were constructed by Fu and Yau [ 18 , 19 ] , taking as \nabla the Chern connection of \omega . The solutions are defined on toric bundles over K3 surfaces. For their construction Fu and Yau used first the result by Goldstein and Prokushkin [ 25 ] that for a Ricci-flat base and an appropriate choice of the principal torus fibration, the total space has trivial canonical bundle and has a balanced metric. Then they showed that particular solutions of the Hull-Strominger system on some principal torus fibrations on K3 manifolds can be reduced to a complex Monge-Ampère type equation for a scalar function u on the base, and solved by a continuity method type argument inspired from the techniques of Yau in [ 46 ] (see also [ 37 , 38 , 39 ] ). Since then, and the work by Li and Yau [ 35 ] , different analytical and geometrical aspects of the Hull-Strominger system have been studied with a relevant influence to non-Kähler complex geometry (see for instance [ 1 , 8 , 10 , 20 , 22 , 23 , 38 ] ). Up to now the biggest set of solutions is provided by the choice of \nabla given by the Chern connection [ 6 , 7 , 11 , 12 , 13 , 14 , 15 , 17 , 36 , 37 , 38 , 40 , 41 ] . Examples of solutions of the Hull-Strominger system on non-Kähler torus bundles over K3 surfaces with the property that the connection \nabla is Hermitian-Yang-Mills have been constructed in [ 21 ] . In [ 15 ] the Fu-Yau result have been generalized to K3 orbisurfaces, providing an extension to Hermitian 3 -folds foliated by non-singular elliptic curves. Here K3 orbisurface is a simply-connected surface with isolated cyclic singularities and trivial canonical bundle - see Section 2 for the details. In particular the following result has been established: let (X,\omega_{X}) be a compact K3 orbisurface equipped with two anti-self-dual (1,1) -forms \omega_{1} and \omega_{2} such that [\omega_{1}],[\omega_{2}]\in H^{2}_{orb}(X,\mathbb{Z}) and the total space X of the principal T^{2} orbifold bundle \pi:M\rightarrow X determined by them is smooth. If there exists a stable vector bundle E of degree 0 over (X,\omega_{X}) satisfying (4) \alpha^{\prime}(e_{orb}(X)-(c_{2}(E)-\frac{1}{2}c_{1}^{2}(E)))=\frac{1}{4\pi^{2}}\int_{X}(\|\omega_{1}\|^{2}+\|\omega_{2}\|^{2})\frac{\omega_{X}^{2}}{2}. where e_{orb}(X) denotes the orbifold Euler number, then there is a Hermitian structure \omega_{u} on the complex 3 -fold M , and a metric h along the fibers of E , such that (M,\omega_{u},V=\pi^{*}E,H=\pi^{*}(h)) solves the Hull- Strominger system. Here \omega_{u} depends on a function u on X satisfying the complex Monge-Ampére equation solved in [ 19 ] . Applying this extended result to a special type of K3 orbisurfaces and using the topological classification in [ 24 ] for compact simply-connected 6-manifolds with a free S^{1} -action, in [ 15 ] it has been shown that, for 13\leq k\leq 22 and 14\leq r\leq 22 , the smooth manifolds S^{1}\times\sharp_{k}(S^{2}\times S^{3}) and \sharp_{r}(S^{2}\times S^{4})\sharp_{r+1}(S^{3}\times S^{3}) admit a solution to the Hull-Strominger system. The cases k=22 and r=22 respectively correspond to the solutions of Fu and Yau. The examples have the structure of a principal S^{1} -bundle over Seifert S^{1} -bundles and for them \alpha^{\prime}>0 . The simply connected examples are obtained starting from a K3 orbisurface with isolated A_{1} singular points and trivial orbifold fundamental group and using partial resolution of singularities by blow-ups. In the present paper we construct new solutions to the Hull-Strominger system considering other types of K3 orbisurfaces. Our main result is the following Theorem 1.1 . Let 4\leq k\leq 22 and 5\leq r\leq 22 . Then the smooth manifolds S^{1}\times\sharp_{k}(S^{2}\times S^{3}) and \sharp_{r}(S^{2}\times S^{4})\sharp_{r+1}(S^{3}\times S^{3}) admit a complex structure with a trivial canonical bundle with a balanced metric and a solution to the Hull-Strominger system via the Fu-Yau ansatz. Note that since a quotient of a manifold under an almost free torus action is an orbifold, we can characterize the underlying simply connected compact smooth manifolds M with almost free T^{2} action which carry a solution of Hull-Strominger system and also have a transversal Calabi-Yau geometry. Since the base is a K3 orbisurface then they are diffeomorphic to \sharp_{r}(S^{2}\times S^{4})\sharp_{r+1}(S^{3}\times S^{3}) for 3\leq r\leq 22 . This follows from [ 24 ] and the topological restriction on the second Betti number b_{2} arising from the fact that the classes of the holomorphic symplectic form and the Kähler form of the base survive under the pull-back to M . The upper bound is realized by the smooth K3 surface. We note that the remaining spaces missing from Theorem 1.1 can not be realized by the methods of Fu-Yau ansatz for the surfaces in [ 30 ] . Moreover, based on the results in [ 21 ] , we note that the constructions in the present paper could likely be applied to obtain solution of the Hull-Strominger system with \nabla Hermitian-Yang-Mills. We describe shortly the structure of the paper. In Section 2 we collect the necessary information on singular K3 surfaces, which we call orbisurfaces. In Section 3 we construct the smooth principal T^{2} -orbifold bundles over the K3 orbisurfaces. In Section 4 we use the Serre construction to find the appropriate stable bundles over the K3 orbisurfaces. Finally in Section 5 we prove Theorem 1.1 by using a result in [ 15 ] and the constructions in Sections 3 and 4 . 2. K3 orbisurfaces We recall that a complex orbifold is a topological space together with a cover of coordinate charts where each element of the cover is homeomorphic to a quotient of an open subset in \mathbb{C}^{n} containing the origin by a finite group G_{x} and the transition functions are covered by holomorphic maps. We consider in the paper complex orbifolds with cyclic isotropy groups G_{x} , where x is the image of the origin after under the action. The (complex) dimension of the orbifold is the number n in \mathbb{C}^{n}/G_{x} . If n=2 the orbifold sometimes is called orbisurface and this is the terminology which we’ll be using throughout the paper. We are interested in orbisurfaces which arise as a complete intersections in weighted projective spaces and follow [ 30 ] for the notations and definitions not provided here. Definition 2.1 . An A_{n} singularity in a complex orbisurface X is an isolated singular point p\in X such that there is a neighborhood W of p with a chart W\cong\mathbb{C}^{2}/\mathbb{Z}_{n+1} , or equivalently, W\cong V(x^{2}+y^{2}+z^{n+1}) Recall that V(x^{2}+y^{2}+z^{n+1}) is the ideal generated by x^{2}+y^{2}+z^{n+1} . The main objects in this paper are orbifold analogues of K3 surfaces. Definition 2.2 . A singular K3 X is surface with only isolated singularities \Sigma=\{p_{1},...,p_{k}\} , such that H^{1}(X,\mathcal{O}_{X})=0 and \omega_{X}\cong\mathcal{O}_{X} , where \omega_{X}:=i_{*}\omega_{X_{reg}} , i:X_{reg}\rightarrow X is the inclusion, and \omega_{X_{reg}} is the canonical sheaf of X_{reg}, where X_{reg}=X\setminus\Sigma . Moreover, if we endow X with an orbifold structure \mathcal{X}=(X,\mathcal{U}) we call \mathcal{X} a K3 orbisurface . Note that the dualizing sheaf of X in our case is isomorphic to i_{*}\omega_{X\setminus\Sigma} , so the definition is consistent. Suitable K3 orbisurfaces with at worst finite isolated A_{n} singularities can be found in [ 43 ] and [ 30 ] . Their orbifold structure is inherited from their respective weighted projective spaces. The examples we consider from [ 30 ] are hypersurfaces of weighted projective spaces, them being X_{30}\subset\mathbb{P}(5,6,8,11) , X_{36}\subset\mathbb{P}(7,8,9,12) , and X_{50}\subset\mathbb{P}(7,8,10,25) , where the d in X_{d} is the degree of the hypersurface and a_{0},a_{1},a_{2},a_{3} in \mathbb{P}(a_{0},a_{1},a_{2},a_{3}) are the weights on the variables of the projective space. The singularities of X_{36} are described in [ 30 ] . Here we describe the singularities of X_{30} and X_{50} . In general, a well-formed \mathbb{P}(a_{0},a_{1},a_{2},a_{3}) has singularities on each of its affine pieces x_{i}\neq 0 , which are isomorphic to \mathbb{C}^{2}/\mathbb{Z}_{n+1} where n+1 is the weight of x_{i} . They have singular edges when the gcd(a_{i},a_{j})=d>1 , and the singular edges are isomorphic to \mathbb{P}(a_{i},a_{j})\cong\mathbb{P}(\frac{a_{i}}{d},\frac{a_{j}}{d}) . The hypersurface X_{30}\subset\mathbb{P}(5,6,8,11) is given by a general polynomial f=w^{6}+x^{5}+yz^{2}+xy^{3}+w^{2}x^{2}y+wxyz with (x,y,z,w) having weights (5,6,8,11) , where for simplicity we consider all the coefficients in the monomials to be ones. This surface is well-formed and quasismooth (see [ 30 ] ), so its only possible singularities are inherited from the ones in \mathbb{P}(5,6,8,11) . Since f has the monomials w^{6} and x^{5} , when we restrict to w\neq 0 and x\neq 0 , f restricts to a polynomial of the form f(1,x,y,z)=1+... and f(w,1,y,z)=1+... thus the singular points (1:0:0:0), and (0:1:0:0) are not in the surface. On the contrary, the polynomial f does not have monomials only on y and z , thus the singular points (0:0:1:0), and (0:0:0:1) are in the surface, which are of type A_{7} , and A_{10} respectively. Along the singular edge \mathbb{P}(6,8)\cong\mathbb{P}(3,4) we have f=x^{4}+y^{3} , and with Lemma 9.4 in [ 30 ] we calculate that X_{30} intersects at \lfloor\frac{12}{3\cdot 4}\rfloor=1 points. Following 5.15 in [ 30 ] this is an A_{1} singularity since 1=gcd(6,8)-1 from \mathbb{P}(6,8) . Similarly the hypersurface X_{50}\subset\mathbb{P}(7,8,10,25) is given by the polynomial f=z^{2}+y^{5}+w^{6}x+x^{5}y+w^{2}+x^{2}+y^{2}+wxyz. By the same reasoning as before, we have the monomials z^{2} and y^{5} , so X_{50} does not contain the singular points (0:0:1:0) and (0:0:0:1) , it does contain the points (1:0:0:0) and (0:1:0:0) with A_{6} and A_{7} singularities. Along the edge \mathbb{P}(8,10)\cong\mathbb{P}(4,5) , we have f=y^{4}+x^{5} , so X_{50} intersects the singular edge at \lfloor\frac{20}{4\cdot 5}\rfloor=1 point. Again from 5.15 in [ 30 ] we conclude that this singular point is of type gcd(8,10)-1 , hence an A_{1} singularity. Along the edge \mathbb{P}(10,25)\cong\mathbb{P}(2,5) , we have f=z^{2}+y^{5} which intersects the singular edge at \lfloor\frac{10}{2\cdot 5}\rfloor=1 point giving an A_{4} singularity. In summary, X_{30} has A_{1},A_{7}, and A_{10} singularities; X_{36} has A_{6},A_{7},A_{3} singularities, and A_{2} ; X_{50} has A_{6},A_{7},A_{1}, and A_{4} singularities. In Theorem 5.1 we’ll need Cartier divisors corresponding to characteristic classes \omega_{1},\omega_{2} used for the construction of the principal T^{2} -bundle \pi:M\rightarrow X . To achieve that we first have to calculate the Picard groups of the K3 orbisurfaces. The following is a known fact, we provide a short proof for completeness. Lemma 2.1 . Let X be a complex orbisurface with at worst isolated A_{n} singularities, and \pi:\tilde{X}\rightarrow X a blow up at a singular point. Then we have two cases: for n>1 we have {\text{Pic}}(\tilde{X})\cong{\text{Pic}}(X)\oplus\langle\mathcal{L},\mathcal{L}^{\prime}\rangle; for n=1 we have {\text{Pic}}(\tilde{X})\cong{\text{Pic}}(X)\oplus\langle\mathcal{G}\rangle, where \mathcal{L} is (n-1)C , \mathcal{L}^{\prime} is C+C^{\prime} in the case of n>1 and \mathcal{G}=C in the case of n=1 . C and C^{\prime} are exceptional curves. Proof. We have an exact sequence 0\rightarrow{\text{Pic}}(X)\underset{\pi^{*}}{\rightarrow}{\text{Pic}}(\tilde{X})\underset{i_{B}^{*}}{\rightarrow}{\text{Pic}}(B)\rightarrow 0 where B:=\pi^{-1}(p) , the exceptional divisor, and i_{B}:B\rightarrow\tilde{X} is the inclusion. We also have a projection map \varphi:\tilde{X}\rightarrow B induced by the inclusion and projection \tilde{X}\subset X\times\mathbb{P}^{2}\rightarrow\mathbb{P}^{2} whose image is B . The pullback \varphi^{*}:{\text{Pic}}(B)\rightarrow{\text{Pic}}(\tilde{X}) splits the exact sequence, thus {\text{Pic}}(\tilde{X})\cong{\text{Pic}}(X)\oplus{\text{Pic}}(B) . Now, in the case of an A_{1} singularity, the exceptional curve is a single curve isomorphic to \mathbb{P}^{1} . Indeed, after blowing up the singular point of the variety V(x^{2}+y^{2}-z^{2}) we obtain the exceptional curve \pi^{-1}(p)\cong V(x^{2}+y^{2}-1). In the case of A_{n} with n>1 the exceptional curve is \pi^{-1}(p)\cong V(x^{2}+y^{2}) , which corresponds to two projective lines C,C^{\prime} intersecting in one point. Computing the intersection (n-1)C\cdot C^{\prime}=1 we conclude that the minimal number to make Cartier divisors from the Weil divisors C or C^{\prime} is n-1. Thus \mathcal{L}=(n-1)C , \mathcal{L}^{\prime}=C+C^{\prime} serve us as a basis for Pic(B) . ∎ We also need the following: Lemma 2.2 . Let X be a generic K3 hypersurface of codimension 1 on a weighted projective space. Then {\text{Pic}}(X)\cong\langle\mathcal{O}_{X}(n)\rangle\cong\mathbb{Z} for some n . Moreover, in the orbifold category we have {\text{Pic}}^{orb}(X)=\langle\mathcal{O}_{\mathcal{X}}(1)\rangle. The calculation {\text{Pic}}(X)\cong\mathbb{Z} for generic K3 hypersurfaces can be found in [ 4 ] , and it follows that {\text{Pic}}^{\mathrm{orb}}(X)=\langle\mathcal{O}_{\mathcal{X}}(1)\rangle . The orbifold Picard group in this case is the group of rational Cartier divisors. Next we calculate the minimal n for which {\text{Pic}}(X)\cong\langle\mathcal{O}_{X}(n)\rangle for the surfaces X_{30},X_{36} , and X_{50} . In general, our approach is to find a polynomial of degree d_{2} such that the curve in our surface of degree d_{1} , X_{d_{1},d_{2}}\subset X_{d_{1}} avoids the singularities of X_{d_{1}} , thus defining a Cartier divisor. In the case of X_{30}\subset\mathbb{P}(5,6,8,11) we have singularities at the points (0:0:1:0) , (0:0:0:1) and one point along the singular edge \mathbb{P}(6,8) . We can avoid the singular points by choosing a polynomial which has monomials of the type y^{11} and z^{8} . Thus we choose the polynomial g=y^{11}+z^{8}+\ldots of degree 88 to avoid the singular points. The curve X_{88,30} will intersect the singular edge \mathbb{P}(6,8) , but we can choose the coefficients of g such that it avoids the point where X_{30} passes through. Thus we claim that the sheaf \mathcal{O}_{X_{30}}(88) is the generator of {\text{Pic}}(X_{30}) . Indeed, it can’t be a lower degree since it would have to divide 88 , but by dividing the degree the corresponding curve would have to go through the singular points. Now, we calculate the self intersection of H:=\mathcal{O}_{X_{30}}(88) . From the adjunction formula we have \omega_{X_{88,30}}=\mathcal{O}_{X_{88,30}}(88), thus the genus \overline{g} of X_{88,30} is \dim H^{0}(\omega_{X_{88,30}})=\dim H^{0}(\mathcal{O}_{X_{88,30}}(88))=45. We obtain the dimension of global sections of \mathcal{O}_{X_{88,30}}(88) by counting all monomials of degree 88 of the ring \mathbb{C}[w,x,y,z]/(f,g) given that w,x,y,z have weights 5,6,8,11 respectively. Finally, we have {\text{deg}}(\omega_{X_{88,30}})=\omega_{X_{30}}\cdot H+H\cdot H, so 2\overline{g}-2=H^{2} , and H^{2}=88. Using the same procedure for X_{36} and X_{50} we have {\text{Pic}}(X_{36})=\langle\mathcal{O}_{X_{36}}(56)\rangle,\text{ and }{\text{Pic}}(X_{50})=\langle\mathcal{O}_{X_{50}}(56)\rangle. The self intersections are \mathcal{O}_{X_{36}}(56)^{2}=18,\text{ and }\mathcal{O}_{X_{50}}(56)^{2}=10. This gives us right away the intersections in the orbifold case, which will be \mathcal{O}_{\mathcal{X}_{30}}(1)^{2}=\frac{1}{88},\,\mathcal{O}_{\mathcal{X}_{36}}(1)^{2}=\frac{9}{1568},\,\mathcal{O}_{\mathcal{X}_{50}}(1)^{2}=\frac{5}{1568}. The following lemma follows from Example 18.3.4b in [ 16 ] . Lemma 2.3 . (Noether’s formula for singular surfaces) Let X be a complex orbisurface with at worst finite A_{n} singularities say \{A_{n_{1}},...,A_{n_{j}}\} . Then \chi(X,\mathcal{O}_{X})=\frac{1}{12}\left(K_{X}^{2}+\chi_{top}(X)+\sum_{i=1}^{j}n_{i}\right), where \chi(X,\mathcal{O}_{X}) is the Euler characteristic of the sheaf cohomology of \mathcal{O}_{X} and \chi_{top}(X) is the topological Euler characteristic of X . As a consequence we get: Corollary 2.1 . Let X be a complex K3 orbisurface with at worst finite A_{n} singularities say \{A_{n_{1}},...,A_{n_{j}}\} . Then \chi_{top}(X)=24-\sum_{i=1}^{j}n_{i}. Moreover, X has the following orbifold Betti numbers: b_{0}=b_{4}=1\text{, }b_{1}=b_{3}=0\text{, and }b_{2}=22-\sum_{i=1}^{j}n_{i}. Proof. Firstly, observe that \chi(X,\mathcal{O}_{X})=2 by the fact that Dolbeault isomorphism is still true for orbifolds [ 2 ] . Indeed, H^{2}(X,\mathcal{O}_{X})\cong H^{0}(X,\omega_{X})\cong H^{0}(X,\mathcal{O}_{X})\cong\mathbb{C}, and H^{1}(X,\mathcal{O}_{X})=0 by definition of K3 orbisurface. So we have 2=\frac{1}{12}\left(\chi_{top}(X)+\sum_{i}n_{i}\right). Now, the Hodge numbers are h^{i,j}=\dim H^{j}(X,\Omega^{i}_{X}), h^{i,j}=h^{j,i} , and b_{k}=\sum_{i+j=k}h^{i,j} . Thus, b_{1}=h^{0,1}+h^{1,0}=2h^{0,1}=2\dim H^{1}(X,\mathcal{O}_{X})=0 and by Poincaré duality b_{1}=b_{3} . Similarly, b_{0}=dimH^{0}(X,\mathcal{O}_{X})=1 and b_{0}=b_{4} . Finally, \chi_{top}(X)=2+b_{2} , thus b_{2}=22-\sum_{i}n_{i}. ∎ Corollary 2.2 . Let X be a K3 orbisurface in Miles Reid’s 95 list [ 43 ] , let \pi:\tilde{X}\rightarrow X be a chain of blow ups of the A_{n} singular points, where k is the number of irreducible exceptional curves. Then {\text{Pic}}(\tilde{X})\cong\langle\pi^{*}H\rangle\oplus\mathbb{Z}^{k}, and {\text{Pic}}^{orb}(\tilde{X})\cong\langle\mathcal{O}_{\tilde{\mathcal{X}}}(1)\rangle\oplus\mathbb{Z}^{k} Proof. This follows from applying Lemma 2.1 as many times as needed. Also, the information from Corollary 2.1 would give us the second orbifold Betti number. This lemmas also apply for the Pic^{orb}(\tilde{\mathcal{X}}) . ∎ Remark 2.1 . Note that the intersections of the exceptional divisors \mathcal{L}=(n-1)C and \mathcal{L}^{\prime}=C+C^{\prime} of the blow up of an A_{n} singular point with n>1 are \mathcal{L}^{2}=-n(n-1) , \mathcal{L}\cdot\mathcal{L}^{\prime}=-(n-1) . Lemma 2.4 . Let X be X_{36}\subset\mathbb{P}(7,8,9,12) or X_{50}\subset P(7,8,10,25) , and let \pi:\tilde{X}\rightarrow X be a chain of blow ups of singular points. Then \tilde{X} has an ample divisor \mathcal{Q} such that \mathcal{Q}^{2}=2 , and the set of effective divisors D of \tilde{X} such that D\cdot\mathcal{Q}\leq 2 is empty. Proof. Let D be an effective divisor of \tilde{X} . The map \pi induces two homomorphims \pi^{*}:{\text{Pic}}(X)\rightarrow{\text{Pic}}(\tilde{X}) and \pi_{*}:Cl(\tilde{X})\rightarrow Cl(X) , where Cl(Y) is the divisor class group of Weil divisors in Y modulo linear equivalence. This last map extends by tensoring over \mathbb{Q} to be \pi_{*}:Cl(\tilde{X})_{\mathbb{Q}}\rightarrow Cl(X)_{\mathbb{Q}} . Let H be the generator of {\text{Pic}}(X) , and note that all divisors are \mathbb{Q} -Cartier [ 5 ] , so \mathbb{Q}\cong{\text{Pic}}(X)_{\mathbb{Q}}\cong Cl(X)_{\mathbb{Q}} . The ample divisor \mathcal{Q} will have the form \mathcal{Q}=a_{0}\pi^{*}H+\ldots, where the rest of the summands will be divisors that come from the blow ups and a_{0} a positive integer. Over Chow rings we have the formula \pi_{*}(\pi^{*}H\cdot[D])=H\cdot\pi_{*}D. The divisor \pi_{*}D is a positive linear combination of elements of Cl(X) or zero, so assume its non zero and let \bar{H} be an arbitrary non-zero summand. Since \bar{H} is \mathbb{Q} -Cartier we have that there are integers m and k such that m\bar{H}=kH with m,k coprime and m>k. Thus we have that \bar{H}=\frac{k}{m}H , and H\cdot\bar{H}=\frac{k}{m}H^{2}\in\mathbb{Z}. Since this number is an integer it means that m divides kH^{2} , but k,m are coprime so m divides H^{2} . Here we have two cases, one for X_{36} and one for X_{50} where H^{2} is 18 and 10 respectively. In both cases the smallest H\cdot\bar{H} we can get is when k=1 and m=H^{2} , which makes H\cdot\bar{H}=1, thus \mathcal{Q}\cdot[D]\geq a_{0}\pi^{*}H\cdot D=a_{0}H\cdot\pi_{*}D=a_{0}H\cdot\bar{H}=a_{0}, so as long as a_{0}>2 , \mathcal{Q}\cdot[D]>2 . Now, if \pi_{*}D=0 , then D is a positive linear combination of exceptional curves. Exceptional curves only intersect non-trivially with the curves that arise from the same blow up. Again, we have separate cases. One case is when it is a blow up of an A_{n} singularity with n>1 , and a second case is when it is a blow up of an A_{1} singularity. We start with the first case. From blowing up a singularity we obtain two generators \mathcal{L},\mathcal{L}^{\prime}\in{\text{Pic}}(\tilde{X}) and two curves C,C^{\prime}\in Cl(\tilde{X}) , with relations \mathcal{L}=(n-1)C and \mathcal{L}^{\prime}=C+C^{\prime} . Their intersections are \mathcal{L}\cdot C=-n,\mathcal{L}\cdot C^{\prime}=1,\mathcal{L}^{\prime}\cdot C=-1,\mathcal{L}^{\prime}\cdot C^{\prime}=-1, so \displaystyle(a\mathcal{L}+b\mathcal{L}^{\prime})\cdot C \displaystyle=-na-b, \displaystyle(a\mathcal{L}+b\mathcal{L}^{\prime})\cdot C^{\prime} \displaystyle=a-b. We must choose a and b such that -na-b>2 and a-b>2 . So a and b as coefficients of \mathcal{L},\mathcal{L}^{\prime} in \mathcal{Q} must satisfy the former inequalities. In the case of A_{1} we only have the \mathcal{L}^{\prime}=C generator, the exceptional divisor. \mathcal{L}^{\prime}\cdot C=-2 , so a\mathcal{L}^{\prime}\cdot C=-2a , hence we need -2a>2 . For all the cases we have found sufficient conditions on \mathcal{Q} such that \mathcal{Q}\cdot D>2 for any effective divisor. Table 1 contains a list of divisors \mathcal{Q} that satisfy the conditions indexed by Euler characteristic of the surface they are over. ∎ 3. Construction of T^{2} -bundles over orbisurfaces In this section we focus on T^{2} -bundles over an orbisurface X . The general theory of such spaces from the foliations view-point is given in [ 27 ] . In particular, such bundles are determined by two rational divisors on the base orbifold and are constructed via Seifert S^{1} -bundles. For the definition of Seifert S^{1} -bundle we refer to [ 5 , Definition 4.7.6] . Roughly speaking, Seifert S^{1} -bundles are spaces with a locally free S^{1} -action, for which the S^{1} -foliation has an orbifold leaf space. A multiple leave is an S^{1} -orbit on which the action is not globally free. We recall the following result, which is Theorem 4.7.3 in [ 5 ] and has been proven by Kollár in [ 32 ] . Theorem 3.1 . Let X be a normal reduced complex space with at worst quotient singularities and \Delta=\sum_{i}(1-\frac{1}{m_{i}})D_{i} be a \mathbb{Q} divisor (this is the data associated to an orbifold). Then there is a one-to-one correspondence between Seifert {\mathbb{C}}^{*} -bundles f:Y\rightarrow(X,\Delta) and the following data: (i) For each D_{i} an integer 0\leq b_{i}1 , and denote \bar{H}:=\mathcal{O}_{\mathcal{X}}(1) as well as its pullbacks through the chain of blow ups. Assume that the Seifert S^{1} -bundle given by c_{1}=\bar{H} is smooth. Then there is an ample divisor \mathcal{E}\in{\text{Pic}}^{orb}(\tilde{\mathcal{X}})\otimes\mathbb{Q} , and \mathcal{D},\mathcal{D}^{\prime}\in{\text{Pic}}^{orb}(\tilde{\mathcal{X}}) such that (1) \mathcal{D},\mathcal{D}^{\prime} are primitive with respect to \mathcal{E} , i.e. \mathcal{D}\cdot\mathcal{E}=0,\mathcal{D}^{\prime}\cdot\mathcal{E}=0 ; (2) if K_{\mathcal{X}}=0 and \pi^{orb}_{1}(X)=1 , then K_{\tilde{\mathcal{X}}}=0 and \pi^{orb}_{1}(\tilde{\mathcal{X}})=1 ; (3) the Seifert S^{1} -bundles f_{1}:Y_{1}\rightarrow\tilde{\mathcal{X}} with c_{1}(Y_{1}/\tilde{\mathcal{X}})=\mathcal{D} , and f_{2}:Y_{2}\rightarrow Y_{1} with c_{1}(Y_{2}/Y_{1})=\pi^{*}\mathcal{D}^{\prime} are smooth and H_{1}(Y_{1},\mathbb{Z})=0 and H_{1}(Y_{2},\mathbb{Z})=0. Proof. We only need to check the case when the last blow up \varphi:\tilde{\mathcal{X}}\rightarrow\mathcal{Y} in the chain of blow ups \pi:\tilde{\mathcal{X}}\rightarrow\mathcal{X} is of an A_{n} singularity with n>1. We consider the divisors to be of the form: \displaystyle\mathcal{E} \displaystyle:=\frac{k(an-b)(bn-a)}{(n-1)\bar{\mathcal{E}}\cdot\bar{H}}(\varphi^{*}\bar{\mathcal{E}})-aC-bC^{\prime}, \displaystyle\mathcal{D} \displaystyle:=\bar{H}-k(bn-a)C, \displaystyle\mathcal{D}^{\prime} \displaystyle:=\bar{H}-k(an-b)C^{\prime}; where n is the index in A_{n} , a=mn+c-1 , b=c+m , with k,m\in\mathbb{Z}^{+} to be determined later. We first check that \mathcal{E} is indeed an ample divisor via the Nakai–Moishezon ampleness criterion which says that \mathcal{E} is ample if \mathcal{E}^{2}>0 and \mathcal{E}\cdot D>0 for any irreducible curve D in \tilde{\mathcal{X}} . Any irreducible curve D will either be the proper preimage of a curve in \mathcal{Y} making \mathcal{E}\cdot D=q(\varphi^{*}\bar{\mathcal{E}}\cdot D)=q(\bar{\mathcal{E}}\cdot\varphi_{*}D)>0 where q is the rational number in front of \varphi^{*}\bar{\mathcal{E}} in \mathcal{E} , or it will be C or C^{\prime} making \mathcal{E}\cdot D=\mathcal{E}\cdot C . The expression q(\bar{\mathcal{E}}\cdot\varphi_{*}D)>0 is true because we assumed that \bar{\mathcal{E}} is ample and q is a positive number. So we only have to check for the other cases, and if \mathcal{E}^{2}>0 we get: \displaystyle\mathcal{E}\cdot C \displaystyle=-aC^{2}-bC\cdot C^{\prime} \displaystyle=\frac{an-b}{n-1}=\frac{n(mn-1)+c(n-1)-m}{n-1}>0 \displaystyle\mathcal{E}\cdot C^{\prime} \displaystyle=-aC\cdot C^{\prime}-bC^{\prime 2} \displaystyle=\frac{bn-a}{n-1}=\frac{c(n-1)+1}{n-1}>0 \displaystyle\mathcal{E}^{2} \displaystyle=\frac{k^{2}(an-b)^{2}(bn-a)^{2}\bar{\mathcal{E}}^{2}}{(n-1)^{2}(\bar{\mathcal{E}}\cdot\bar{H})^{2}}-\left(\frac{a(an-b)+b(bn-a)}{n-1}\right)>0, the last inequality is because we can choose k to be big enough. Now we show that \mathcal{D} and \mathcal{D}^{\prime} are primitive with respect to \mathcal{E} . This follows from: \displaystyle\mathcal{E}\cdot\mathcal{D} \displaystyle=\mathcal{E}\cdot\bar{H}-k(bn-a)\mathcal{E}\cdot C \displaystyle=\frac{k(an-b)(bn-a)}{(n-1)\bar{\mathcal{E}}\cdot\bar{H}}\bar{\mathcal{E}}\cdot\bar{H}-\frac{k(bn-a)(an-b)}{n-1}=0, \displaystyle\ \mathcal{E}\cdot\mathcal{D}^{\prime} \displaystyle=\mathcal{E}\cdot\bar{H}-k(an-b)\mathcal{E}\cdot C^{\prime} \displaystyle=\frac{k(an-b)(bn-a)}{n-1}-k(an-b)\frac{bn-a}{n-1}=0. The second property, i.e. K_{\tilde{\mathcal{X}}}=0 and \pi^{orb}_{1}(\tilde{\mathcal{X}})=1 , follows from the fact that A_{n} singularities are characterized by the property K_{\tilde{\mathcal{X}}}\cong\pi^{*}K_{\mathcal{X}} for \pi:\tilde{\mathcal{X}}\rightarrow\mathcal{X} a blow-up at the singularity. By definition of \pi^{orb}_{1} in [ 33 ] and [ 45 ] , blowing up can only remove classes from the orbifold fundamental group of the base. Since we have \pi^{orb}_{1}(\mathcal{X})=1 , then \pi^{orb}_{1}(\tilde{\mathcal{X}})=1 . The condition of smoothness of S^{1} -Seifert bundles is given in Kollar [ 33 ] and Boyer-Galicki [ 5 ] . We have to check if \mathcal{D} and \mathcal{D}^{\prime} are generators of the local class groups at all the points of \tilde{X} . For the singular points not involved in the blow up, \bar{H} is a generator of their local class group since \bar{H} defines a smooth S^{1} -Seifert bundle by assumption. \mathcal{D} and \mathcal{D}^{\prime} restricted to those points are equal to \bar{H} . For the remaining singular point (that is when n>2 since if n=2 the blow-up of that singular point will be smooth in that neighborhood), we just need to check if bn-a and an-b are coprime to n-1 since the local class group of the remaining singularity is \mathbb{Z}/(n-1)\mathbb{Z} . We control k which is multiplying both bn-a and an-b so we can just make it coprime to n-1 . Observe that bn-a=c(n-1)+1 , and an-b=c(n-1)+(n^{2}-1)m-n , so an-b\text{ (mod }n-1)=-n\text{ (mod }n-1)=n-2\text{ (mod }n-1); and bn-a\text{ (mod }n-1)=c(n-1)+1\text{ (mod }n-1)=1\text{ (mod }n-1). Thus the S^{1} -Seifert bundles are smooth. Finally, we have to check if \mathcal{D} and \mathcal{D}^{\prime} define S^{1} -Seifert bundles over \tilde{X} that are simply connected. The conditions for this are given by Kollár [ 33 ] , and for our case we just have to check if there are elements \alpha,\beta in H_{2}(\tilde{X},\mathbb{Z}) such that gcd(\mathcal{D}\cap\alpha,\mathcal{D}\cap\beta)=1 , same for gcd(\mathcal{D}^{\prime}\cap\alpha,\mathcal{D}^{\prime}\cap\beta)=1 , with \alpha,\beta not necessarily the same for \mathcal{D} and \mathcal{D}^{\prime} . Consider H and (n-1)C^{\prime} for \mathcal{D} , we have \mathcal{D}\cdot H=H^{2}=c and -\mathcal{D}\cdot(n-1)C^{\prime}=(bn-a)C\cdot(n-1)C^{\prime}=bn-a=c(n-1)+1, fulfilling the requirement. Now consider H and (n-1)C for \mathcal{D}^{\prime} , we have \mathcal{D}^{\prime}\cdot H=H^{2}=c and -\mathcal{D}^{\prime}\cdot(n-1)C=(an-b)C^{\prime}\cdot(n-1)C=an-b. The expression an-b=c(n-1)+(n^{2}-1)m-n\text{ (mod }c)=(n^{2}-1)m-n\text{ (mod }c), so we need (n^{2}-1)m-n to be coprime to c . By Dirichlet’s theorem on arithmetic progressions, which applies since gcd(n^{2}-1,n)=1 , we have that there are infinitely many primes for different m of the form (n^{2}-1)m-n , thus we can always choose one such that (n^{2}-1)m-n is coprime to c . ∎ Remark 3.1 . Note that, by the classification in [ 24 ] , the T^{2} -bundles over K3 orbisurfaces X we consider must be connected sums of the form as in Theorem 1.1, where k and r correspond respectively to b_{2}^{orb}(X)-1 and b_{2}^{orb}(X)-2 . Using the lists in [ 30 ] and [ 43 ] we see that k and r respectively are at least 4 and 5 . 4. Existence of Stable Bundle The solutions of Hull-Strominger system involve a stable bundle with a Hermitian-Einstein structure. So in this section we will consider a complex algebraic orbisurface X and use the Serre construction to find the appropriate stable bundles over it. On one side if we consider X simply as a complex algebraic surface, we have the category of sheaves over X , and the abelian category of coherent sheaves over X . There are analogues if we give X an orbifold structure. We begin with the definition of stability in the case of normal singular surfaces. For this we need a good definition of c_{1}(\mathcal{F}) for a coherent sheaf \mathcal{F} as in [ 3 ] . Observe that for the divisor class group we have an isomorphism Cl(X)\cong Cl(X_{reg}) since codim(\Sigma)>1 . So, we can define c_{1}(\mathcal{F}) as the image under the isomorphism of c_{1}(\mathcal{F}|_{X_{reg}}) , where c_{1}(\mathcal{F}|_{X_{reg}})=c_{1}(det(\mathcal{F}|_{X_{reg}}^{\vee\vee})) . Definition 4.1 . Let \mathcal{F} be a coherent orbisheaf over the orbisurface X , and let \mathcal{Q} be an ample divisor over X . Then the slope of \mathcal{F} with respect of \mathcal{Q} is \mu_{\mathcal{Q}}(\mathcal{F}):=\frac{c_{1}(\mathcal{F})\cdot\mathcal{Q}}{rk(\mathcal{F})}. A locally free sheaf \mathcal{E} is called stable if for any subsheaf \mathcal{F}\subset\mathcal{E} with rk(\mathcal{F})1 there exists on \tilde{X}_{k} a stable bundle E of rank 2 with c_{1}(E)=0 and c_{2}(E)=c for any c\geq 5 . The case when k=1 will be discussed after this lemma. In that case we blow up an A_{1} singularity, and it becomes impossible to use the following methods to find an ample divisor Q such that c_{2}(E)=2+Q^{2}\ell_{1} , then by the Cayley–Bacharach condition implies that E^{\prime} is a locally free sheaf of rank 2 . Suppose that N\subset E^{\prime} was a destabilizing line bundle. Then \mu_{Q}(N)\geq\mu_{Q}(E^{\prime})=Q^{2}>0=\mu_{Q}(\mathcal{O}_{\tilde{X}}), which implies that N can not be contained in \mathcal{O}_{\tilde{X}}. As a consequence the composition map N\rightarrow E^{\prime}\rightarrow 2Q\otimes\mathcal{I}_{Z} is non zero and it vanishes along D with Z\subset D and D\cdot Q=\mu_{Q}(2Q)-\mu_{Q}(N)\leq Q^{2}=2. By the Lemma 2.4 we assumed such D does not exist. ∎ We note that the lemma is independent of the orbifold structure. That is, the vector bundle E (locally free sheaf) is independent of the orbifold structure over \tilde{X}_{1} with k>1 . It may construct a stable vector bundle for the case when k=1 , but in that case any ample divisor Q would have Q^{2}>2 , which in turn will make c_{2}(E)>5 and e_{orb}(\tilde{X}_{1})-c_{2}(E)<0 thus making the slope parameter \alpha^{\prime}<0 . We note here that we can construct many stable bundles E for which \alpha^{\prime}<0 . Existence of solutions for the Hull-Strominger system in this case follows in the same way as for \alpha^{\prime}>0 in [ 18 ] . By Theorem 1 in [ 40 ] one even has a solution for which the function e^{u} can have pre-assigned integral as long as it is large enough. However, as in [ 18 ] we consider only the case \alpha^{\prime}>0 . The following lemma solved this issue. Lemma 4.2 . Let X=X_{30}\subset\mathbb{P}(5,6,8,11) which has A_{1},A_{7},A_{10} and let \pi:\tilde{X}\rightarrow X the blow up of the A_{1} singularity, and let \tilde{\mathcal{X}}=(\tilde{X},\mathcal{U}) be the “canonical” orbifold structure. Then there exist a stable vector orbibundle (called also V-bundle) E over \tilde{\mathcal{X}} with c_{1}^{orb}(E)=0 and c_{2}^{orb}(E)=c-\frac{2}{11} for every c\geq 3 . Proof. Let \bar{H}=\mathcal{O}_{\mathcal{X}}(1) , and let Q=40\pi^{*}\bar{H}-3\mathcal{L}^{\prime} , where \mathcal{L}^{\prime} is the exceptional divisor of \pi . The divisor Q is ample and Q^{2}=\frac{2}{11} . Just as before, let \ell_{1}=h^{0}(2Q)=\chi(2Q) and \displaystyle\chi(2Q) \displaystyle=\chi_{orb}(2Q)+\mu_{Sing}(2Q) \displaystyle=\chi_{orb}(\mathcal{O}_{\tilde{\mathcal{X}}})+\frac{1}{2}(2Q)^{2}+\mu_{Sing}(2Q) \displaystyle=\chi(\mathcal{O}_{\tilde{\mathcal{X}}})-\mu_{Sing}(\mathcal{O}_{\tilde{\mathcal{X}}})+2Q^{2}+\mu_{Sing}(2Q) \displaystyle=2-\left(\frac{10}{11}+\frac{21}{32}\right)+\frac{4}{11}+\mu_{Sing}(2Q), where \mu_{sing} is the correction term in the orbifold Riemann-Roch formula (see Remark 4.3 ). Observe that \mu_{Sing}(2Q)\leq\mu_{Sing}(\mathcal{O}_{\tilde{\mathcal{X}}})=\frac{10}{11}+\frac{21}{32} , so \displaystyle\chi(2Q) \displaystyle\leq 2-\mu_{Sing}(\mathcal{O}_{\tilde{\mathcal{X}}})+\frac{4}{11}+\mu_{Sing}(\mathcal{O}_{\tilde{\mathcal{X}}}) \displaystyle=2+\frac{4}{11}<3=\ell. If we apply Serre’s construction to L=\mathcal{O}_{\tilde{\mathcal{X}}} , N=2Q , and Z such that \ell(Z)=3 , then we obtain a vector orbibundle E^{\prime} with c_{1}^{orb}(E^{\prime})=2Q , and c_{2}^{orb}(E^{\prime})=\ell(Z)=3 . Same as before, assume there is an destabilizing sheaf N\subset E^{\prime} . The map N\rightarrow E^{\prime}\rightarrow 2Q\otimes\mathcal{I}_{Z} vanishes at an effective divisor D with D\cdot Q=2Q\cdot Q-N\cdot Q\leq 2Q\cdot Q=\frac{4}{11} but there are no effective divisors with D\cdot Q\leq\frac{4}{11} . The tensor product of a stable bundle by a line bundle is still stable, so define E:=E^{\prime}\otimes(Q^{\vee}) , so c_{1}^{orb}(E)=0 , and c_{2}^{orb}(E)=\ell(Z)-Q^{2}=3-\frac{2}{11}. ∎ Remark 4.2 . We check that Q is ample by the Nakai-Moishezon criterion. In the case of \tilde{\mathcal{X}} we just have to check Q\cdot\bar{H}=\frac{19}{88} , Q\cdot\mathcal{L}^{\prime}=18 , and Q^{2}=\frac{2}{11} . Also, observe that any effective divisor D will have Q\cdot D\geq Q\cdot\bar{H}=\frac{40}{88}>\frac{4}{11}=2Q^{2}. We obtained the divisor Q the following way. Let Q=a\pi^{*}H_{0}-b\mathcal{L}^{\prime} such that Q\cdot\bar{H}>2Q^{2} , so we found positive integer solutions for \frac{1}{88}a>\frac{1}{44}a^{2}-4b^{2}>0 . Remark 4.3 . From [ 3 ] we get that the correction term in the orbifold Riemann-Roch formula is given by \mu_{Sing}(\mathcal{F}):=\underset{x\in Sing}{\sum}\mu_{\mathcal{X},x}(\mathcal{F)}, where \mu_{\mathcal{X},x}(\mathcal{F}):=\frac{1}{\#G_{x}}\cdot\sum_{g\in G_{x}-\{\text{id}\}}\frac{\text{trace}(\rho(g))}{\det(I_{n}-g)}, and \#G_{x} denotes the cardinality of the finite group G_{x} . Proposition 2.6 in [ 3 ] establishes a bijection of V-free sheaves or free orbisheaves over a quotient germ (X,x) , and representations of the local isotropic group G_{x} up to isomorphism. In the formula \rho is the representation corresponding to \mathcal{F} and g in the local action of g in the coordinates of the local smoothing of the orbifold. For the given surface \tilde{\mathcal{X}} which has A_{7},A_{10} singularities, the maximum possible \mu_{Sing}(\mathcal{F}) for \mathcal{F} rank 1 is \frac{10}{11}+\frac{21}{32} . This is achieved by computing all possible \mu_{\mathcal{X},x}(\mathcal{F}) , which are \#G many. Remark 4.4 . Integrating equation ( 3 ) we get \alpha^{\prime}\int_{\tilde{\mathcal{X}}}(\text{tr}(R\wedge R)-\text{tr}(F_{H}\wedge F_{H}))=\alpha^{\prime}(e_{orb}(\tilde{\mathcal{X}})-c_{2}^{orb}(E)), and the expression e_{orb}(\tilde{\mathcal{X}})-c_{2}^{orb}(E)=7-\left(\frac{10}{11}+\frac{7}{8}\right)-3+\frac{2}{11}=\frac{211}{88}>0. 5. Proof of Theorem 1.1 The topology of T^{2} bundles over complex surfaces is determined in [ 26 ] . The same results hold without change of the orbisurfaces we consider. The method was also used to determine the diffeomorphism type of the spaces M in [ 15 ] . We start by recalling the following result: Theorem 5.1 ( [ 15 ] ) . Let X be a compact K3 orbisurface with a Ricci-flat Kähler form \omega_{X} and orbifold Euler number e_{orb}(X) . Let \omega_{1} and \omega_{2} be anti-self-dual (1,1) -forms on X such that [\omega_{1}],[\omega_{2}]\in H^{2}_{\text{orb}}(X,\mathbb{Z}) and the total space M of the principal T^{2} orbifold bundle \pi:M\to X determined by them is smooth. Let E be a stable vector bundle of degree 0 over (X,\omega_{X}) such that \alpha^{\prime}(e_{orb}(X)-(c_{2}(E)-\frac{1}{2}c_{1}^{2}(E)))=\frac{1}{4\pi^{2}}\int_{X}\left(\|\omega_{1}\|^{2}+\|\omega_{2}\|^{2}\right)\frac{\omega_{X}^{2}}{2}. Then M has a Hermitian structure \omega_{u} and there is a metric h along the fibers of E such that (M,\omega_{u},V=\pi^{*}E,H=\pi^{*}(h)) solves the Hull-Strominger system with \nabla being the Chern connection on M . In Table 3 at the end we list the values e_{orb}(\tilde{X}_{k})-(c_{2}(E)-\frac{1}{2}c_{1}^{2}(E)) for each k. Now Theorem 3.1 , Lemma 4.1 , and Lemma 4.2 give us sufficient conditions to apply Theorem 5.1 as a last step to prove Theorem 1.1 . To use Theorem 5.1 it remains to calculate e_{orb}(X)-(c_{2}(E)-\frac{1}{2}c_{1}^{2}(E)) which will end up being a positive rational number. The case for a stable bundle over \tilde{X}_{1} with b_{2}(\tilde{X}_{1})=5 is explained in Remark 4.1 . In [ 5 ] and [ 3 ] , the orbifold Euler number e_{orb}(X) is calculated the following way e_{orb}(X)=e(X)-\sum_{i}\frac{n_{i}}{n_{i}+1}, where n_{i} are the A_{n_{i}} singularities. The left hand side of the integral comes from the integral \alpha^{\prime}\int_{X}(\text{tr}(R\wedge R)-\text{tr}(F_{h}\wedge F_{h}))=\alpha^{\prime}\left(e_{orb}(X)-\left(\int_{X}c_{2}^{orb}(E)-\frac{1}{2}c_{1}^{orb}(E)^{2}\right)\right). In our construction of E we get that c_{1}^{orb}(E)=0 , so the integral reduces to \alpha^{\prime}\left(e_{orb}(X)-\int_{X}c_{2}^{orb}(E)\right). In order to know the number \int_{X}c_{2}^{orb}(E) we need to calculate the orbifold Euler characteristic of the bundle, that is \chi_{orb}(X,E)=2\cdot\chi_{orb}(X,\mathcal{O}_{X})-\int_{X}c^{orb}_{2}(E). From [ 3 ] we get the formula \chi(X,F)=\chi_{orb}(X,F)+\mu_{Sing}(F) where \mu_{Sing}(F):=\sum_{x\in\Sigma}\mu_{X,x}(F), for any vector bundle F . the map \mu_{Sing} is a group homomorphism, so by the exact sequence defining E , that is 0\rightarrow\mathcal{Q}^{\vee}\rightarrow E\rightarrow\mathcal{Q}\otimes\mathcal{I}_{Z}\rightarrow 0, and the fact that Z\cap\Sigma=\emptyset , we have that \mu_{Sing}(E)=\mu_{Sing}(\mathcal{Q})+\mu_{Sing}(\mathcal{Q}^{\vee})=2\mu_{Sing}(\mathcal{O}_{X}) . Thus we have \chi_{orb}(X,E)=2\cdot\chi_{orb}(X,\mathcal{O}_{X})-\int_{X}c^{orb}_{2}(E), \chi(X,E)-\mu_{Sing}(E)=2(\chi(X,\mathcal{O}_{X})-\mu_{Sing}(\mathcal{O}_{X}))-\int_{X}c^{orb}_{2}(E), \int_{X}c^{orb}_{2}(E)=2\chi(X,\mathcal{O}_{X})-\chi(X,E)=c_{2}(E)=\ell(Z)-\mathcal{Q}^{2}=\ell(Z)-2. Here Z is chosen such that \ell(Z)>\chi(2\mathcal{Q})=2+2\mathcal{Q}^{2} say \ell(Z)=3+2\mathcal{Q}^{2}=7. Thus \int_{X}c^{orb}_{2}(E)=5, and this holds for any of our bundles. Note that these cases are independent of the orbifold structure as opposed to the case of Lemma 4.2 and Remark 4.3 . The Table 1 at the end of the paper lists the divisors \mathcal{Q} such that \mathcal{Q} is ample and \mathcal{Q}^{2}=2 in blow ups of X_{36}\subset\mathbb{P}(7,8,9,12) which has A_{6},A_{7},A_{3},A_{2} , and Table 2 lists the ones for \mathcal{Q} such that \mathcal{Q}^{2}=2 in blow ups of X_{30}\subset\mathbb{P}(5,6,8,11) which has A_{1},A_{7},A_{10} . Using the notations of Remark 4.1 we have that the Picard group of the surface \tilde{X}_{k} with b_{2}(\tilde{X}_{k})=k+4 will be {\text{Pic}}(\tilde{X}_{k})=\langle H,\mathcal{L}_{1},\mathcal{L}_{1}^{\prime},\ldots,\mathcal{L}_{k/2},\mathcal{L}_{k/2}^{\prime}\rangle in the case where k is even, and {\text{Pic}}(\tilde{X}_{k})=\langle H,\mathcal{G},\mathcal{L}_{1},\mathcal{L}_{1}^{\prime},\ldots,\mathcal{L}_{(k-1)/2},\mathcal{L}_{(k-1)/2}^{\prime}\rangle in the case where k is odd. Here \mathcal{L}_{i} and \mathcal{L}^{\prime}_{i} come from blow-ups of A_{n} singularities with n>1 , and \mathcal{G} is the exceptional divisor of the blow-up of an A_{1} singularity, see Lemma 2.1. Now we consider the cases in Table 1 for even k . Let the k -tuple (a_{0},...,a_{k-1}) in \mathbb{Z}^{k}_{+} denote the divisor \mathcal{Q}=a_{0}H-a_{1}\mathcal{L}_{1}-a_{2}\mathcal{L}_{1}^{\prime}-...-a_{k-2}\mathcal{L}_{k/2}-a_{k-1}\mathcal{L}_{k/2}^{\prime}, such that a_{2i}>a_{2i-1} . We check that \mathcal{Q} is ample using Nakai-Moichezon criteria. Note that a class [D]\in Cl(\tilde{X}_{k}) will be represented by H , C_{i} , or C_{i}^{\prime} for some 1\leq i\leq k/2 where \mathcal{L}_{i}^{\prime}=C_{i}+C_{i}^{\prime} , which is the exceptional divisor of the blow up of an A_{n_{i}} singularity. Thus we have to check the cases \mathcal{Q}\cdot[D]=\mathcal{Q}\cdot H=a_{0}H^{2}>0, \mathcal{Q}\cdot C_{i}=-a_{2i-1}\mathcal{L}_{i}-a_{2i}\mathcal{L}_{i}^{\prime} . For the first case we have \mathcal{Q}\cdot[D]=\begin{cases}\mathcal{Q}\cdot H=a_{0}H^{2}>0,\\ \mathcal{Q}\cdot C_{i}=-a_{2i-1}\mathcal{L}_{i}\cdot C_{i}-a_{2i}\mathcal{L}_{i}^{\prime}\cdot C_{i}=a_{2i-1}n_{i}+a_{2i}>0,\\ \mathcal{Q}\cdot C_{i}^{\prime}=-a_{2i-1}\mathcal{L}_{i}\cdot C_{i}^{\prime}-a_{2i}\mathcal{L}_{i}^{\prime}\cdot C_{i}^{\prime}=a_{2i}-a_{2i-1}>0.\end{cases} Now, we consider the cases in Table 2 for odd k>1 . Let the k -tuple (a_{0},...,a_{k-1}) in \mathbb{Z}^{k}_{+} denote the divisor \mathcal{Q}=a_{0}H-a_{1}\mathcal{G}-a_{2}\mathcal{L}_{1}-a_{3}\mathcal{L}_{1}^{\prime}-...-a_{k-2}\mathcal{L}_{(k-1)/2}-a_{k-1}\mathcal{L}_{(k-1)/2}^{\prime}. Again we check for ampleness, the class [D]\in Cl(\tilde{X}_{k}) will be represented by H , C_{i} , C_{i}^{\prime} , or G , where H , C_{i} , and C_{i}^{\prime} are the same as before; and G is the class representing the exceptional curve of the blow up of an A_{1} singularity, and \mathcal{G} the corresponding line bundle. For this case we have \mathcal{Q}\cdot[D]=\begin{cases}\mathcal{Q}\cdot H=a_{0}H^{2}>0\\ \mathcal{Q}\cdot C_{i}=-a_{2i}\mathcal{L}_{i}\cdot C_{i}-a_{2i+1}\mathcal{L}_{i}^{\prime}\cdot C_{i}=a_{2i}n_{i}+a_{2i+1}>0\\ \mathcal{Q}\cdot C_{i}^{\prime}=-a_{2i}\mathcal{L}_{i}\cdot C_{i}^{\prime}-a_{2i+1}\mathcal{L}_{i}^{\prime}\cdot C_{i}^{\prime}=a_{2i+1}-a_{2i}>0,\\ \mathcal{Q}\cdot G=-a_{1}\mathcal{G}\cdot G=2a_{1}>0.\end{cases} This concludes the proof of the main Theorem. Below are the three tables, to which we referred above in this Section. Table 1. Surfaces obtained by blowing up singularities A_{6},A_{7},A_{3},A_{2} of X_{36}\subset\mathbb{P}(7,8,9,12) Surface Blow up Singularities \mathcal{Q} \tilde{X}_{2} A_{3} A_{6},A_{7},A_{1},A_{2} (14,11,28) \tilde{X}_{4} A_{2} A_{6},A_{7},A_{1} (6,2,10,6,9) \tilde{X}_{6} A_{6} A_{4},A_{7},A_{1} (5,1,5,2,11,1,3) \tilde{X}_{8} A_{4} A_{2},A_{7},A_{1} (5,3,7,1,6,1,3,1,3) \tilde{X}_{10} A_{2} A_{7},A_{1} (5,1,6,3,8,1,3,1,3,1,3) \tilde{X}_{12} A_{7} A_{5},A_{1} (5,1,4,1,8,1,3,1,3,1,3,1,3) \tilde{X}_{14} A_{5} A_{3},A_{1} (5,1,5,1,5,1,3,1,3,1,3,1,3,1,3) \tilde{X}_{16} A_{3} A_{1},A_{1} (6,1,5,4,8,1,3,1,3,1,3,1,3,1,3,1,3) Table 2. Surfaces obtained by blowing up singularities A_{6},A_{7},A_{1},A_{4} of X_{50}\subset\mathbb{P}(7,8,10,25) Surface Blow up Singularities \mathcal{Q} \tilde{X}_{3} A_{1},A_{4} A_{6},A_{7},A_{2} (5,3,2,7) \tilde{X}_{5} A_{2} A_{6},A_{7} (5,2,1,4,2,6) \tilde{X}_{7} A_{6} A_{4},A_{7} (5,2,1,4,1,4,1,5) \tilde{X}_{9} A_{4} A_{2},A_{7} (6,2,1,4,2,6,1,5,1,3) \tilde{X}_{11} A_{2} A_{7} (6,2,1,4,2,5,1,5,1,3,1,3) \tilde{X}_{13} A_{7} A_{5} (8,2,1,4,1,12,1,3,1,3,1,3,1,3) \tilde{X}_{15} A_{5} A_{3} (8,3,1,5,3,6,1,3,1,3,1,3,1,3,1,3) \tilde{X}_{17} A_{3} A_{1} (7,2,1,5,1,4,1,3,1,3,1,3,1,3,1,3,1,3) Table 3. Values of e_{orb}(X)-c_{2}^{rb}(E) with \mathcal{Q}^{2}=2 ; \tilde{X}_{1} is the blow up of the A_{1} singularity of X_{30}\subset\mathbb{P}(5,6,8,11) , the rest of the odd entries are blow ups of X_{50}\subset\mathbb{P}(7,8,10,25) , and the even entries are blow ups of X_{36}\subset\mathbb{P}(7,8,9,12) . Surface Singularities e(X) e_{orb}(X)-c_{2}^{orb}(E) \tilde{X}_{1} A_{7},A_{10} 7 211/88 \tilde{X}_{2} A_{6},A_{7},A_{1},A_{2} 8 17/168 \tilde{X}_{3} A_{6},A_{7},A_{2} 9 269/168 \tilde{X}_{4} A_{6},A_{7},A_{1} 10 155/56 \tilde{X}_{5} A_{6},A_{7} 11 239/56 \tilde{X}_{6} A_{4},A_{7},A_{1} 12 139/40 \tilde{X}_{7} A_{4},A_{7} 13 253/40 \tilde{X}_{8} A_{2},A_{7},A_{1} 14 57/8 \tilde{X}_{9} A_{2},A_{7} 15 203/24 \tilde{X}_{10} A_{7},A_{1} 16 77/8 \tilde{X}_{11} A_{7} 17 89/8 \tilde{X}_{12} A_{5},A_{1} 18 35/3 \tilde{X}_{13} A_{5} 19 79/6 \tilde{X}_{14} A_{3},A_{1} 20 55/4 \tilde{X}_{15} A_{3} 21 61/4 \tilde{X}_{16} A_{1},A_{1} 22 16 \tilde{X}_{17} A_{1} 23 35/2