We will show that minimally supersymmetric SU(N+2) SQCD models in the middle of the conformal window can be engineered by compactifying certain 6d SCFTs on three punctured spheres. The geometric construction of the 4d theories predicts numerous interesting strong coupling effects, such as IR symmetry enhancements and duality. We will discuss this interplay between simple geometric and group theoretic considerations and complicated field theoretic strong coupling phenomena. For example, one of the dualities arising geometrically from different pair-of-pants decompositions of a four punctured sphere  is an $SU(N+2)$ generalization of the Intriligator-Pouliot duality of $SU(2)$ SQCD with $N_f=4$, which is a degenerate, $N=0$, instance of our discussion. 

This is a joint work with C.Daw in progress. We study the L_{omega_1,omega}-theory of the modular functions j_n: H -> Y(n). In other words, H is seen here as the universal cover in the class of modular curves. The setting is different from one considered before by Adam Harris and Chris Daw: GL(2,Q) is given here as the sort without naming its individual elements. As usual in the study of 'pseudo-analytic cover structures', the statement of categoricity is equivalent to certain arithmetic conditions, the most challenging of which is to determine the Galois action on CM-points. This turns out to be equivalent to determining the Galois action on SL(2,\hat{Z})/(-1), that is a subgroup of

Out SL(2,\hat{Z})/(-1)   induced by the action of  Gal_Q. We find by direct matrix calculations a subgroup Out_* of the outer automorphisms group which contains the image of Gal_Q. Moreover, we prove that Out_* is the image of Drinfeld's group GT (Grothendieck-Teichmuller group) under a natural homomorphism.

It is a reasonable to conjecture that Out_* is equal to the image of Gal_Q, which would imply the categoricity statement. It follows from the above that our conjecture is a consequence of Drinfeld's conjecture which states that GT is isomorphic to Gal_Q.  

Abstract: We use a deep neural network to generate controllers for optimal trading on high frequency data. For the first time, a neural network learns the mapping between the preferences of the trader, i.e. risk aversion parameters, and the optimal controls. An important challenge in learning this mapping is that in intraday trading, trader's actions influence price dynamics in closed loop via the market impact. The exploration--exploitation tradeoff generated by the efficient execution is addressed by tuning the trader's preferences to ensure long enough trajectories are produced during the learning phase. The issue of scarcity of financial data is solved by transfer learning: the neural network is first trained on trajectories generated thanks to a Monte-Carlo scheme, leading to a good initialization before training on historical trajectories. Moreover, to answer to genuine requests of financial regulators on the explainability of machine learning generated controls, we project the obtained ``blackbox controls'' on the space usually spanned by the closed-form solution of the stylized optimal trading problem, leading to a transparent structure. For more realistic loss functions that have no closed-form solution, we show that the average distance between the generated controls and their explainable version remains small. This opens the door to the acceptance of ML-generated controls by financial regulators.