We introduce a novel differentiable hybrid traffic simulator , which simulates using model of both macroscopic and microscopic models can be directly integrated into neural network for control flow optimization. This is the first that compute gradients states across time steps inhomogeneous lanes. To gradient between two types in framework, we present intermediate conversion component bridges l...

We present a novel, fast differentiable simulator for soft-body learning and control applications. Existing simulators can be classified into two categories based on their time integration methods: Simulators using explicit timestepping schemes require tiny timesteps to avoid numerical instabilities in gradient computation, implicit typically compute gradients by employing the adjoint method so...

There are various statements in the physics literature about stratification of quantum states, for example into orbits a unitary group, and generalized differentiable structures on it. Our aim is to clarify make precise some these statements. For A an arbitrary finite-dimensional C*-algebra U(A) group elements A, we observe that partition state space S(A) not decomposition orbit types (its piec...

We introduce a new model, the Recurrent Entity Network (EntNet). It is equipped with a dynamic long-term memory which allows it to maintain and update a representation of the state of the world as it receives new data. For language understanding tasks, it can reason on-the-fly as it reads text, not just when it is required to answer a question or respond as is the case for a Memory Network (Suk...

In the recent paper Communications in Nonlinear Science and Numerical Simulation. Vol.18. No.11. (2013) 2945-2948, it was demonstrated that a violation of the Leibniz rule is a characteristic property of derivatives of non-integer orders. It was proved that all fractional derivatives Dα, which satisfy the Leibniz rule Dα(fg) = (Dαf) g+ f (Dαg), should have the integer order α = 1, i.e. fraction...

Cost functions formulated in four-dimensional variational data assimilation (4DVAR) are nonsmooth in the presence of discontinuous physical processes (i.e., the presence of ‘‘on–off’’ switches in NWP models). The adjoint model integration produces values of subgradients, instead of gradients, of these cost functions with respect to the model’s control variables at discontinuous points. Minimiza...

This work presents a continuous and differentiable approximation of a Tantalum oxide memristor model which is suited for robust numerical simulations in software. The original model was recently developed at Hewlett Packard labs on the basis of experiments carried out on a memristor manufactured in house. The Hewlett Packard model of the nano-scale device is accurate and may be taken as referen...

The existence of infinitely many solutions for an anisotropic discrete non-linear problem with variable exponent according to p(k)–Laplacian operator with Dirichlet boundary value condition, under appropriate behaviors of the non-linear term, is investigated. The technical approach is based on a local minimum theorem for differentiable functionals due to Ricceri. We point out a theorem as a spe...

conclusion despite multiple exposures, ed control is more effective if wss is performed using apmdr in the aec mode. results using apmdr, the mean number of exposures was 6.1 for ap and 6.5 for lat projections. lat projections required more exposures (6.55%) than ap projections. the mean ed was 0.6276 msv (ap) and 0.6716 msv (lat). the mean ed for lat projections was 0.6061 msv in automatic exp...