Vulkan and Cuda: Memory Sharing

I’m interested in overwriting the default framebuffer vkImages that are grabbed when refreshing the swap chain and replacing them with a 2d buffer managed by Cuda, such that I only need to present the chain and not render to show an image computed and stored in that buffer.

Is the void pointer argument in the vkMapMemory function able to be written to and read by a Cuda kernel? If this is not the case, are there any elegant solutions to share memory between the two as there once was between OpenGL and Cuda?