Based on the popular Moomins series of children's books, Snufkin: Melody of Moominvalley might be the gentlest game about anarchy and ecoterrorism ever made. You play as the titular Snufkin, a free-spirited, harmonica-playing nomad and friend to the doughy, innocent Moomins. Upon venturing back to the Moomins' village to visit his best friend, the earnest and eager Moomintroll, he finds that the once-beautiful valley has been ravaged by the Park Keeper, a dopey bureaucrat who has caged animals, sectioned off private parks and generally tried to tame the wild into something tidier and more civilized. That, of course, will not do. So you take it upon yourself to undermine the Park Keeper's efforts, putting out the fires he and his fellow cops have caused (sometimes literally) and letting nature be nature again.

In recent years, we've seen an influx of self-proclaimed "cozy games," video games explicitly designed to invoke good vibes. To help those who could use some help winding down, we've rounded up a selection of games that purposefully deemphasize fail states, violence, overwhelming grinds, intense competition and other aggressive urges, but aren't overly cute for the sake of it or so stripped-down that they're boring. This open-ended sim has you fix up a dilapidated farm and interact with nearby townsfolk. Apart from being one of our favorite couch co-op games, the farming life sim Stardew Valley is also notable for its relaxing qualities. It's a game that's willing to meet you at your pace: If you want to putter around your farm, casually chat up townsfolk, brew beer or fish for a few hours, you can.

Emerging applications in multi-agent environments such as internet-of-things, networked sensing, autonomous systems and federated learning, call for decentralized algorithms for finite-sum optimizations that are resource-efficient in terms of both computation and communication. In this paper, we consider the prototypical setting where the agents work collaboratively to minimize the sum of local loss functions by only communicating with their neighbors over a predetermined network topology. We develop a new algorithm, called DEcentralized STochastic REcurSive gradient methodS (DESTRESS) for nonconvex finite-sum optimization, which matches the optimal incremental first-order oracle (IFO) complexity of centralized algorithms for finding first-order stationary points, while maintaining communication efficiency. Detailed theoretical and numerical comparisons corroborate that the resource efficiencies of DESTRESS improve upon prior decentralized algorithms over a wide range of parameter regimes. DESTRESS leverages several key algorithm design ideas including stochastic recursive gradient updates with mini-batches for local computation, gradient tracking with extra mixing (i.e., multiple gossiping rounds) for per-iteration communication, together with careful choices of hyper-parameters and new analysis frameworks to provably achieve a desirable computation-communication trade-off.