We report results from a comprehensive study of the nearby M dwarf Proxima Centauri with the XMM-Newton satellite , using simultaneously its X-ray detectors and the Optical Monitor with its U band filter .
We find strongly variable coronal X-ray emission , with flares ranging over a factor of 100 in peak flux .
The low-level emission is found to be continuously variable on at least three time scales ( a slow decay of several hours , modulation on a time scale of 1 hr , and weak flares with time scales of a few minutes ) .
Several weak flares are characteristically preceded by an optical burst , compatible with predictions from standard solar flare models .
We propose that the U band bursts are proxies for the elusive stellar non-thermal hard X-ray bursts suggested from solar observations .
In the course of the observation , a very large X-ray flare started and was observed essentially in its entirety .
Its peak luminosity reached 3.9 \times 10 ^ { 28 } erg s ^ { -1 } 
[ 0.15–10 keV ] , and the total X-ray energy released in the same band is derived to be 1.5 \times 10 ^ { 32 } ergs .
This flare has for the first time allowed to measure significant density variations across several phases of the flare from X-ray spectroscopy of the O vii He-like triplet ; we find peak densities reaching up to 4 \times 10 ^ { 11 } cm ^ { -3 } for plasma of about 1 - 5 MK .
Abundance ratios show little variability in time , with a tendency of elements with a high first ionization potential to be overabundant relative to solar photospheric values .
Using Fe xvii lines with different oscillator strengths , we do not find significant effects due to opacity during the flare , indicating that large opacity increases are not the rule even in extreme flares .
We model the large flare in terms of an analytic 2-Ribbon flare model and find that the flaring loop system should have large characteristic sizes ( \approx 1 R _ { * } ) within the framework of this simplistic model .
These results are supported by full hydrodynamic simulations .
Comparing the large flare to flares of similar size occurring much more frequently on more active stars , we propose that the X-ray properties of active stars are a consequence of superimposed flares such as the example analyzed in this paper .
Since larger flares produce hotter plasma , such a model also explains why , during episodes of low-level emission , more active stars show hotter plasma than less active stars .