VPIC simulations of SRS reflectivity vs laser intensity scaling at under NIF plasma conditions (immobile ions are used). The reflectivity is a running time average obtained during an interval of many reflected laser pulses. In the top frame, two sets of reflectivity values are shown for an f/4 speckle: Crosses indicate 3D simulations and diamonds indicate 2D simulations. The middle frame displays reflectivity values for an f/8 speckle from 2D (diamonds) and 3D simulations (squares). In the lower frame, reflectivity values from 2D simulations are compared between an f/8 speckle (squares) and an f/4 speckle (diamonds).
Bowing results from a 3D simulation of an f/4 beam with NIF relevant plasma conditions at , a case corresponding to the cross at in the upper frame in Fig. 2. Results are displayed at a time during the first SRS burst. (a) Contours of at . [(b) and (c)] Transverse cuts of at and , respectively. [(d) and (e)] Profiles of (at ) as a function of corresponding to (b) and (c), respectively. (f) Isosurfaces of constant electric field with color indicating (electromagnetic field) (enhanced online). [URL: http://dx.doi.org/10.1063/1.3250928.1]10.1063/1.3250928.1
Results from a 3D simulation of an f/4 beam with NIF relevant plasma conditions at and at with mobile ions. Middle left frame: the time history of the instantaneous reflectivity showing that many SRS bursts occur during the simulation. Middle right frame: contours of at and showing the EPW field structure during a SRS burst (the magnitude normalized to ). Top eight frames [(a)–(h)]: electron velocity distributions in space integrated over velocity and averaged over each processor’s spatial domain, as marked correspondingly by the rectangles in the middle right frame, along the speckle -axis with and in the vicinity of [(a)–(d)] and in regions on the speckle side [(e)–(h); also around ]. Bottom frame: contours of magnetic field at associated with the axial current from hot electrons (the magnitude of is shown in terms of by the color bar on the right); overlaid with are profiles of at in arbitrary units.
Self-focusing of EPWs from a 3D SRS simulation for an f/8 speckle with and at (corresponding to the case in the middle frame of Fig. 2, indicated by the square). Top frame: isosurfaces of with color indicating at when SRS has saturated (the color bars display and in simulation units). [(a) and (b)] EPW filament necklace and the corresponding profile of at . [(c) and (d)] Contours of at and the wave amplitude at in the region between and , as indicated by the white arrow in the top frame. [(e)–(h)] Contours of at and , taken near the intersection region of the two focusing cones, as indicated by the white arrow, and corresponding profiles of at .
Power gain exponent as a function of 1D gain for a range of 3D laser speckles with Gaussian shape. Results are shown for two different values of corresponding to different plasma density and temperature values. Note that does not exceed 20 until the 1D gain exponent in 3D [triangles and circles ]. At each value of , several independent random boundary condition phases were used; the spread of data points is a measure of intrinsic gain fluctuations. In 3D, these fluctuations are greater when diffraction is stronger for the scattered light than the laser speckle, as measured by . 2D results for (stars) show a much larger gain.
A comparison of total reflectivity from high intensity speckles obtained from the model in Sec. V for two cases of f/8 laser beams propagating through 0.5 cm of NIF hohlraum plasma with . The first is for a beam conditioned with a top-hat random phase plate; the second is for an identical beam with polarization smoothing. The three curves shown in each plot are for threshold values of of 2 (solid), 4 (dashed), and 6 (dotted). In each, the maximum saturated reflectivity from a single speckle has a value . Polarization smoothing is found to reduce the number of large-amplitude speckles and allow for operation at much higher average intensity .
A comparison of reflectivity scaling calculation for f/4 beam in plasma with conditions as shown in Fig. 2: squares are computed on a large Opteron supercomputer using 25 particles/cell (except the lowest intensity case), diamonds on Roadrunner using 60 particles/cell. Essentially the same scaling was obtained in the two sets of calculations; the Roadrunner calculations required 1/10 of the time.
VPIC exhibits nearly a perfect weak scaling on up to almost the entire Roadrunner machine. The largest calculations ran on 12 240 Cell chips and employed more than computational macroparticles using more than 90% of available system memory. The physics problems labeled “LPI science problems 1 and 2” relate to two single-speckle, high resolution LPI calculations as defined in Ref. 3. The two problems are identical, except that they have different numbers of particles per cell (2000 and 6420, respectively); LPI science problem 2 is more particle dominated and thus has a higher overall performance on Roadrunner since a larger percentage of the computation is performed on the Cell SPE.
Article metrics loading...
Full text loading...