table formatting attempt #139

src/docs/JOSS/paper.md

@@ -163,7 +163,7 @@ In cases where such tight coupling is not required, one may decompose the monoli
 +===================+============+=====================+
 -->
 
-| Nodes | Ranks | DoF/rank | $\huge \frac{\text{Newton iter}}{\text{timestep}}$ | $\huge \frac{\textbf{GMRES iter}}{\textbf{Newton iter}}$ | Setup <br/> [s]  | Solve <br/> [s] | Efficiency |
+| Nodes | Ranks | DoF/rank | $\HUGE \frac{\text{Newton iter}}{\text{timestep}}$ | $\HUGE \frac{\text{GMRES iter}}{\text{Newton iter}}$ | Setup <br/> [s]  | Solve <br/> [s] | Efficiency |
 | :---: | :---: | :------: |                    :-------------:                       |      :-------------:       | :-----:          | :---:           | ----------: |
 | 2     | 72    | 226K     |                          7.5                             |           30.3             |  6,709           |  13,725         |      100%   |
 | 4     | 144   | 113K     |                          7.5                             |           30.9             |  3,816           |  7,479          |      90%    |
@@ -172,7 +172,7 @@ In cases where such tight coupling is not required, one may decompose the monoli
 Table 1: Strong scaling on LLNL/Quartz (Intel Xeon E5-2695 v4)
 
 
-| Nodes | Ranks | DoF/rank | $\huge \frac{\text{Newton iter}}{\text{timestep}}$ | $\huge \frac{\textbf{GMRES iter}}{\textbf{Newton iter}}$ | Setup <br/> [s]  | Solve <br/> [s] | Efficiency |
+| Nodes | Ranks | DoF/rank | $\HUGE \frac{\text{Newton iter}}{\text{timestep}}$ | $\HUGE \frac{\text{GMRES iter}}{\text{Newton iter}}$ | Setup <br/> [s]  | Solve <br/> [s] | Efficiency |
 | ---:  | ---:  |--------: | :-------: | :-------: | :-----: |  -----: | ----------: |
 | 4     | 16    |   1,018K |    7.7    |   44.3    |  2,830  |  4,364  |      100%   |
 | 8     | 32    |   509K   |    7.5    |   46.2    |  1,834  |  3,636  |      66%    |