To do

[shinaoka]

Let us discuss what should be implemented in SpM in near future.
Here are my proposals.

1. Log mesh for real frequency
   This may be useful in the analytical continuation of data at very low temperatures.

2. covariance matrix
   More accurate treatment of error bars is necessary.

3. Python interface
   More user-friendly interface. But it may not be so urgent.

[j-otsuki]

I agree that 1 and 2 should be implemented for practical use. In particular, covariance matrix would improve the result.