Theta Proxy does not start showing: “A problem occured while an add-in was being initialized (InitializeIntegration failed).”
Possible Solution: This happens if the Theta Proxy Add-in was started from a paths which MS Excel does not trust, e.g. a network share. Copy the Add-in “ThetaProxy-packed.xll” to a trusted location, e.g. the desktop and start from there.
The ribbon for Theta Proxy disappeared.
Solution: This sometimes happens after Excel closes with an error. Follow the instructions for the installation and choose “disabled Items” instead of “Excel Add-Ins” in step 3.
The learning phase takes very long.
Background: This may happen for training of very slow functions. A training a two dimensions may take e.g. 1000 function evaluations. If a single function evaluation takes one second, the total training takes 1000s, which is about 17 minutes.
Approach: The time required for a training depends on several parameters: The time a single function evaluation takes, the desired maximal error, the number of parameters and their ranges of the slow function trained and of cause the speed of your computer. If possible, try to use a larger maximal error, fewer parameters and train within smaller ranges.
My function trains “for ever” and the estimates of the maximal error do not decrease.
Solution: You might be training a function which has jumps or peaks or has other discontinuities. E.g. on option pricing with a tree model (binomial/trinomial tree) can have these jumps in the option value due to the implementation. Make sure that your option pricing model is called with a precision which is better than the maximal error, then the peaks will not appear and the function will train quickly.
In General: Use enough steps resp. iterations in your tree or PDE model to meet the maximal error you desire.