Say in a cash game like this, 5 players, 7500 buyin, SB/BB/Ante 20/40/500
Game setup: http://puu.sh/kmwM3.png

I tried to calculate the Nash range in HoldemResources with the model of push-fold and limp-push-fold, and get these results:
push-fold: http://puu.sh/klN7v.png
limp-push-fold: http://puu.sh/klNiM.png

It makes sense in push-fold model, we can push very wide as the ante is so big with a lot of dead money. But in limp-push-fold model, the limping range is so wide and we can only push a few hands.

Intuitively, I think the best strategy is push a little bit tighter than push-fold range and limp the remaining. As the pushing range is constructed with all hands with EVpush>EVfold, but we also have EVlimp with limping a real option, so we would only push those hands with EVpush>EVfold and EVpush>EVlimp, but EVlimp is hard to calculate and may not be accurate as assuming all checking down in HRC's calculation. But it still surprised me we need to tight up pushing range so much with the results from HRC.

Any ideas on the best strategy in practice? Thank you and I appreciate your help.

How the hell does limp-push-fold work? I mean, what if you limp and no one raises? Then you have to play poker so how does it estimate your EV for those branches?

Is his premise that with antes, a more optimal strategy will include more shoving pre?

Check down. The feature was really just made to allowed people to set up situations with limps so they could edit, not to find Nash ranges with limps included

Here are the full HRC results for the limp/push/fold:
http://hands.holdemresources.net/?id=1vus4t1tjdqq8

@Rusty, the calculation assumes the pots are checked down postflop.

I definitely wouldn't recommend to use the limp version in practice. As SiQ wrote: This feature is mostly there to allow setting up previous action with limps, not to calculate full limping games. Both the postflop assumption and the 3-player restriction are very inaccurate in this specific scenario.

*) UTG-Limp, CO-Limp, BU-shove plays out a full 61.1% of the time. (SB/BB are forced folds in that line due to 3-max restriction.)
*) UTG-Limp, CO-Limp, 2 folds, BB check plays out ~9%. Due to the 40:2640 pot odds, this is enough to make even weak hands +EV.

[If you want to mess around with limp games in HRC, please update to the latest Beta btw. We recently fixed a bug that would affect the postflop evaluation of the BB check lines in this hand. I'd still recommend against calculating "full" limp games anyway though.]